148 Commits

Include renames

Author	SHA1	Message	Date
jaap-jan	25407756c3	Look for a newer build the moment the application starts ... The first pass of the update loop waited two minutes. Every pass after it came six hours apart, which is the right interval for a product that ships rarely — but the delay in front of the first one quietly excluded a whole way of using this application. A client opened to reach one host and closed again is over before the two minutes are. Used that way, it never checks at all: not once, not slowly, never. That is precisely the machine ADR 0011 names as the real cost of distributing outside a store — quietly a year behind — and the galling part is that the mechanism to fix it was switched on the whole time and simply never reached. The delay's own argument is recorded in the diff it is being removed from, and it was not a bad one: nothing anybody does in their first two minutes depends on an update, and launch is already contending for the network with a schema migration, a resumed sign-in and a first sync, at the one moment somebody is watching the window. What it weighed was the cost of checking early against the benefit of checking early. It never weighed the cost of not checking at all. ◆ THE YIELD IS WHAT KEEPS THIS OFF THE LAUNCH PATH, AND IT IS NOT DECORATION. Start() is called from MainWindowViewModel.StartAsync ahead of the migration, so an inline first pass would run whatever the channel does before its own first await — Velopack reads the install layout from disk — between the user and their window. Yielding hands the rest of launch back and puts the check in a later turn, which is the same moment in every sense anybody can perceive and none of the cost. So the answer to the delay's argument is not that it was wrong; it is that a yield buys most of what two minutes bought. Task.Yield takes no token where Task.Delay did, so the loop body now observes cancellation at its head. Without that, an application closed during launch spends its last moment asking a release channel about a build it will not run. Two things deliberately not changed. The AUTOMATIC UPDATE CHECKS preference still gates the pass — "on start" means every start, not regardless of what the user asked for, and that setting is already on by default. And the data cost is unchanged rather than merely acceptable: a check is a few hundred bytes and the download only follows if something newer exists, so this moves the same traffic earlier without adding any. That matters most on the phone, where the same loop runs against AndroidUpdateChannel. TheFirstPassRunsAtStart_RatherThanOnADelay drives the real loop rather than CheckOnceAsync, which is the one thing that file otherwise avoids — and here it is the point, because the claim is about when the pass happens rather than what it does. It waits on the pass and not on a clock, so there is nothing to be flaky about: a regression that puts a delay back does not fail on a margin, it spins until the suite's own cancellation ends it. DisposingStopsTheLoop keeps its assertion and gains a note that it is now a race rather than a formality. §16.7 of the manual checks gains the sentence that reopening the application does what CHECK NOW does. It is the step somebody following that section would otherwise discover by accident. 447 App tests and 153 layout tests pass.	2026-08-12 12:03:54 +02:00
jaap-jan	ccaf7a8e72	Give the terminal back the width and the keyboard the session shell took ... Seven things reported from a day's use of v5b's session shell, and they are one commit because five of them are the same complaint from different angles: the window spends too much of itself on chrome describing the session, and the parts that are not chrome do not behave. ◆ THE HOST HEADER IS GONE, and that is a deliberate departure from the design. Terminal.dc.html and SFTP.dc.html both draw a 60-pixel row above the pane: the address on the left, a cross-surface button on the right. Both facts are worth having and the strip they sat in was not — the tab already names the host, and a full-width bar repeating it was the cheapest 60 pixels in the layout to give back. The address is now the first line of the sidebar and the button is stretched across the column under it, so nothing is lost and the pane is taller. Which word that button carries and which command it runs used to be handed in from the two usage sites in MainWindow.axaml, because the control was drawn twice. One sidebar cannot do that, so SessionCrossSurfaceLabel and OpenOtherSurfaceCommand resolve it in the shell — the same place SessionAddress already decides which surface's fact to read. The two directions underneath are untouched: SelectFilesHostAsync for a tab's host, OpenTerminalForFilesHostAsync for a fresh terminal at whatever SFTP has open. SessionHeader.axaml is deleted rather than left unused, and LayoutHarness stops subtracting its 60 pixels from every session screen's budget — the same treatment the retired window-wide tab strip got, and for the same reason: a constant for chrome nobody draws is a suite quietly measuring the wrong rectangle. ◆ AND THE SIDEBAR CLOSES, which the design has no state for at all. 300 pixels of an 1081-pixel minimum is a great deal to spend on a list that is often two rows long. The column now folds to a 34-pixel rail carrying the chevron that brings it back — a rail rather than nothing, because a panel that vanishes leaving no trace is one people report as lost rather than as closed. Both states live in the one control and swap on IsSessionSidebarOpen, so MainWindow's own "Auto" column takes whichever width is showing without knowing the state exists. Written through to ClientSettings.SessionSidebarOpen rather than held for the session. It is a decision about how much of the window a terminal gets, and one that had to be made again on every launch would not really be on offer. ---- THE FOUR SMALLER ONES ---- A SNIP LANDED IN A TERMINAL NOBODY COULD TYPE AT, and looked selected when it got there. Two causes with nothing in common. The click moved Win32 focus onto the sidebar row, and term.focus() in the page cannot take it back — only the host can, so the shell raises TerminalFocusRequested and the window answers with the same posted focus every other path here uses. The highlight was bash: xterm wraps a paste in bracketed-paste markers, readline marks what arrives inside them as an active region, and it stays in reverse video until the next keystroke. Right for a clipboard paste, wrong for a snippet picked off a sidebar. Single-line snips are typed rather than pasted now, which needs no markers; multi-line still pastes, because "runs three commands unasked" is the worse of the two failures and the markers are the whole of what prevents it. A BLACK BAR UNDER THE TERMINAL, on Windows. xterm.css paints its scrolling viewport #000 — its own comment explains why, and it is a macOS scrollbar concern. Everywhere else that black is covered by the rows, except along the bottom: the fit addon floors the row count, so the remainder below the last whole row is bare viewport, up to a line tall, against this page's #171a26. The light square at its right-hand end is where WebView2's classic scrollbar corner lands. The viewport is repainted in the page's own background, and the scrollbar with it — thin and in these colours rather than a grey Windows channel down the side of a near-black terminal, and kept rather than hidden, because a surface that scrolls with no sign that it does is worse than a quiet bar. THE PINS ROW DREW A TOFU BOX. U+E946 is not in the embedded Material Icons face at all — that file is the 2019 build and its cmap skips E944 and E946 — so the rail's Pins row and the hosts screen's own pin badge have both been drawing a missing-glyph rectangle since v5b picked the codepoint. push_pin in that vintage is U+F10D, verified against the file rather than against a codepoints table for a later release of the font. Every other icon codepoint in the repository was audited the same way; this was the only miss. THE KBD CHIP CUT THE CHORD IN HALF. 34 pixels is the design's width for a chip reading ⌘K, and this build substitutes CTRL K — six characters and a space, wider than 34 at 10.5 mono. MinWidth and padding instead, so the design's footprint survives for the day this face has a ⌘ to draw. ---- AND THE POPOVER UNDER THE USER CHIP ---- Reported as not matching the design, and it was not: Button.poprow set a corner radius and a padding and never touched the Background, so every row wore the Fluent theme's own #33FFFFFF button fill. Six raised pills stacked in a menu the design draws as six lines of text — and the hover rule underneath was already correct and simply invisible against a fill that never went away. Set on the ContentPresenter as well as on the Button, the same as Button.flat, because the theme binds its brush there and a Background set only on the control loses to it. The panel itself gets this window's own radius-12 card treatment through a FlyoutPresenter class rather than by widening the shared context-menu rule, and Vaults and Preferences stop being drawn one step dimmer than Settings and Logout, which read as two disabled entries in a menu of five live ones. ---- WHAT PROVES IT ---- Three tests in the layout suite, two of them checked against the defect they describe: the popover row's resting fill (fails with #33ffffff without the style), and the kbd chip against the natural width of its own text, measured on a detached copy because a TextBlock's DesiredSize is already clipped to what it was given and reports 34 inside a 34-pixel chip either way. SessionSidebarTests is new — the sidebar has never been laid out by a test, and it now holds a string of unbounded length beside a button that has to stay clickable. In the shell suite: the cross-surface row in both directions, the closed state surviving to disk, and the focus request being made when a snip lands and not made when it does not.	2026-08-10 16:49:44 +02:00
jaap-jan	e41eca01a8	Stop the SSH suite's server refusing connections at random ... The suite fails intermittently with SshConnectionException "The connection was closed by the remote host", within tens of milliseconds, on whichever test happens to connect first. It has been seen in CI and reproduces locally. This raises sshd's MaxStartups in the fixture, which is the most likely cause and is worth doing regardless. sshd's compiled-in default is 10:30:100: past ten unauthenticated connections in flight it refuses new ones at random, thirty percent of the time, rising to always at a hundred. The image ships the line commented out, so that default was what ran. xUnit runs test classes in parallel and most of the classes here open a connection, so ten in flight is reachable during the opening seconds — and a refusal presents to the client exactly as observed, because a dropped connection and a server that never answered are indistinguishable from that end. ◆ IT IS A MITIGATION AND NOT A DEMONSTRATED CURE, AND THE COMMENT SAYS SO. The flake rate could not be measured. On the Windows development machine the identical unmodified suite ran 85/85 clean and, an hour later, failed 13 runs out of 15; a Linux container gave 30/30 clean and then failed on the first run of the next batch. Docker throughput on that host swings far enough to swamp the effect, so every before/after comparison taken there was noise — including two that were briefly believed. It is committed on the narrower argument that it is right either way. A connection throttle is hardening this suite has no interest in reproducing: it exists to test an SSH client, not to survive a rate limit, and a test server that drops connections at random is a bad test server whether or not it is the cause of this particular flake. The other candidate was the reload window — pkill returns when SIGHUP is delivered, not when sshd has finished closing its listeners and re-execing, so a connection immediately afterwards can be refused the same way. A wait that required three consecutive banner reads before returning was written and then removed: it could not be shown to change anything either, and a fixture carrying two unproven fixes for one symptom is worse than one, because the next person has to disprove both. Both candidates, and how to tell them apart with sshd's own log, are recorded in the fixture and in docs/platform-flags.md.	2026-08-10 11:46:14 +02:00
jaap-jan	ca081af209	Merge branch 'main' into claude/macos-build-release-2a8a0d	2026-08-10 08:58:50 +00:00
jaap-jan	890a5f2246	Give the desktop a macOS head, signed from the first release ... The same application, the same Velopack and the same two-phase person-run release as Windows, with four things forced to differ. Signing is a precondition rather than an improvement: Gatekeeper refuses an un-notarized download outright instead of warning about it, so there was never the "unsigned for now" that ADR 0013 decision 8 argues for on Windows, and release-macos.sh refuses to start without the identities. The packaging split is narrower than it first looked, and the old claim at the foot of ci.yml is why it was worth checking rather than assuming. vpk cross-compiles when told to: 'vpk [osx] bundle' builds a real .app on any platform, and CI now publishes osx-arm64 and bundles it on every main and tag build, which is what catches a restore graph with no macOS native asset. There is no '[osx] pack' off a Mac, and that part is correct — pack drives codesign, notarytool and stapler, which exist nowhere else. The dylib signing loop in the script looks redundant beside vpk's own pass and is not. vpk signs with 'codesign --deep', which is the shape Apple documents as wrong for nested code, and platform-flags has recorded a notarization rejection that names no file since before any of this existed. Signing each native binary inside-out first leaves that pass nothing to get wrong. MacDeviceKeyStore reaches ADR 0007's conclusion through different hardware: a P-256 key in the Secure Enclave under an access control requiring user presence, so the platform enforces the gate rather than this process — which is the whole point of that ADR's amendment. The enclave holds no other kind of key, hence ECIES where Windows uses RSA-OAEP, and the shape that falls out is better than the Windows one: sealing needs only the public half and is silent, so only unlock prompts. IsSupported probes rather than infers, because three ordinary Macs answer no — an Intel machine without a T2, one with no login password, and every unsigned development build, since enclave keys need a signing identity. Two decisions worth stating because they are reversible. arm64 only: a second channel is small work and nobody here has an Intel Mac to walk Phase 18 on, and an x64 package would be the only artefact in this repository reaching users unverified. And the pack id stays DodoSSH.Desktop even though vpk names the bundle after it, so /Applications holds DodoSSH.Desktop.app: decision 2's reasoning binds harder here, because a pack id of DodoSSH would put Velopack's install root on top of ClientPaths.DataDirectory and let an uninstall take the user's un-synced outbox with it. CFBundleDisplayName puts the product name back in front of a person. Measured rather than assumed, since none of it is obvious: the publish and the bundle were both run, LSMinimumSystemVersion is 12.0 because that is the minos in the apphost's own LC_BUILD_VERSION, and vpk copies a custom Info.plist verbatim with no substitution at all — which is why the plist is a template the script renders and not a committed file. What is not done is the half that needs the hardware. There is no macOS runner, so nothing past "it bundles" has ever run. Phase 18 is the whole of the verification, and the two checks most likely to fail are the terminal against WKWebView and the enclave interop, neither of which has executed once.	2026-08-10 10:43:28 +02:00
jaap-jan	9bc9069425	Post the terminal's focus return past the dispatch that steals it ... The first fix handed Android's focus back from inside the keys' Click handlers — which fire inside the UP event's dispatch, and Avalonia's own view requests focus for itself after every handled touch dispatch returns (AvaloniaView.DispatchTouchEvent, decompiled from 12.1.1). So the platform's request ran after ours and undid it microseconds later, which is exactly what the phone showed: the terminal still lost focus. The return is now posted onto the main looper, landing one message after the dispatch that stole, and it is wired at the row for both halves of a press — DOWN steals too, and Click only exists for UP, so a keyboard detached at DOWN would otherwise stay detached for the whole length of the press. Check 11.10a now also says what a tolerable blink looks like against a failure that stays.	2026-08-09 21:35:08 +02:00
jaap-jan	e936ab4646	Announce a session's end when it is actually over, and for closes too ... The phone's notification kept saying '1 shell connected' after the shell was gone, and both close routes were at fault. A shell exiting on its own raised SessionEnded from inside its run's finally block — where the run task is by definition not yet complete, so the LiveSessionCount the keep-alive reads still counted the dead shell, and nothing fired later to correct it. A tab closed by hand announced nothing at all, by a recorded decision that assumed every subscriber was the closer; the keep-alive is not, and a close it never heard about left the notification claiming a shell over nothing. The end is now announced from a continuation after the run completes, and CloseSessionAsync announces after its own drain — every subscriber was already a reconcile-to-reality handler, so the echo the old remark feared costs nothing. Shutdown stays silent: it is dismantling the subscribers along with the sessions.	2026-08-09 13:01:14 +02:00
jaap-jan	506d2803a2	Hand Android's own focus back to the terminal after an accessory key ... Focusable=false was only ever half the fix, and its remark now says so: Avalonia's focus stays on the NativeWebView, but the touch that presses a key still hands Android's native focus to Avalonia's input view — the platform moves it before Avalonia decides anything. The WebView's input connection dies with it, the keyboard swaps to its no-input layout, and the inset churn parks it over the very row that was tapped. Each key now returns that focus once its byte is on the wire, through a sibling of SoftKeyboard that walks the decor view to the one WebView this application has. Free when nothing moved. Check 11.10a is the phone-in-hand proof.	2026-08-09 11:35:15 +02:00
jaap-jan	3f5979d639	Record the renderer-reattach correction and its phone checks ... The port notes carry the third correction of this round: the data plane assumed a renderer that attaches once and lives forever, which no foreground service can make true of Android's separate WebView renderer process. Phase 11 gains the two checks a phone can run — close and reopen a connection, and a backgrounded shell surviving its renderer being killed, banner and all.	2026-08-09 10:54:45 +02:00
jaap-jan	3977f68870	Record the keep-alive corrections in the port notes and the manual checks ... The port doc's backgrounding decision now carries the four corrections rather than describing a wiring that was not true, and Phase 14 gains the checks a phone can actually run: a backgrounded shell surviving, an idle Files connection surviving, the permission ask arriving at the first thing worth showing, and a refusal costing the notification and nothing else.	2026-08-09 10:14:24 +02:00
jaap-jan	dbf6ce1bcf	Give the phone its pins: an editor section and chips on Files ... The data was never the gap — HostSecret.PinnedPaths syncs and merges on both heads, and the desktop's drawer has staged it since v5 — the phone just had nowhere to add, remove or use a pin. Now it has both halves. The host editor page gains a QUICK ACCESS section over the same shared staging the drawer binds (EditorPinnedPaths, AddEditorPin, RemoveEditorPin), with the remove target at this head's 44dp touch floor rather than the desktop's 22-pixel close box, and no folder glyph because this head embeds no icon font for one. The page also gains a Status line of its own: the add command's five refusals speak through Status, and this page covers the screen that normally draws it — a refusal nothing shows is no refusal at all. The Files screen draws the connected host's pins as chips between the breadcrumb and the listing, each running GoRemoteCommand exactly as a crumb does. They are captured at connect, like ConnectedTo and the session facts before them; a bucket gets none, having no HostSecret to pin anything on. Covered headlessly in ShellFlowTests — connect populates, disconnect clears, a bucket stays empty — and by manual checks 8.18 and 8.19, whose phase preamble also stops claiming thirteen checks when it lists twenty-one.	2026-08-08 23:09:57 +02:00
jaap-jan	009b35e069	Cover the mixed-keychain regroup refusal headlessly ... The two-keychain branch of VaultViewModel.RegroupChosenHosts had no test: 7.6a's manual walk was the only thing asserting that a mixed set gets the sentence instead of the picker. A ShellFlowTests case now ticks a host in each of two vaults, reads the refusal off the status line, and shows the same command opening the picker once the set is one keychain's again. Check 7.6a cites the test and keeps only the popup wiring for the eye.	2026-08-08 22:19:46 +02:00
jaap-jan	d32f5609e3	Rewrite checks 7.6/7.6a for the picker that replaced the drag ... The drag onto a group card went with v5's flat sections; filing a set is the chosen-hosts menu's "Change group..." picker on both heads now. The two checks walk that route instead and say honestly what ShellFlowTests and ScreenLayoutTests already cover, what only a real popup can show, and that nothing automated raises the mixed-keychain refusal. The numbering preamble's example swaps to citations that still exist.	2026-08-08 22:12:44 +02:00
jaap-jan	9d5ff9f23a	Draw what authenticated, and over what, on the session status bar ... The v5b design's own row: the negotiated cipher, then the host key's algorithm and the name of the key or credential that authenticated, as one mono run beside CONNECTED — on both surfaces, off MainWindowViewModel's surface-aware SessionCipher and SessionIdentityText, the same shape SessionAddress set. The identity's name comes out of TryBuildAuthentication, the one resolution point that always had it in scope and always threw it away; it rides HostAuthentication to the tab and to the SFTP connect alike. Three deviations, recorded in the gaps doc: the algorithm prints as negotiated rather than shortened, the run is plain text because no pin-details modal exists for an open session, and a typed password shows the algorithm alone — there is no item behind the dot. A dead terminal tab keeps its facts for the scrollback still on screen; an SFTP disconnect, with no scrollback, clears them.	2026-08-08 20:54:56 +02:00
jaap-jan	8915650a0d	Record the phone catching up in the design-import log	2026-08-08 15:22:04 +02:00
jaap-jan	c59b517fdf	Record v5c in the design-import log, and true up the manual checks	2026-08-08 14:17:20 +02:00
jaap-jan	422d5ca10e	Record v5b in the design-import log, and true up the manual checks	2026-08-08 00:49:59 +02:00
jaap-jan	7d64027972	Record v5 in the design-import log, and let the working notes go	2026-08-07 18:14:59 +02:00
jaap-jan	281e828e25	Sweep out the group-card navigation nothing reaches any more	2026-08-07 18:14:54 +02:00
jaap-jan	82966af37b	Let a connection be reached through a proxy on this machine's loopback ... Step 1 of docs/reaching-a-host-you-cannot-dial.md, and it is not the step that document said it was. SshConnectionRequest carries an optional SshLoopbackProxy and BuildConnectionInfo hands SSH.NET its proxy ConnectionInfo when there is one. Nothing passes one yet: the callers are jump hosts and the relay, which are steps 2 and 3. ◆ THE BRIDGE WAS THE WRONG FIRST STEP, AND BUILDING IT WOULD HAVE BEEN THE MISTAKE THIS DOCUMENT IS ABOUT. ADR 0004 says the relay's loopback bridge "also provides ProxyJump via a SOCKS5 dynamic forward — one mechanism, two features", and the plan took that to mean the bridge was the shared foundation. It is not: ForwardedPortDynamic *is* the listener for a jump host — SSH.NET accepts on it, speaks SOCKS5 on it and tunnels through the bastion — so nothing is left for a bridge of ours to do on that path. The relay is the case with no SshClient to hang a forward off, so it is the bridge's only consumer, and the bridge belongs in the commit that uses it. What the two actually share is one level down and a tenth of the size: being told to reach a target through a loopback proxy while staying about the target. That is what this is. Three properties, one test each. A port and nothing else, so a proxy anywhere but loopback cannot be expressed. The failure that shape rules out is an open SOCKS proxy on the user's network for the life of a shell, which nothing would report — so it is made unrepresentable rather than validated, on the same grounds AuthenticationChoice carries a kind. SOCKS5 rather than a dumb pipe, which is what keeps host key pinning honest. The target's own name and port stay in the request, travel to the proxy in the CONNECT, and are what the gate pins — so a machine reached through a bastion is pinned under its own name instead of under 127.0.0.1 on whatever ephemeral port that day's forward got, which is not an identity at all. A pipe would have meant handing SSH.NET a stand-in and remembering everywhere else that it was one. And a proxy that is not listening fails as a connection error rather than as an unknown host key. The gate turns "no host key seen" into a fingerprint prompt, and a connection that never reached a server has seen none either; the prompt would offer to fix the wrong thing, with no fingerprint to show. TWO THINGS THE TESTS MEASURED RATHER THAN ASSUMED, both found by the first run failing. The target is resolved at the *bastion*, not here — a SOCKS CONNECT names it and the far end looks it up. So the test asks for localhost:2222, the address inside the container, and the published port this host would use means nothing there. That is not a quirk of the fixture; it is what ProxyJump means, and it is why an ssh_config writes the target's internal address beside its jump host. Getting it wrong is a SOCKS "general failure" that names neither end. And the test server refuses forwarding. linuxserver/openssh-server ships AllowTcpForwarding no, which a dynamic forward does not notice — opening one asks the server nothing — so every connection through it is refused at channel-open and reported as the same general failure. The fixture patches it and HUPs sshd. There are two sshd_config files in that image and the running server uses /config/sshd/sshd_config; the first attempt patched /etc/ssh/sshd_config, which is the one a search finds first, changed the text and nothing else, and left the failure exactly where it was. VERIFIED. Build clean with no new warnings, 85 tests in Client.Ssh.Tests against the real sshd, and the solution builds. The proxy test was seen to fail — proxy.Port + 1 in BuildConnectionInfo — and seen green again. An earlier mutation attempt did not compile, and the log said 85 passing because the run never started and the previous log was still on disk; the second attempt deletes the log first, which is worth copying whenever a mutation "passes". dotnet format reports one pre-existing IDE1006 in DodoSSH.Api/Features/Events/EventsEndpoint.cs, in a project nothing here touches. Left alone.	2026-08-07 13:48:15 +02:00
jaap-jan	575a9a9f5e	Stop the relay checkbox promising a connection this client cannot make ... Ticking "Connect through the server relay" moved the host's address and port out of the encrypted payload into plaintext columns on the server — the single deliberate privacy concession in the design, per ADR 0004 — and then the client dialled the address directly, exactly as it does with the box clear. VaultViewModel builds SshConnectionRequest(hostname, port, username, credential) and nothing on this side reads RelayEnabled at all. The connection failed the way it always had, for a machine the laptop could not reach, with nothing saying the box had done nothing. The server half is built and shipped: tickets, the WebSocket, the deny list, the CHECK constraint that enforces a non-null address for a relay-enabled host. What does not exist is the client's path to it, so this is an unfinished feature rather than a broken one — but the control in front of it was collecting the cost of the finished version. Both heads now say so, in the label and in the first sentence of the paragraph under it. Not disabled, and that is the one decision here worth stating: a host somebody has already ticked has to be able to lose the flag, and a control greyed out with the concession switched on would trap it there. Tickable and honest beats untickable and stuck. This is step 0 of docs/reaching-a-host-you-cannot-dial.md, and the only step of it that should ship alone — the sentence is written to be deleted when the bridge lands. VERIFIED. Build clean, 112 layout tests. The drawer's paragraph is longer than it was and the host editor is measured with the drawer open at the window's minimum, so the wrap is held inside the column rather than assumed to fit.	2026-08-07 08:47:02 +02:00
jaap-jan	6185d74800	Write down the two things a user is promised and does not get ... Two plans, both for the same class of defect: a control or a code that a user is told to rely on, backed by storage and by nothing else. Neither is started; what follows is the reasoning, so that starting is not where it gets thought about. ── UNLOCKING WITHOUT THE PASSPHRASE ───────────────────────────────────────────────────────────────────── Every account is issued a recovery code at enrollment. The client generates it, wraps the identity bundle under KEK_rc, the server stores that wrap as UserKeyWrapKind.Recovery, and both heads work to make sure the user writes it down — the phone raises FLAG_SECURE for that screen alone and will not let anybody past it. Nothing can use it. SessionOpener has UnlockAsync and UnlockWithDeviceAsync, and there is no third. Walk the failure through: forget the passphrase, and the bundle cannot be unwrapped, so no vault key opens and every item is unreadable. Signing out and back in returns the same passphrase wrap. The device key would be the other door, and sign-out withdraws it — which is the advice the unlock screen gives for exactly this situation. The loss is total and permanent, and the thing built to prevent it is inert. The docs already disagree with each other about this, which is how it surfaced. manual-checks §10.2 calls the code "the only thing standing between a forgotten passphrase and an unrecoverable vault"; android-port says losing it *along with* the passphrase is what makes a vault unrecoverable; README says signing out is the only answer and nothing can recover one. The third is the true one today. More than half the work is already done and one piece of it was done on purpose: LocalCacheKey derives from the identity bundle rather than from MK — crypto.md §3.2, changed 2026-07-30 — specifically so an unlock that never computes MK can still read the cache it wrote. What is missing is an endpoint to serve the wrap, an unlock path, and a way to set a new passphrase afterwards, without which the account unlocks with a one-time code forever. That last step is the same re-wrap a change-passphrase feature needs, so it delivers both. Two traps are recorded because both would produce a code that verifies nowhere. The derivation uses the displayed string *including its dashes*, so the unlock must canonicalise to the printed form rather than strip it; and the recovery wrap uses a different Argon2 profile to the passphrase one (64 MiB against 256), so it must derive from the parameters served with the wrap rather than from a profile constant. ── REACHING A HOST YOU CANNOT DIAL ────────────────────────────────────────────────────────────────────── This started as "delete the dead jump-host field" and inverted twice. HostSecret.JumpHostIds is stored, validated, encoded and three-way merged, and nothing reads it or writes it — the ssh_config importer looks like the writer and is not; it records ProxyJump as an option and a note saying DodoSSH cannot honour it. The first draft recommended deleting it. That was wrong twice over. ADR 0004's last consequence had already designed the implementation — a loopback TCP bridge for the relay, and "the same bridge provides ProxyJump via a SOCKS5 dynamic forward", one mechanism and two features — which the pinned SSH.NET 2025.1.0 supports through ForwardedPortDynamic and ProxyTypes.Socks5, checked in Renci.SshNet.xml rather than remembered. And the stored shape is right: an ordered list of host ids is what a chain is, the merge arm is correct, and the missing schema version is a line to add. ◆ Looking properly found the same shape one field over, where it costs something. RelayEnabled is also stored, merged and never read by the connect path — but it is user-settable, and both heads draw a checkbox promising it. Ticking it moves the host's address and port out of the encrypted payload into plaintext columns, which ADR 0004 calls the single deliberate concession in the design, and then the client dials directly anyway. The privacy is spent and the feature is not delivered. That is a defect rather than a gap, and it is step 0. The comparison the plan turns on: the relay reaches what the *deployment* can reach and the jump host reaches what a *machine in the keychain* can reach, so they are not substitutes. On a self-hosted box outside the target's network the relay reaches nothing the laptop could not. And the privacy ordering is the opposite way round from the ADR's framing — the relay costs a plaintext address, the jump host costs nothing, because the operator is not in it. Both documents carry a section on what their own earlier reasoning got wrong, which for the second one is the load-bearing part: "nothing reads this field" was read as evidence of a mistake when it was evidence of an unfinished feature — and the same sentence one field over would have found the checkbox that is lying.	2026-08-07 08:44:21 +02:00
jaap-jan	88809f0d66	Stop the docs claiming absences that have since been built ... An audit of README.md, the seven docs and the fourteen ADRs against the code, looking for what is described as absent or planned. Most of it held. What did not is here, and it clusters: every stale claim but one is downstream of the settings file arriving without this document noticing. design-import-gaps said the client has no preferences store and writes exactly two files. It writes three — ClientSettings is in settings.json beside the cache — and two preferences are saved through it. From that one error followed four more: the terminal font size row said "fixed at the renderer's 13px" when it has been 8 to 32 from a screen and three chords for some time; the transfer-resume row and the per-host last directory row both blamed a store that now exists, when what they actually want is a table and a scalar file is the wrong shape for one; and the Preferences table asserted no preference could be saved at all. It also said TerminalServerOpcode has four values and none carries an option. It has eight, and one of them is FontSize — which is the interesting part rather than a counting error, because that opcode is the proof that the frame these rows say is missing can be built. The rows now say what each one would actually take, which for three of the four is a setting, an opcode and a control, and for the Backspace row is a reason: which byte backspace sends is a fact about the remote's stty, so a client-side switch fixes a mismatch by hiding it. THE SAME TWO ERRORS WERE SHIPPED IN THE INTERFACE. The preferences screen carries a NOT BUILT YET list, so that what the screen does not do is as legible as what it does — and it said terminal size was hard-coded a hundred lines below a working size control, and said there is one release channel a month after the nightly shipped. A list of absences is only worth having if it is true, and a screen contradicting itself in the same scroll is worse than no list. Both lines are corrected rather than removed: the first now says which three of the four are genuinely hard-coded, and the second says what is actually missing, which is a way to change channel from inside the application rather than by installing the other build. docs/adding-hosts-on-the-phone.md is deleted. It was a work plan whose own header says "Status: built. All six steps." — nothing links to it, and the decisions it records are in the code it produced, including the one it is proudest of: HostSecret.AsksForPassword carries its own remark on why naming neither binding had to stop meaning "ask me". What was left was step ordering and per-test instructions for work that shipped. Git keeps it. crypto.md is deliberately untouched. It is normative and frozen, and its claims are about the DSH1 format rather than about this build — including the one that reads oddly next to the code, that a passphrase is one of four ways to open a vault. Under the spec it is. What is missing is a statement about what this build can open, and that belongs beside the spec rather than inside it; see docs/unlocking-without-the-passphrase.md. VERIFIED. Build clean, 112 layout tests, 354 app tests. The preferences screen is measured by the layout suite, so the longer copy is held inside the window at the minimum size rather than assumed to fit.	2026-08-07 08:43:49 +02:00
jaap-jan	507cd9ff88	Choose more than one host card on the desktop, the way the phone already can ... The chosen-hosts set has been in VaultViewModel since the phone's connect card became a contextual action bar: a set of entity ids, a tick on the row, and seven things that can be done to it. Only one head could fill it. The desktop's grid answered a press with one selection — the card the drawer, CONNECT and the context menu are about — so filing eleven imported machines under a heading was eleven drags, and clearing out a vault was eleven rounds of the deletion question. So the pointer gets three ways into the same set. Ctrl-clicks a card to tick it, Shift-clicks to tick the run between the anchor and the card, and drags a band out over the space between and below the cards to tick everything it touches. Esc, CLEAR, a plain click on a card and a click on the empty space each drop it, and Ctrl+A takes every card being drawn — VisibleHosts, so with something in the find box that is the ones on screen and not the ones it is hiding, which is the version of that shortcut whose result can be checked before Delete is pressed. TWO SELECTIONS ON ONE SCREEN, AND KEEPING THEM FROM DISAGREEING IS MOST OF THE CHANGE. Ctrl and Shift are answered on the tunnel and marked handled, so the ListBox never moves its own mark onto the card: a Ctrl-click that also selected would light the card it had just unticked and open the drawer on a machine somebody is removing from a set. A plain press drops the set unless it lands on a ticked card, and that case is deferred to the release, because the press may be the start of a drag of all of it. After any ordinary click exactly one card is in play, which is what makes every command on the screen unambiguous again. The context menu is where the seven live, and it is one markup with two halves gated on IsChoosingHosts. Connect, Browse files and Edit… are drawn only for a single ticked host, as the phone's sheet collapses them and for the same reason; the other four read better for a count. A right click on a card outside the set drops the set first, so a Delete… about the card under the pointer can never be offered while six sit ticked behind the menu — the same rule OnContextRequested has always enforced for the selection, reached from the other direction. No bar of buttons: the phone raises one because it has no other way to hold seven entries, and a strip repeating a menu that already exists would be a second home for the wording that matters most. What the desktop gains instead is a count beside the HOSTS heading, CLEAR, and a sentence saying where the actions are. A drag that starts on a ticked card carries every ticked card. The payload is a list rather than a row now, and a drop of more than one goes through FileChosenHostsUnder, which makes the refusals once — an open editor, and a group belonging to one keychain — and reports a count instead of forty status lines. Moving whichever card the pointer happened to be holding and leaving the other five where they are is a gesture that quietly does a fraction of what it looks like it does, and the five left behind look filed. The three panels the set's actions raise had never been drawn in a window: the vault picker with its key question, the group picker, and the deletion question. All three sit above the grid rather than over it, which is the arrangement the GROUPS section and the phone's list already use and for the reason written there — the ticked cards are the information the question exists to give, so the grid shortens instead. A DEFECT FOUND BEHIND IT, AND IT WAS ALREADY LIVE ON THE PHONE. The deletion question names a count and the run that answers it reads the set again, and nothing kept the two the same set: the panel is deliberately above a live list, so one more tick between "Delete these 6 hosts?" and pressing DELETE deleted seven, with the seventh named in nothing the user had read. It needed a deliberate act on a phone and a second's work with a band, which is what turned it up. VaultViewModel now remembers which hosts the question was asked about and drops the question when the set stops being them — the question rather than the set, because what somebody has just chosen is what they meant. It also covers the case nobody performs: a colleague's deletion arriving mid-question and shrinking the set under it. VERIFIED. 354 tests in App.Tests and 111 in App.Layout.Tests, build clean, no new warnings. Six gesture tests drive real pointer and key input through the headless window — the modifier click and what it must not do to the selection, the run and its re-measurement from the anchor, the band and the click that drops the set, Ctrl+A under a filter, and the menu's two halves — plus a DragOver carrying two hosts. Four layout tests measure the strip and the three panels at the window's minimum; the vault panel binds a key to its host first, or it would measure the short shape and certify the tall one. Two flow tests cover the multi-drop's write and its refusal, and the deletion question dropping itself. manual-checks gains 7.6a for dragging a set, which no test can see for the reason 7.6 gives, and 7.7a for the gestures — the rectangle actually being painted and the tick and the fill being legible together are the two things the harness cannot look at.	2026-08-06 14:11:48 +02:00
jaap-jan	1e8a1f2e83	Merge branch 'claude/trust-connect-popup-56abec'	2026-08-06 12:27:20 +02:00
jaap-jan	4f9faa2fe3	Ask about a host key where the connection was made, not on the host list ... The trust prompt was two banners at the top of the desktop's hosts screen, so the shell navigated there before letting a handshake raise one: Screen = Hosts, Surface = Page, in OnVaultConnectionFailed and again in the palette's own connect. The reason was sound — a connection can be started from Ctrl+K on any screen, and a question behind whatever somebody is looking at is a question nobody can answer — and it was answered the wrong way round. Rather than making the decision reachable from where the user is, it moved the user to where the decision was, and charged every screen for it. It is worst for the one connection that has no host at all. A machine typed into the phone's connect box is deliberately in no keychain, so a first contact from there judged it on a list it does not appear on, after taking the box that dialled it away. So both heads now draw the decision over the surface. HostKeyCard is the desktop's, and is the counterpart of the phone's HostKeySheet: a scrim with no press handler, because a question with two named answers must not be answerable by missing; the unknown key offering TRUST AND CONNECT, because judging a fingerprint against what an operator published is a decision a person is entitled to make and the only moment they can make it; and the changed key offering no way forward at all, because a button beside that warning is "continue anyway" with two clicks instead of one. The phone needed no new markup — its sheet was already a shell-level overlay, so deleting the navigation is what puts it over the Connections screen. IsHostKeyDecisionShowing is on the shell rather than on a screen because the answer decides an occlusion. A second connection can be refused while a first one is open, so this card is routinely raised over a live terminal, and that rectangle is a native child window: layered over it the card would be sliced at its left edge with TRUST AND CONNECT taking no clicks, which for the most safety-critical question in the product is the worst place for that class of bug to land. IsTerminalShowing gives the rectangle up instead. The banners are gone rather than copied. One prompt in two markups is two copies of the most safety-critical wording here, and the second is the one that goes stale. TWO DEFECTS FOUND BEHIND IT. VaultViewModel.RejectHostKey cleared only the pending key and never the mismatch, so the changed-key refusal had no working exit. That was invisible for as long as it was a banner nothing was drawn over — nothing was trapped, and the next attempt cleared it — and it was already live on the phone, where that refusal is an opaque full-screen panel whose one button runs this command: pressing it left the panel up over every screen the user went to next, including the host editor the panel tells them to open. TransfersViewModel.RejectHostKey has always cleared both; the vault's was the outlier. Its button said BACK TO HOSTS, which was wrong twice over, and now says BACK. And an assertion written for this change could not fail: the palette test asserted the renderer was collapsed in a scenario whose only tab had just been removed, so it was collapsed for want of a session whatever the occlusion rule said. It is gone, with a note pointing at the test that can fail on it. VERIFIED. 1580 tests, build clean, no new warnings, format clean. Three mutations each seen to fail and then seen green again: dropping !IsHostKeyDecisionShowing from IsTerminalShowing, caught by AChangedHostKey_CollapsesTheTerminalItIsRefusedOver; reverting RejectHostKey to clear one flag, caught by RefusingAHostKeyDecision_TakesItOffTheScreen(false) and by that same test; and dropping the two host-key arms from OnVaultPropertyChanged, caught by TheHostKeyDecision_IsAnnouncedToTheWindowWhenItArrivesAndWhenItGoes. That last one is the first test in this repository to watch PropertyChanged, and it is worth being the first: every other assertion about the flag reads it directly, and a direct read passes with the subscription deleted — while the card would never go away. The two layout tests moved with the prompts, from the hosts screen to the card. manual-checks gains 7.4a for the occlusion, 7.4b for getting out of a refusal and 11.7a for the hand-typed case, none of which a test can see; 1.5 and 7.4 are corrected rather than left describing a window that no longer moves. ONE ROUGH EDGE, DELIBERATELY LEFT. On the desktop, refusing a first contact whose tab was the only one leaves the terminal surface with no tabs — a blank rectangle under the strip's "no terminals open · press + or Ctrl+K", which is the one sentence near that rectangle Avalonia can draw. The alternative was falling back to the page, and on the phone that means the host list, which is the bug this commit is about. A desktop connect page would close it properly.	2026-08-06 12:25:37 +02:00
jaap-jan	6d6edb02c1	Keep an open editor's pickers in step with the vault ... The host editor's four pickers were snapshots taken when it opened, and the comment on EditorAuthenticationChoices said why: a picker whose contents move under somebody halfway through a form is worse than a list a minute stale, and only one editor could be open at a time anyway, so the only way to add a key was to close this one. The second half of that stopped being true when AHostEditorIsInTheWay was split from AVaultEditorIsInTheWay. The host editor is the Hosts screen's business and the keychain's editors are the Vault screen's; neither refuses the other now, which was the right split — it stopped three quarters of a screen going inert over an editor the user was not looking at — but it left the assumption those snapshots rested on false and nothing to notice. So the ordinary way of using the feature was the broken one. Somebody starts editing a host, finds there is no key to bind it to, goes to KEYS, makes one, and comes back to a picker that does not have it — with the fix being to throw the form away and start again. The same for a password, a tag, a group, and for a whole vault made on the Teams screen because the host being typed belongs to the team rather than to the person typing it: the vault they had just made for it was the one place they could not file it. RefreshOpenEditors refills whichever editor is open, and it hangs off ReloadAsync rather than off the twenty-odd commands that write to the vault. That is the choice worth stating, because it is what makes a sync count as well as a save: a key pulled from another machine reaches the open editor by the same path a key typed here does, and a place that wrote to the vault without refreshing the editor would be a bug nobody would find for months. What the old comment was protecting against is real, so every picker is put back onto what it was already showing, by id, and not one typed field is touched. An editor that reset its own bindings because a background sync landed would be a worse bug than the stale list this fixes — it would rebind a host as a side effect of somebody else's work. The placeholder entries go back too, which is the case 3.4 measures: a group deleted on another machine mid-edit still cannot unfile the host when the form is saved. The group editor gets the same treatment for the same reasons; it shares the drawer, and its default binding is lent to every host under it. The snippet editor's vault picker was the same copy of the same list and went stale the same way. It watches TargetVaults rather than the reload, because that screen has always been a wrapper over the vault's collections and has no reload of its own to hang off — which is how it already follows Snippets. The move panels are deliberately left alone. A vault arriving from a sync while one is open still will not appear in it, but a move panel is opened by the act that fills it and its picker resets its selection to the first entry on every rebuild, so refreshing it would move a destination somebody had chosen. Same class of bug, different answer, and not this change. Five tests, and four of them were checked failing with the RefreshOpenEditors call commented out: a key reaching the open host editor and binding when chosen, an item arriving without moving a selection that was already made, a tag arriving as an unworn chip, a key reaching the group editor, and a vault reaching the host and snippet editors without moving either. Manual check 7.12 sits beside 7.11, which is this same bug on the files screen's picker, and says what the worse failure would look like: a picker that moves rather than one that does not notice.	2026-08-06 12:08:25 +02:00
jaap-jan	f1d6499bb5	Merge branch 'main' ... Two of main's changes land in files this branch rewrote, and both needed carrying across by hand rather than by the merge. The phone's nav staying up on Connections with nothing running is a fourth input to RefreshChrome, which this branch had already given two more — whether hosts are ticked and whether the host editor is filling the screen. They compose: the rail and the bottom bar now ask (pages || connectPage) && !editing, so a page-shaped terminal surface keeps its way off the screen and the editor still takes the whole display. The key question under the host's move panel is the harder one, because this branch deleted the panel it was added to. The connect card is gone and the phone's only route to a move is the action bar, so leaving the merge to take this side would have removed a capability main had just shipped — silently, since nothing would fail to build. It is asked in the action bar's own picker instead, in two shapes fewer than the desktop's: one host, because which key to carry is a fact about one machine and a selection of six has six answers, and a move rather than a copy, because taking the key out from under an original that is staying put would leave that original unable to connect. BindingOfTheMovingHost splits into MovableBindingOf so both heads answer it the same way from different panels. Main also fixed a real trap in the same commit — a host that only inherited its key from its group arrived in the destination naming nothing at all, because the group stays behind — and the batch move had the same bug for the same reason. It goes through Detached now, which is where that fix lives. The carried host is written as the carry left it rather than being detached again, which is the one thing worth measuring: the key takes a new id over there, so a run that rebuilt the payload from the row would send the machine across naming a tombstone. Both directions are pinned, along with the rule about which shapes the question is asked in at all.	2026-08-06 09:30:00 +02:00
jaap-jan	c882fa0cd3	Give the phone a selection instead of a card under the list ... A long press on a host raised a connect card over the bottom of the list: a password box, CONNECT, EDIT, MOVE and DELETE. It was the right idea in the wrong place. It covered rows, it had room for five things and never a sixth, and every one of them was about exactly one machine — so filing eleven imported hosts under a group was eleven trips through a form, and there was nowhere to put a sixth action if anybody wanted one. A long press now chooses the host it landed on, and the actions move into a bar across the top of the screen, in the vault header's place rather than beside it. That is where Android has put them since contextual action bars existed, and it is the one strip a list can never grow into — but the real reason for it is that while it is up the screen is unambiguously about the ticked hosts and nothing else, which is what lets the count in the middle of it mean something. Left to right: the cross that leaves the mode, the count, the pencil, and a ⋯ holding Connect, Connect via SFTP, Move to vault, Copy to vault, Change group, Duplicate and Remove. A tap still connects and still raises nothing. Once anything is ticked it ticks and unticks instead, which is what every Android list does and is not merely a convention worth following: a tap that connected while five machines sat ticked would open a terminal on top of a selection somebody was halfway through building. Unticking the last host leaves the mode, so there are two ways out of it and the cross is only one of them. Both gestures now read the row from the element under the finger rather than from the list's selection, and that is a correctness change rather than tidying. A tap on a group heading moves the selection and the view model bounces it straight back to whichever host was chosen before — which answered "a host, or nothing" for free while a tap only ever connected. It stops answering it the moment a tap can tick one: the heading would tick a machine the user was not pointing at, into a set they are about to delete. Three of the seven entries are about one machine and are drawn only for one. A terminal, a file-transfer session and a form each have no reading over six, so they are collapsed rather than refused. The other four read better for a count than without one — it is the reason the set exists — and each of them says afterwards how many hosts it wrote and how many it left alone. Skipping beats refusing the whole run: a selection of eleven with one read-only row would otherwise do nothing at all and then report about the wrong ten. Copy to vault and Duplicate are new, and the difference between them is what each can safely carry. A copy crosses a key boundary, so it drops the group and the tags exactly as a move does — both are items of the vault being left, and a host arriving with either would point at something the destination does not contain, resolvable on the machine that sent it and dangling for everybody else. A duplicate stays in the same keychain, so everything it points at is still there and it keeps both. Change group is the write dragging a card onto a group already makes on the desktop, run over a selection; it refuses one spanning two keychains rather than half-filing it, which is the refusal a drop across that boundary already makes one host at a time. Connect via SFTP is the one action that leaves the vault. Which machine is a decrypted item and so is this object's business; the screen it leads to and the transfers view model behind it are the shell's — so it is an event, on the same division SessionOpened already draws for a shell. The host is re-found in that screen's own copy of the list, because the picker binds to rows in that copy and handing it the vault's object would select nothing. What is left of the card is the password box, and only because it had nowhere else to go: a host that authenticates with a typed password cannot be reached by a tap alone. That tap now raises a sheet rather than the bar, and the difference is that a sheet is up only while a question is on screen — the bar was raised by a long press and stayed, so it was a password box sitting over the list whether or not anything was being asked. Dismissing it empties the box, which is not tidiness either: a secret left behind would satisfy the emptiness check that decides whether to raise the sheet at all, so the next tap would dial with somebody else's password. The pencil moving into that bar takes the host editor with it. It was a card in the list's own row, under the search box and the sync line — twenty controls sharing a screen with two rows of chrome about the list it had replaced. It is a page now, and PhoneShell stands all four of its rows down for it, which is what "opens with all the options" means at 360dp. That needed a second subscription in that control: two of its flags are questions about the vault rather than about the shell, and the shell does not forward the vault's notifications. The ticks are held as entity ids rather than as rows, and written back onto the rows after every reload. Every row object in the list is replaced on every filter keystroke and every synchronisation pass, so a set of rows would empty itself once a minute under somebody choosing what to do with eleven machines. Ids that no longer resolve are dropped, so a colleague's deletion arriving mid-selection leaves a count that matches what is on screen. One caller had to change with it. ConnectToRecent opened the pane about a host, which was the desktop's drawer and the phone's card; the phone's answer is now a tick, and nothing on that list means "selected" any more — so arriving with the host merely selected would be arriving at a screen with nothing to press. Both are raised together, and the one the head in front of the user does not draw is inert.	2026-08-06 09:15:37 +02:00
jaap-jan	69858f82d1	Merge branch 'claude/vault-key-sync-sharing-d098aa'	2026-08-06 07:39:32 +02:00
jaap-jan	185790fb14	Let a key move to another vault, and ask whether it goes with the host ... Keys sync and keys are shared: SshKey is in the sync registry on both sides, the material rides in the sealed payload, and every generation of the vault key is wrapped to a new member. What was missing was the way in. Hosts and groups could move between vaults and keychain items could not, so a key typed into a personal vault before the team existed stayed there for good — and moving a host into the team's vault left it authenticating with something nobody else in that vault can read. The code said so and could do nothing about it: "the answer is usually to put a copy of that key in the destination vault", which meant pasting the private half into a second item and deleting the first. A private key on a clipboard, and two items nobody can tell apart afterwards. MoveAsync already existed on the generic repository and is now exposed for keys and passwords as it is for hosts and groups. What had to be built around it is the re-aim. An item re-sealed under another vault's key lands with an id of that vault's making, so every host bound to the old one and every group lending it as a default is left naming a tombstone — and a host bound to something its vault no longer holds refuses to connect rather than falling back to a typed password. A move without the re-aim would look like a success and break every machine on that key. It runs over every vault this session can write to, because a binding resolves across all of them, and it counts what it could not rewrite: an item from a newer client, or one in a vault this account may only read. Those are said in the sentence afterwards rather than swallowed. The host's move asks the question rather than deciding it. A binding resolves across vaults, so the moved host goes on working for the person who moved it whichever way this is answered; it is the colleagues they have just joined who hold one vault's key and cannot connect with a host whose key stayed behind. Unticked, and it stays that way on purpose: moving a key into a team's vault hands it to everybody holding that key, and this design does not default anybody into a disclosure. Under the box is the count of everything else that authenticates with that key, because a key twenty machines use is a different decision from one nothing else touches, and neither number is visible from the panel otherwise. The question is answered against the vault in the picker, so choosing a different destination re-asks it and a key already in the destination offers nothing. One thing fixed on the way. A host that inherited its key from its group arrived in the destination naming nothing at all — the group belongs to the vault it left — so a machine that connected before the move refused after it, with no sentence anywhere saying why. The resolved binding is now written onto the host as it crosses, and the stranded-binding warning reads the resolved binding too, which is the case where somebody is least likely to know a key is involved. MOVE is on both heads, for keys and passwords only: a tag, a bucket and a pin are read from the active vault alone, so "another vault" is not a question any of them has. Four tests cover the move and its re-aim, the host's move with the key brought and without it, and the inherited binding.	2026-08-06 07:39:15 +02:00
jaap-jan	cddfeb1f55	Keep the phone's nav under Connections when nothing is running ... The chrome stands down for a shell, and it was standing down for the whole terminal surface. Those parted company when that surface gained a connect page: with no tabs open it draws a box, a CONNECT button and the machines connected to before, which is a page in everything but which enum it is in. A third of the display is worth giving to a shell and is not worth giving to that. Worse, it is the one screen somebody arrives at by closing their last tab — so the state the collapsed bar was most likely to be seen in was the state where it left the system back gesture as the only route to Hosts or Settings. So RefreshChrome reads one more question. IsTerminalSurface with no tabs joins the pages in both flags, which keeps the rail and the bar in step: above 600dp the rail is the bar, and fixing only the narrow layout would leave an unfolded device on the same screen with the same nothing. The vault header is deliberately not part of it. The surface draws its own bar with back and the +, and a header above that is the second row of chrome this head exists to avoid. The Connections entry lights for the first time, on IsTerminalSurface. It was left unbound on the argument that the bar was never drawn while that surface was up, so a lit state was unreachable — that argument is now false, and the flag is unambiguous on a control that is only drawn in two situations: false on every page, true on the connect page, and never read while a shell is showing. A bar sitting under a screen it does not point at is the entry looking broken instead. Nothing here is testable on this head — the phone's rectangles have no coverage, for the reasons Phase 8 of manual-checks records — so 11.7 gains the check that the bar is there with Connections lit, and 11.1 keeps the one that it is gone with a shell up, which is the half that pays for the arrangement.	2026-08-06 07:38:31 +02:00
jaap-jan	174ef7c420	Merge branch 'claude/android-release'	2026-08-05 22:49:51 +02:00
jaap-jan	af0e29a98b	Give the desktop a nightly channel, the way the phone has one ... ADR 0014 gave the phone a nightly and ADR 0013 rule 3 gave the desktop none, so the two heads had different answers to the same question — how does somebody try what is on main? — for no reason except the order the work happened in. This is the desktop's answer: CI publishes a build from main on every push, and it installs beside the release one rather than over it. The phone gets its separation from the platform. Android refuses an update signed by a different key, so its two channels cannot replace one another whatever anybody does. Nothing refuses anything here: Velopack applies what its feed serves and verifies no signature. So all of it is construction, and there are four separations because each closes a different door. A pack id each, so the two install in different directories and neither feed's package can be applied to the other's install. A Velopack channel each — win and win-nightly — so neither build ever reads the other's release index; the name reaches the wire as releases.win-nightly.json, which is why the constant in VelopackUpdateChannel and the argument in ci.yml have to agree or the channel answers nothing forever with no error. A prerelease flag, so the release channel cannot see the nightly even by accident. And a profile directory each, which is the one that is easy to skip and would hurt most: the cache schema is migrated on every launch, before unlock, so a shared profile means a nightly quietly upgrading a database the release build then opens. Both are installed at once by design, so that is an ordinary Tuesday rather than a corner case. The prerelease flag turns out to be load-bearing across heads as well. The phone's release channel reads releases/latest, which skips prereleases — so a desktop nightly published as a stable release would become the newest release in this repository and every phone on the release channel would start failing its check against a release carrying no Android manifest. Which build this is arrives as assembly metadata, the same mechanism and the same reasoning as the Android head: the updater needs the string rather than a branch, and a value baked into the assembly is one a crash report can be asked for. Three things read it — the feed, the prerelease flag, and the profile — and one more shows it: the titlebar says DodoSSH Nightly. Everything else that distinguishes the two is somewhere nobody is looking while typing a passphrase into one of them. The version needed a floor and it is applied to the whole build rather than to the packaging. MinVer answers 0.0.0-alpha.0.N until the first v* tag and vpk refuses anything below 0.0.1, so the job lifts the patch digit and keeps the height — through MinVerVersionOverride, so the assemblies carry the same number the installer does. Packing a version the assembly disagreed with would put one string on the preferences screen and another in the feed, which is the screen somebody reads when asked which nightly they are on. Two things found by running it rather than reading it. -t:MinVer needs a restore first, because the target arrives with the package and MSB4057 on a clean checkout reads like a typo in the workflow rather than a missing restore; the release script had the same gap and now restores before it reads. And vpk rejects an empty --packVersion loudly, which is how a broken version handoff announces itself rather than shipping a package called 1.0.0. Rule 3 is untouched. The release channel still has no job, no token and no runner, and the two channels cannot see each other. What a nightly costs is written where somebody reads it before installing one: whoever can write a release here can put a build on every nightly machine, which is fine for a build being tried and is not fine for a build holding somebody's infrastructure credentials. Verified by running the job's own steps against a clone in a Linux container: DodoSSH.Desktop.Nightly-win-nightly-Setup.exe, and an index naming pack id DodoSSH.Desktop.Nightly at 0.0.1-alpha.0.144. The upload itself is the one step not exercised — it needs a real forge and a write token, and check 16.10 is what walks the half no runner can.	2026-08-05 22:35:51 +02:00
jaap-jan	8591035170	Merge branch 'claude/pipeline-curl-not-found-97e70f'	2026-08-05 22:31:42 +02:00
jaap-jan	ca7fee2358	Start the confirmation Android hands back, so an update can install ... Pressing INSTALL closed the application, installed nothing and said nothing. That is two independent faults in one method, either of which breaks it on its own, and they hid each other: the first kills the process before the second can be observed, and the second is silent by construction. The pending intent handed to commit was implicit — an action string with no component behind it. A mutable pending intent may not wrap one of those from API 34, and this head targets 36, so every current phone threw IllegalArgumentException before commit was reached. Nothing caught it, so it left the command handler, passed the dispatcher and took the process with it. That is the closing. Below 34, where it did not throw, it still installed nothing. An application holding REQUEST_INSTALL_PACKAGES rather than the privileged INSTALL_PACKAGES gets no verdict back from a commit: what the platform answers first is STATUS_PENDING_USER_ACTION, carrying the activity that draws the dialogue in EXTRA_INTENT for the application to start. Android does not draw it on its own. The comment here asserted the opposite — that a pending intent is required whether or not anything listens, and that nothing needed to — so no receiver was ever written, and the session was written, committed and left staged forever. So there is a receiver now, not exported because the only sender is this application's own commit, and the intent naming it is explicit, which is the same change that stops the throw. Sessions are abandoned when anything fails, since one created and neither committed nor abandoned stays staged against a per-application cap — a repeating fault would have started failing at CreateSession instead, which is the same bug wearing a completely unrelated face. The reporting is the part worth keeping even after the cause is gone. Where applying ends the process an exception has nowhere to go; where it does not, which is this head's whole shape, it goes out through the dispatcher. RestartNowAsync now answers the way CheckNowAsync already did, and the regression test asserts the absence of a throw rather than the presence of one. ADR 0014 rule 6 gets the correction in place: "asks Android to ask" is one step longer than it reads. Check 17.5 needed no rewording — it asks for the installer appearing by name, which is exactly the thing that never happened — so what it gets instead is the two symptoms named, because both present as a dead button. It is the only thing in the project that can catch either, and it plainly was never run against a real pair of builds. Note for whoever takes the next nightly: a broken updater cannot install its own fix. The phone is running the code this commit replaces, so the first build carrying it has to be sideloaded by hand; the ones after that install normally. Compile-verified and manifest-verified — the receiver reaches the generated manifest — and 321 tests pass. Not run on a device, which is what 17.5 is for.	2026-08-05 22:31:33 +02:00
jaap-jan	3d3d0bc95f	Package the Windows client in CI, on the runner that could not ... The build job published a win-x64 tree and stopped there, so the half of a release that fails in ways a compile cannot see was proved by nobody until a person was midway through cutting one. It now packs as well: vpk opens the published binaries and verifies the main executable really calls VelopackApp.Build().Run(), which is the check worth having — a refactor that drops that call compiles, tests green, and produces an application that silently never updates itself. The file said this was impossible on Linux, and also said it was fine, in comments forty lines apart. The claim that vpk needs Windows tooling to stamp the Setup.exe stub is the one that was wrong: vpk cross-compiles when told to, and the telling is a bracketed directive before the verb rather than a flag. Plain `vpk pack --runtime win-x64` on a Linux host refuses outright and says so in the message that names the fix. `[win]` must be quoted, or the shell reads it as a glob matching any one of w, i and n. Only signing needs Windows, and nothing here is signed yet. Fixing that does not move ADR 0013 rule 3 an inch, which is why the two reasons were recorded separately in the first place. What may not live on a runner is the token, not the build: Velopack clients apply what their feed serves without verifying a signature, so whoever can write a release can ship an update every install runs. The packages go to RUNNER_TEMP and die with the job. They are not offered as workflow artefacts either — an installer nobody has run should not sit somewhere that invites passing it on. Written out as shell rather than by calling scripts/release-windows.ps1. That script is a person's procedure and holds things a runner must not have and must not skip: it refuses a dirty tree, insists HEAD is tagged, downloads the previous release for deltas, and asks for the forge token. Calling it would mean either weakening it with CI switches or having CI satisfy conditions that only make sense at a desk. The constants the two now share — pack id, title, authors, channel, icon — are a contract with VelopackUpdateChannel and with every installed client, and both sides say so. Two things fell out of running the steps rather than reading them, and both were in code nothing had ever executed: dotnet msbuild -getProperty:Version answers 1.0.0. Without a target named it evaluates the project and runs nothing, and MinVer computes inside a target — so the read comes back as the SDK default on a full checkout with every tag present. That line is the tag check in this file, which is `if:` a tag ref, and there are no tags yet: the first release ever cut would have been refused by its own guard, which would then have blamed fetch-depth. release-windows.ps1 had the same line and would have demanded HEAD be tagged v1.0.0. Both now pass -t:MinVer. And MinVer answers 0.0.0-alpha.0.N until that first tag exists, which vpk rejects outright as below 0.0.1 — so packing the true version could not have worked on any build made today. The patch digit is lifted for the throwaway package only. The release script gets no such floor and must not: its version is the one users compare against, and there the refusal is the right outcome. Verified by extracting both steps from this file and running them against a real clone in a dotnet SDK container: Setup.exe, the portable zip, the .nupkg and releases.win.json, from a machine that is not Windows.	2026-08-05 21:56:45 +02:00
jaap-jan	5cb361ea13	Lay the phone out like the desktop when the surface is not a phone ... Three destinations in a bar and everything else behind SETTINGS is the right shape at 360dp, where a fourth entry costs the width of the three that are there. On a tablet, an unfolded foldable or a landscape phone it is the wrong one: there is room for every destination at once, and the hub becomes an extra tap between somebody and a screen they can already see space for. So at 600dp — Android's own boundary between a compact window and a medium one, in the density-independent units Avalonia lays out in — the bar stands down and PhoneRail takes the left edge with all nine on it. It is the desktop's NavRail arrangement rather than its file: the two heads cannot share a view, and this one draws the phone's destination set with the phone's palette and touch targets. The flags are computed in code rather than assembled in the markup because none of them is a single question any more, and Avalonia's bindings have no "and" — and the header's condition is an "or", which not even a wrapper can express. That header is the one worth reading twice: narrow it stands down behind SETTINGS, so the hub's screens can draw their own; wide there is no hub to be behind, so it stays up everywhere. Losing it on the keychain would be losing the only LOCK button on the surface. Removing the hub means removing the routes into it, and there were four kinds. The rail has no SETTINGS entry, because that screen is a menu of the rail. The five back arrows in the screens under it are hidden, since an arrow to a screen the layout removed is the one control on a header that leads nowhere. The system back gesture goes to Hosts instead. And unfolding while sitting on the hub moves to Hosts, rather than leaving somebody on a list of things now visible beside it. One bug fixed on the way: OnBodyResized returned early unless the keyboard was open, so a foldable would have opened to a phone layout until somebody typed something. The chrome is refreshed first and unconditionally; the early return belongs to the older job below it. What this does not do is use the width *inside* a screen — the host list is one column at any size. Two columns needs the row model to change, because that list is headings and hosts in one sequence and a heading has to span, and that model is shared with the desktop. Check 8.1 walks the rail; nothing here is covered by a test, for the reason 8.0 exists.	2026-08-05 20:50:18 +02:00
jaap-jan	0c61ea3a97	Let a vault be shared from the phone, not only read there ... This screen's own comment argued ADD out: an address typed into a box, a directory lookup, a role picker and a paragraph saying what adding somebody did not do, for an act a colleague at a desktop is already performing. That was a cost argument and it was wrong about who is holding what. The person who needs to let somebody into a vault is often the one away from their desk, and answering them with "go and find a desktop" is the thing this head exists to stop doing. Making a vault was already here on exactly that reasoning. Nothing shared changed — AddMemberCommand, the role and the chips are the same members the desktop binds — so what this is, is markup and the argument it reverses. Four rows under the members list: the box, three role chips rather than a picker because the answer is one of three short words, ADD, and the paragraph. Gated on being able to administer the vault, so a plain member sees nothing rather than a button whose only outcome is a 403. The paragraph is not the optional part. Adding somebody changes what the server will serve and nothing else; the key is still wrapped by a machine that holds one — which on an unlocked phone is this one, in the same press. A screen that offered the first and stayed quiet about the second would imply the server can hand out access, which is the single claim this product is built to refuse. What the phone still does not draw is anything that takes access away. REMOVE and WITHDRAW KEY act on the first press, and an irreversible revocation under a thumb with its explanation in a tooltip no touch screen can show is the wrong trade — which is the line this file already drew and this does not move. Check 12.4 walks it, including the locked-keychain case: the membership is made and the line says the key could not be wrapped, which is a state somebody can act on rather than silence.	2026-08-05 19:10:25 +02:00
jaap-jan	dc1ebf6afa	Let the recovery code be copied, and give the phone a clipboard to copy to ... Both screens had made the code selectable and both said why: a person who cannot get it out of the box photographs the screen, and a screenshot is a worse home for it than a clipboard. This finishes that argument. Selecting 64 characters of letter-spaced monospace with a thumb is the version of "possible" people give up on halfway — and on the phone the screen blocks screenshots, so the honest remaining options were retyping it or losing it. It is the one secret this application deliberately offers to a clipboard, and the contrast with the keychain's copy is the whole argument rather than an inconsistency. There, copying the private half is refused outright, because installing a key means pasting the public one and the private one has no business leaving the vault. Here there is no better route: the code exists for one screen, is stored nowhere, and has to reach a password manager. The clipboard is the intended destination rather than a way round the design. The sentence afterwards matters as much as the copy, and is asserted: a clipboard is a staging post, this screen is the only place the code exists, and the next thing copied replaces it. Somebody who copies and does nothing has not saved it. The phone had no clipboard delegate at all — the desktop passed one and this head passed null — so COPY PUBLIC KEY on the keychain answered "this machine has no clipboard" on a device that plainly has one. Nothing about that was platform shaped: Android has a clipboard and Avalonia surfaces it through the same TopLevel. Wiring it fixes that copy too. The test fixture built its shell without a clipboard, which modelled the bug rather than the product, so it has one now and the public-key test asserts what lands there instead of the refusal. The refusal keeps its own test, on a shell built without one, because the view model reads the delegate's absence rather than an empty result — and because a button that silently does nothing on this screen is worse than one that refuses.	2026-08-05 18:14:34 +02:00
jaap-jan	253c72d2b7	Name the organisation the repository actually lives in ... It moved to DodoTech-Public, and every address in the product still said DodoTech. That looked like it worked, which is the part worth writing down: Gitea leaves a 301 at the old path and HttpClient follows a redirect on a GET, so both update channels would have kept polling through it. What a 301 does not survive is a POST. `vpk upload gitea` publishes the desktop release by POSTing to that URL, so the stale address would have failed at the one step the whole feature depends on — and a redirect is a thing an operator can delete, which turns "works today" into the same silent outage this session has already spent two commits on. So both channel constants, both release scripts, the workflow's REPO, the image's source label and the curl in phase 16 all name the live path. The local remote too, which had been printing a redirect warning on every push. Measured after the move: the org, the repo and the nightly release all answer 200 anonymously, and that release now carries both assets — the manifest and a 54 MB APK. The upload going through also answers the open question about the reverse proxy's body-size limit, which nothing local could test.	2026-08-05 12:51:19 +02:00
jaap-jan	23f1db9dc8	Stop the terminal's accessory keys taking the keyboard off it ... Ctrl, Esc, Tab, the arrows and the two text-size keys were ordinary Avalonia buttons sitting over a NativeWebView. An ordinary button takes focus on tap, which takes it off the WebView — and the package's own OnLostFocus then calls the adapter's ResignFocus(). So pressing Tab handed the terminal one byte and took the keyboard away from it, and everything typed afterwards went nowhere. What makes it worth more than a one-line fix is the symptom. The row goes on working, because its keys are pressed rather than typed into, so what you see is a terminal that answers the buttons and ignores the keyboard — which reads as the session having died rather than as anything to do with focus. Focusable = false is what a toolbar button is: these keys are an extension of the keyboard, not a place it should go. The focused element then never changes, so nothing resigns and nothing has to be handed back — which matters, because the hand-back is the direction platform-flags already records as the hard one. The flags file gains the phone's half of that entry, and check 11.10 is the measurement: this needs a paired hardware keyboard and there is no test on this head that could stand in for one.	2026-08-05 12:45:17 +02:00
jaap-jan	33d4c3ff48	Check the organisation as well, because in Gitea the organisation wins ... Setting the repository public changed nothing: /api/v1/orgs/DodoTech answers 404, and a Gitea org's own visibility gates everything under it — a public repository inside a Limited or Private org is invisible to anyone not signed in. So 16.0 now checks both, and says which answer means which. It also names the signal that tells this apart from an instance requiring sign-in for everything: /explore/repos answering 200, which this one does, so what is hidden is hidden on purpose rather than by policy.	2026-08-05 12:06:09 +02:00
jaap-jan	e0655dbb31	Look the host list up by name, rather than off a field that is never assigned ... Nightly 0.0.0-alpha.0.133 died before its first frame. The long press I added attached itself in HostsScreen's constructor through the field the Avalonia name generator declares for `x:Name` — and that field is assigned by the generated InitializeComponent, which no view in this repository calls. Every one of them loads its XAML directly. So the field compiles, resolves in the editor, and is null at run time; PhoneShell builds this control on the way up, so the NullReferenceException took the launch rather than the hosts screen. PhoneShell and TerminalScreen both use FindControl, and PhoneShell carries a <remarks> saying exactly this and naming exactly this consequence. I read neither and wrote the field. So the rule is in docs/platform-flags.md now as well. A comment on the control that already got it right is not where somebody writing a new one is looking, which is the whole of why two correct examples and one warning were not enough. And phase 8 opens with "it launches at all". Nothing on this head is covered by a test — no test project, no headless surface — so a view that throws while being built takes the launch with it and no gate anywhere says so. Thirty seconds, and it would have caught this one before it was published.	2026-08-05 12:00:20 +02:00
jaap-jan	9a7e3bbd5c	Let a failed update check say so, instead of reporting good news ... The phone reported every build as current because the release repository is private. Gitea answers 404 rather than 403 for a repo you cannot see, the client reads that address anonymously, and AndroidUpdateChannel caught the failure and returned null — which IUpdateChannel documented as meaning "this build is the latest". The check had never once succeeded on any phone and nothing anywhere said so. Two faults, and the second is why the first lasted. The seam said null was the honest answer for an unreachable channel, on the reasoning that the caller does the same thing either way. That is true of the six-hourly pass and false of CHECK NOW. UpdateViewModel already draws the line correctly — silent on the timer, the exception's message on the button — and it could only ever draw the first half, because nothing was ever thrown at it. The desktop's channel does not catch, so the interface described neither implementation. So CheckAsync throws now, and null means one thing. A release that is reachable but missing its manifest or the APK it names throws too: "you are up to date" about a half-published feed is the same lie in a smaller costume, and the self-healing that argument protected is untouched, since the timer still swallows everything. The precondition is written down where somebody would look, rather than left as a sentence about where a token could live. ADR 0013 §4 already said a private release repository was incompatible with this design; nobody checked which side of it this repository was on. It is one curl, and manual-checks phase 16 now opens with it — pointedly not against /api/v1/version, which answers 200 from a forge that is up whatever is readable on it, and which is what made this look like nothing was wrong. Phone check 17.4 was the one that passed all along. It now presses CHECK NOW with the network off as well as on, because two different answers are the whole of what makes the first one worth reading.	2026-08-05 11:17:44 +02:00
jaap-jan	ca07d63585	Offer to bring the keys an ssh_config points at ... An import that recorded a key path and left every host asking for a password was an import whose result did not connect. The answer to that was a manual paste per key, which is the sort of thing people do once and then stop importing. So there is a tick, and it starts off. With it off nothing changes: an IdentityFile becomes a note and the host asks for a password. With it on, IMPORT reads each host's first IdentityFile out of ~/.ssh, stores it in the vault encrypted like any other key, and binds the host to it. Three things about how it is drawn are load-bearing rather than tidy. It is a default nobody arrives at by accident. The sentence beside it names the directory rather than saying "your keys", because that is what somebody is agreeing to. And nothing is read during SCAN — tick it, read what it says, untick it, and no private key has been opened. This is the only place the application opens key material out of a directory the user did not point at file by file, and the whole of what makes that acceptable is that it took a deliberate press. One vault key per file, however many entries named it: an ssh_config pointing twelve hosts at one id_ed25519 is the ordinary shape, and twelve copies would be twelve things to rotate and eleven to forget. A file whose material is already in the keychain is bound to rather than stored again, which is what makes running the import twice harmless. What cannot be read off a disk is a passphrase, so a protected key arrives without one — and the report under the button names those files rather than leaving a host to fail at connect time with a message about a malformed key. Telling them apart means decoding for OpenSSH's own container, whose cipher name is the first field inside the base64 rather than anything in the armour, and that is the format ssh-keygen has written by default for years. The 88 base64 characters it decodes need 66 bytes, not 64: with the smaller span every protected key came back unprotected, which the tests now pin. A path that is not on this machine leaves its host imported and unbound, exactly as it would have been with the tick off, and is named in the same report. A config carried from another machine is the ordinary case, not an error.	2026-08-05 08:58:28 +02:00
jaap-jan	746711da9d	Let a tap on the phone's host list mean connect ... Choosing a machine raised the connect bar over the bottom of the list: a password box, CONNECT, EDIT, MOVE and DELETE. Five controls in the way of the one thing a tap on a machine's name obviously means. So the gestures split. A tap connects. A long press raises the bar, with all five. The pencil in the phone's header — its only persistent chrome — edits whichever host is chosen, which is the one of the five common enough to be worth a control that is always in the same place. The flag doing it is the desktop's own IsHostPaneOpen rather than a second one. That head made exactly this move when a selection stopped opening its drawer, and the question both are asking is "has somebody asked about this host" — answering it twice is how two heads come to disagree about what a selection means. One tap cannot finish: a host that authenticates with a typed password has nowhere on a list to be given one. That tap raises the bar with the box in it and says so, and a second tap with the box filled in connects. The branch is in the view model rather than in the head, because "can this machine be reached without asking for anything" is the same question the bar's own password box answers, and a copy of it in a view would be a second reading of a binding chain that has one. Two mechanics worth knowing. Avalonia raises Tapped on release whatever the press lasted, so a long press would open the bar and then connect — one touch firing both gestures — which is why HostsScreen tracks the hold and swallows the tap it precedes. And Holding only fires once IsHoldingEnabled is set, so that and the handler are attached together rather than one in markup and one in code. ConnectToRecent now opens the pane rather than selecting the row. On the phone it has to: a selection alone raises nothing now, so going back to a recent machine would land on a screen with nothing to press.	2026-08-05 08:43:33 +02:00
jaap-jan	69bc9e270b	Let a team be joined only by somebody who is already here ... An invitation decided access from an assertion about an address. Everything else in this model decides it from something a person did — an admin naming an account, a key holder wrapping a vault key to a key they verified — and this was the one place a token's email claim was the thing that let somebody in. It was guarded as tightly as that can be guarded: the claim was refused outright on an unverified or absent `email_verified`, with no setting to relax it. But the guard and the risk were the same shape. The whole defence was one boolean sent by a system the deployment does not control. So `POST /teams/{id}/members` is the only way in, and an address with no account is refused with `no-such-account` — which is now the end of the road rather than the signal to invite. Both clients say the remedy: that person signs in here once, which is what creates the account, and then they can be added. The desktop leaves the address in the box, because a message telling you to come back later is one you act on later. Gone with it: the `team_invitation` table, the claim hook in the sign-in path, and `Oidc:EmailVerifiedClaim`, which that hook was the only reader of. Nothing in the server now reads the email claim to decide anything. Pending invitations are dropped rather than converted. Converting one would mean creating a membership because an address matched, which is the property being removed — and an invitation to an address that did have an account here had already been claimed by the hourly sweep, so what is left is offers to people who never arrived. Two tests carry the property rather than the feature: the endpoint inventory asserts the three routes are absent, and the API suite adds an address that has no account, watches the refusal, then signs that address in and checks it joined nothing. Without the second half, a server that merely renamed the deferred path would pass.	2026-08-05 08:28:57 +02:00
jaap-jan	a18ca56fde	Copy the checkout into the container, because a socket is not a shared filesystem ... The container started and could not see the repository: MSBUILD : error MSB1009: Project file does not exist. Switch: src/DodoSSH.Client.Android/DodoSSH.Client.Android.csproj `-v "$PWD:/build"` cannot work here. This runner is itself a container holding the host's Docker socket, so the workspace path it reports — /root/.cache/act/<hash>/… — exists in the runner and not on the daemon's host, which is where Docker resolves a bind source. It finds nothing, creates an empty directory, and mounts that. Nothing about the failure says so. Nothing earlier in this workflow would have caught it either, and that is the part worth keeping: the image job's `docker build` sends its context over the API and Testcontainers mounts nothing, so neither of the two places this repository already used Docker proves a bind mount would work. I read a working daemon as a shared filesystem, and they are not the same claim. `docker cp` goes over the same API and so does not care where the daemon lives. In with the whole checkout, .git included, since MinVer and the versionCode both read it — the repository is well under a megabyte packed. Out with the staged package. The NuGet cache becomes a named volume for the same reason: it lives on the daemon and needs no path either side has to agree on. The two container steps collapse into one, since with a copy in and a copy out there is nothing to be gained by paying for both twice, and scripts/ci-android.sh is now what runs inside — a file that can be read and executed on its own rather than a heredoc inside a workflow. Exercised locally against a real clone, every step as the job runs it: create, cp in, start --attach, cp out, parse the manifest on the outside. package: name='dev.dodotech.dodossh.nightly' versionCode='196' versionName='0.0.0-alpha.0.129' The second run took 2m25s against the first run's 8m, which is the named volume doing its job.	2026-08-05 07:29:53 +02:00
jaap-jan	30a3edb1d4	Build the phone in a container, because this runner cannot build it at all ... The runner is Alpine, and .NET for Android does not work on musl. Not "needs setting up" — the SDK's own MSBuild tasks pull glibc shared objects out of the workload pack into the build process, and a musl-linked dotnet will not load one: error XARLP7000: Error relocating .../libZipSharpNative-3-3.so: __snprintf_chk: symbol not found That is a glibc fortify symbol musl does not implement, reached through a DllImport rather than an exec, so gcompat is no help: it gets a glibc *executable* started, which is a different problem. There is no musl variant of the pack. Everything this job did on the host to make Android work was therefore treatment of symptoms, mine included. The missing aapt2 was present. The "unsupported version" was of a binary that had never run. Both were this one sentence in a different accent, and the loader was the accent, not the sentence. So the toolchain moves into build/android-build.Dockerfile — Microsoft's own sdk:10.0-noble plus a JDK, the Android SDK and the workload — and the job keeps on the host only what the host is good at: checkout, git, publishing. The daemon needed no arranging, since the image job already builds with it and every Testcontainers suite reaches it over the socket. The image is tagged by the digest of the Dockerfile that made it, so on a persistent runner every run after the first is a cache hit, and a change to the toolchain is the only thing that buys a new one. Built rather than pulled: a community image with the Android SDK already in it would put a stranger in the path of a package this project signs and publishes. Eleven lines of apt and sdkmanager is the cheaper trade. Verified end to end in that image against a real clone rather than reasoned about, which after three rounds of reasoning seemed the least I could do. Restore under locked mode, Release build, then SignAndroidPackage: package: name='dev.dodotech.dodossh.nightly' versionCode='195' versionName='0.0.0-alpha.0.128' Signer #1 certificate SHA-256 digest: a9f067877724ddb0fdc04b637fbd5bfb97df753976616f100b48b522e132ba22 which is the keystore in build/. The versionName carries MinVer's height, so the csproj's target fires in the container too, and the manifest the feed publishes parses back on the host. Staging moves from RUNNER_TEMP to artifacts/, which is forced rather than preferred: the package is made inside a container and read outside one, so it has to land under the bind-mounted checkout.	2026-08-04 23:28:47 +02:00