aboutsummaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
...
* smtp: replace smtp-starttls with schema optionMarcin Serwin2023-03-264-55/+53
| | | | | | | | | The "smtp-starttls" options is now ignored in favor of more detailed schema specification, similarly to IMAP. Fixes: https://todo.sr.ht/~rjarry/aerc/149 Signed-off-by: Marcin Serwin <marcin.serwin0@protonmail.com> Acked-by: Robin Jarry <robin@jarry.cc>
* ci: fix lint if notmuch is not installedMoritz Poldrack2023-03-162-3/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | | The linter goes completely crazy when notmuch is not installed and the notmuch build tag is enabled. With an otherwise successful lint, the following output is generated: worker/notmuch/lib/database.go:14:10: could not import github.com/zenhack/go.notmuch (-: # github.com/zenhack/go.notmuch ../../go/pkg/mod/github.com/brunnre8/go.notmuch@v0.0.0-20201126061756-caa2daf7093c/configList.go:9:11: fatal error: notmuch.h: No such file or directory 9 | // #include <notmuch.h> | ^~~~~~~~~~~ compilation terminated.) (typecheck) notmuch "github.com/zenhack/go.notmuch" ^ commands/eml.go:4:2: "bytes" imported and not used (typecheck) "bytes" ^ aerc.go:39:7: previous case (typecheck) case *widgets.AccountView: ^ exit status 1 make: *** [Makefile:73: lint] Error 1 Only include the notmuch build tag in the lint process if it is actually installed. Signed-off-by: Moritz Poldrack <git@moritz.sh> Signed-off-by: Robin Jarry <robin@jarry.cc>
* doc: merge duplicate :filter in aerc(1)witcher2023-03-161-6/+2
| | | | | | | | | | | :filter was given twice in the "MESSAGE LIST COMMANDS" section of aerc(1), both incomplete/wrong. Merge both to reference aerc-search(1) and say it actually *filters* the message list. Fixes: af63bd0188d1 ("doc: homogenize scdoc markup") References: https://todo.sr.ht/~rjarry/aerc/144 Signed-off-by: witcher <witcher@wiredspace.de> Acked-by: Moritz Poldrack <moritz@poldrack.dev>
* main: rename aerc.go -> main.goRobin Jarry2023-03-151-0/+0
| | | | | | | | I like it better that way. And it will not piss me off every time I need to use auto completion. Signed-off-by: Robin Jarry <robin@jarry.cc> Acked-by: Moritz Poldrack <moritz@poldrack.dev>
* config: fix default value of next-message-on-deleteRobin Jarry2023-03-151-1/+1
| | | | | | | | | The default value should be true as stated in the docs. Fixes: b63c93563c62 ("config: use reflection to map ini keys to struct fields") Reported-by: Kirill Chibisov <contact@kchibisov.com> Signed-off-by: Robin Jarry <robin@jarry.cc> Tested-by: Kirill Chibisov <contact@kchibisov.com>
* move/archive: make sure to close existing viewerRobin Jarry2023-03-151-0/+2
| | | | | | | | | | When running :move or :archive with a message viewer opened, the viewer is never closed properly. This leads to resources leak (pager processes left running). Fix that. Reported-by: Kirill Chibisov <contact@kchibisov.com> Signed-off-by: Robin Jarry <robin@jarry.cc> Tested-by: Kirill Chibisov <contact@kchibisov.com>
* openers: support basic shell globbingRobin Jarry2023-03-135-16/+38
| | | | | | | Allow wild cards for MIME types like in filters. Signed-off-by: Robin Jarry <robin@jarry.cc> Tested-by: Bence Ferdinandy <bence@ferdinandy.com>
* openers: add mime type handling for open-linkRobin Jarry2023-03-137-17/+33
| | | | | | | | | Provide a way to configure link openers. Based on the URL mime type: x-scheme-handler/$scheme. Signed-off-by: Kirill Chibisov <contact@kchibisov.com> Signed-off-by: Robin Jarry <robin@jarry.cc> Tested-by: Bence Ferdinandy <bence@ferdinandy.com>
* doc: improve aerc-templates examplesRobin Jarry2023-03-131-2/+5
| | | | | | Adjust some examples with real world use cases. Signed-off-by: Robin Jarry <robin@jarry.cc>
* fswatcher: fix bsd supportRobin Jarry2023-03-106-20/+14
| | | | | | | | | | | | | | | Fix the following error when running maildir on freebsd: could not create file system watcher: Unsupported OS: freebsd Do not register based on os type. Register based on supported API. Rename linux -> inotify and darwin -> fsevents. Only build fsevents on darwin, and inotify on all other platforms. Fixes: a0935a3de0ce ("worker/lib: implement an fswatcher interface") Reported-by: Jens Grassel <jens@wegtam.com> Signed-off-by: Robin Jarry <robin@jarry.cc> Tested-by: Jens Grassel <jens@wegtam.com>
* account: fix crash/race on initRobin Jarry2023-03-091-0/+3
| | | | | | | | | | | | | | | | | | | | | | | | | | When an account is initialized too quickly, setTitle may be called before acct.tab is initialized, leading to a crash: runtime error: invalid memory address or nil pointer dereference goroutine 1 [running]: panic({0xa49820, 0x1059510}) runtime/panic.go:890 +0x262 git.sr.ht/~rjarry/aerc/lib/ui.(*Tab).SetTitle(...) git.sr.ht/~rjarry/aerc/lib/ui/tab.go:37 git.sr.ht/~rjarry/aerc/widgets.(*AccountView).setTitle(0xc000838000) git.sr.ht/~rjarry/aerc/widgets/account.go:617 +0x2a0 git.sr.ht/~rjarry/aerc/widgets.(*AccountView).onMessage(0xc000838000, {0xbffc60?, 0xc0000f1080?}) git.sr.ht/~rjarry/aerc/widgets/account.go:364 +0x13e8 git.sr.ht/~rjarry/aerc/widgets.(*Aerc).HandleMessage(0xc0000f4000, {0xbffc60, 0xc0000f1080}) git.sr.ht/~rjarry/aerc/widgets/aerc.go:174 +0x5f main.main() git.sr.ht/~rjarry/aerc/aerc.go:252 +0x965 Avoid this. Fixes: 6b39c0dae1e1 ("tabs: use template for account tab name") Signed-off-by: Robin Jarry <robin@jarry.cc>
* goflags.sh: silence errorsRobin Jarry2023-03-091-1/+1
| | | | | | | No need to print errors when notmuch.h is not found. The only point of this script to check if it is available or not. Signed-off-by: Robin Jarry <robin@jarry.cc>
* doc: add missing references to other manpagesTom Schwindl2023-03-091-2/+3
| | | | | | | | aerc(1) is most likely the entry point for a user who wants to read the docs. Thus, reference all other manpages to improve accessibility. Signed-off-by: Tom Schwindl <schwindl@posteo.de> Acked-by: Robin Jarry <robin@jarry.cc>
* templates: add conditional style methodRobin Jarry2023-03-084-1/+25
| | | | | | | | | | | | | Add .StyleSwitch that takes a string and an arbitrary number of cases (regexp, style). Reuse some of the constructs introduced by previous commit. The style of the first regular expression that matches will be applied. If the string does not match any of the expressions, it will be left as-is. Signed-off-by: Robin Jarry <robin@jarry.cc> Acked-by: Tim Culverhouse <tim@timculverhouse.com>
* templates: add switch/case functionsRobin Jarry2023-03-083-0/+45
| | | | | | | This allows much shorter templates. Signed-off-by: Robin Jarry <robin@jarry.cc> Acked-by: Tim Culverhouse <tim@timculverhouse.com>
* templates: add match functionRobin Jarry2023-03-083-0/+40
| | | | | | | | | | | | | | Add a match function that returns true if a string matches the provided regular expression. The compiled regular expressions are cached in a sync.Map to avoid repetitive compilations of the same expressions. Caveat: if the user mixes the arguments order, it may cause the cache to consume a lot of memory filled with garbage regular expression objects. Ideally, we would need to limit the size of this cache and/or use a LRU cache implementation. Maybe someday? Signed-off-by: Robin Jarry <robin@jarry.cc> Acked-by: Tim Culverhouse <tim@timculverhouse.com>
* templates: reverse args order for .StyleRobin Jarry2023-03-082-2/+2
| | | | | | | | In preparation for .StyleMatch put the content first, followed by the style name. Signed-off-by: Robin Jarry <robin@jarry.cc> Acked-by: Tim Culverhouse <tim@timculverhouse.com>
* templates: make SubjectBase official apiRobin Jarry2023-03-084-1/+16
| | | | | | | | Document .SubjectBase. Fix its value along the way. Do not reuse d.Subject() which may be empty if d.threadSameSubject is true. Signed-off-by: Robin Jarry <robin@jarry.cc> Acked-by: Tim Culverhouse <tim@timculverhouse.com>
* templates: separate thread prefix from subjectRobin Jarry2023-03-088-16/+27
| | | | | | | | Extract {{.ThreadPrefix}} from {{.Subject}} so that the prefix can be styled in a different color. Signed-off-by: Robin Jarry <robin@jarry.cc> Acked-by: Tim Culverhouse <tim@timculverhouse.com>
* msglist: fix inconsistent thread subject deduplicationRobin Jarry2023-03-081-1/+11
| | | | | | | | | | | | | | | | | | | | | | | | | | | | When threading is enabled, some messages which have different base subjects are hidden whereas some others which have similar subjects are displayed: [PATCH aerc v3 1/3] mk: remove smart rebuild when GOFLAGS have changed ├─>[PATCH aerc v3 2/3] mk: only install changed/missing files │ └─>Re: [PATCH aerc v3 2/3] mk: only install changed/missing files │ └─>Re: [PATCH aerc v3 2/3] mk: only install changed/missing files ├─> └─>Re: [PATCH aerc v3 1/3] mk: remove smart rebuild when GOFLAGS have changed This happens because we are using the "previous" message to get the subject base instead of the current one. Compute the base subject from the current message. Here is the result: [PATCH aerc v3 1/3] mk: remove smart rebuild when GOFLAGS have changed ├─>[PATCH aerc v3 2/3] mk: only install changed/missing files │ └─> │ └─> ├─>[PATCH aerc v3 3/3] mk: speed up notmuch detection └─>Re: [PATCH aerc v3 1/3] mk: remove smart rebuild when GOFLAGS have changed Fixes: 535300cfdbfc ("config: add columns based index format") Signed-off-by: Robin Jarry <robin@jarry.cc> Acked-by: Tim Culverhouse <tim@timculverhouse.com>
* templates: avoid error when using .Style with unknown style nameRobin Jarry2023-03-081-1/+4
| | | | | | | | | | | | | | | | | When using {{.Style "foo" bar}} and "foo" is not defined in the styleset under the [user] section, there is an obscure error: template: column-right:1:18: executing "column-right" at <.Style>: error calling Style: runtime error: invalid memory address or nil pointer dereference Which is most of the time truncated and can only be read in logs. Referencing an undefined style is not a serious problem. Return the default tcell style in that case and avoid the error. Signed-off-by: Robin Jarry <robin@jarry.cc> Acked-by: Tim Culverhouse <tim@timculverhouse.com>
* templates: allow layered fg & bg color for inline stylesRobin Jarry2023-03-081-5/+7
| | | | | | | | If a user style has no fg and/or no bg color defined, use the color from the underlying style. Signed-off-by: Robin Jarry <robin@jarry.cc> Acked-by: Tim Culverhouse <tim@timculverhouse.com>
* contrib: fix sendemail validate scriptRobin Jarry2023-03-083-10/+29
| | | | | | | | | | | | | | | | | | The current sendemail-validate script cannot be used for patch series. It works on each patch independently from the others and make it impossible to use when there are inter dependencies. I have submitted an alternate validate script that works on whole series which is still waiting for reviews. In the meantime, propose to use my script which can work. To install it in your environment: curl -Lo ~/.local/bin/git-send-email \ https://paste.sr.ht/blob/cedcf24383022949c8dc5bc1300cf5cbc879b36f echo 'export GIT_EXEC_PATH=$PATH:$(git --exec-path)' >> ~/.profile Link: https://lore.kernel.org/git/20230103231133.64050-1-robin@jarry.cc/ Link: https://paste.sr.ht/~rjarry/7e9a98866ba2c6f34e4169debf4f2c14b574613e Signed-off-by: Robin Jarry <robin@jarry.cc>
* mk: use fixed versions for gofumpt and golangci-lintRobin Jarry2023-03-081-3/+5
| | | | | | @latest is flaky and does not resist GOPROXY hiccups. Signed-off-by: Robin Jarry <robin@jarry.cc>
* statusline: avoid races when updating status stackTim Culverhouse2023-03-071-0/+10
| | | | | | | | | The statusline stack is written and read from separate goroutines without a lock. Make the statusline thread-safe by adding locks for access to the stack. Signed-off-by: Tim Culverhouse <tim@timculverhouse.com> Acked-by: Robin Jarry <robin@jarry.cc>
* notmuch: use fswatcher to trigger directory count updateTim Culverhouse2023-03-073-13/+37
| | | | | | | | | | | Use fswatcher to watch the underlying notmuch db to trigger directory updates. Remove unused argument in the handleUpdateDirCounts function. Call this function when listing directories to initialize counts when starting aerc. Signed-off-by: Tim Culverhouse <tim@timculverhouse.com> Tested-by: Ben Lee-Cohen <ben@lee-cohen.com> Acked-by: Robin Jarry <robin@jarry.cc>
* notmuch/checkmail: simplify check-mail functionTim Culverhouse2023-03-071-19/+9
| | | | | | | | | Simplify the check-mail function in the notmuch worker. Add account name to the logs. Signed-off-by: Tim Culverhouse <tim@timculverhouse.com> Tested-by: Ben Lee-Cohen <ben@lee-cohen.com> Acked-by: Robin Jarry <robin@jarry.cc>
* fswatcher: add a darwin fswatcher implementationTim Culverhouse2023-03-075-0/+97
| | | | | | | | | | Add a darwin implementation of FSWatcher using the fsevents package. The implementation is behind a darwin build flag. Co-authored-by: Ben Cohen <ben@bencohen.net> Signed-off-by: Tim Culverhouse <tim@timculverhouse.com> Tested-by: Ben Lee-Cohen <ben@lee-cohen.com> Acked-by: Robin Jarry <robin@jarry.cc>
* maildir: use FSWatcher interfaceTim Culverhouse2023-03-072-13/+8
| | | | | | | | Use the FSWatcher interface to create the watcher. Signed-off-by: Tim Culverhouse <tim@timculverhouse.com> Tested-by: Ben Lee-Cohen <ben@lee-cohen.com> Acked-by: Robin Jarry <robin@jarry.cc>
* worker/lib: implement an fswatcher interfaceTim Culverhouse2023-03-073-0/+113
| | | | | | | | | | | Implement an FSWatcher interface. The interface is used to abstract away file system watchers, which have implementation specific backends. The initial interface has one implementation: inotify for linux. Subsequent commits will add a macOS watcher. Signed-off-by: Tim Culverhouse <tim@timculverhouse.com> Tested-by: Ben Lee-Cohen <ben@lee-cohen.com> Acked-by: Robin Jarry <robin@jarry.cc>
* change-tab: document "focus previous tab" functionalityinwit2023-03-071-2/+3
| | | | | | | | | | The current :change-tab command allows to be called with a single "-" as a parameter, in which case it changes focus to the previously focused tab. This is useful to switch back and forth between a pair of tabs, and is undocumented. Document this functionality. Signed-off-by: inwit <inwit@sindominio.net> Acked-by: Robin Jarry <robin@jarry.cc>
* ipc: stay silent if command succeededMoritz Poldrack2023-03-072-3/+2
| | | | | | | | It would be nicer if aerc behaved the same way most other tools behave in that no news is good news. Signed-off-by: Moritz Poldrack <git@moritz.sh> Acked-by: Robin Jarry <robin@jarry.cc>
* ipc: allow triggering internal commands via IPCMoritz Poldrack2023-03-077-0/+34
| | | | | | | | | | | | | | | | | | | | | In order to make automation easier, it's useful to be able to send commands to aerc via IPC. This can be done by calling the aerc binary followed by a colon and the command to run. For example: aerc :read && aerc :remark && aerc :archive month Security to ensure no malicious commands are run is deferred to the user. By default the socket is only writable by the user. This is considered sufficient as the potential harm an attacker gaining write-access to a user's session can cause is significantly greater than "can delete some emails". To ensure users with an according threat model, it is possible to disable command IPC. mailto-handling is unaffected even though it works over IPC as it is absolutely non-destructive. Signed-off-by: Moritz Poldrack <git@moritz.sh> Acked-by: Robin Jarry <robin@jarry.cc>
* ipc: use an interface instead of a function listMoritz Poldrack2023-03-073-10/+14
| | | | | | | | | Currently, every function has to be mapped to the according handler with our god-object. To make adding new handlers require less changes, change this mapping into an interface that is satisfied by *widgets.Aerc Signed-off-by: Moritz Poldrack <git@moritz.sh> Signed-off-by: Robin Jarry <robin@jarry.cc>
* ipc: change protocol to JSONMoritz Poldrack2023-03-074-36/+108
| | | | | | | | | | | | | | | | | | | | | | In overhauling the IPC, it has become necessary to switch to a more extendable message format, to ensure more complex commands can be sent. Messages have the following basic structure and must not contain linebreaks, as these are used to delimit separate messages from one another. {"arguments": ["mailto:moritz@poldrack.dev"]} The responses have the following structure: {"error": "epic fail"} If the IPC request was successful, "error" will be empty. {"error": ""} Signed-off-by: Moritz Poldrack <git@moritz.sh> Signed-off-by: Robin Jarry <robin@jarry.cc>
* socket: refactor existing codeMoritz Poldrack2023-03-073-146/+147
| | | | | | | | | | | | | There are several //TODO comments in the socket package, these should be fixed before expanding it. Put send logic into it's own file and rename receiver code. Fix the rather inelegant error handling when shutting down the server. Make sure to close sockets. Signed-off-by: Moritz Poldrack <git@moritz.sh> Signed-off-by: Robin Jarry <robin@jarry.cc>
* socket: extract functionality into ipc packageMoritz Poldrack2023-03-072-5/+5
| | | | | | | | | | | | The current model of files dumped into lib does not follow the general advice not to have "util" classes. Naming the package "lib" does not change that it is a random assortments of functions that have some utility. Extracts socket functionality into it's own package. Signed-off-by: Moritz Poldrack <git@moritz.sh> Acked-by: Robin Jarry <robin@jarry.cc>
* mk: speed up notmuch detectionRobin Jarry2023-03-064-31/+28
| | | | | | | | | | | | | | | Use gcc instead of go to build a basic program and determine if notmuch is available. Building a minimal go program takes more than 300ms on a fast machine. A minimal C counterpart takes less than 100ms. To avoid lag when doing bash completion, avoid running any shell commands directly during make evaluation. Rename check-notmuch.sh to goflags.sh and make that script print the goflags directly. Signed-off-by: Robin Jarry <robin@jarry.cc> Tested-by: Inwit <inwit@sindominio.net>
* mk: remove smart rebuild when GOFLAGS have changedRobin Jarry2023-03-063-18/+4
| | | | | | | | | | | This was a nice idea but it is very annoying since the flags change every time the git commit changes even if there are no changes to the source files nor the GOFLAGS... Building with different GOFLAGS now requires cleaning first. Signed-off-by: Robin Jarry <robin@jarry.cc> Acked-by: Moritz Poldrack <moritz@poldrack.dev>
* table: fix center paddingRobin Jarry2023-03-042-14/+29
| | | | | | | | | | | When a column has ALIGN_CENTER and the number of white space character of padding is not a multiple of two, the last cell(character) is not padded and that can cause coloring glitches. Make sure to pad all the way. Fixes: 49de9b09cacc ("ui: parse strings for ansi styles") Signed-off-by: Robin Jarry <robin@jarry.cc> Acked-by: Tim Culverhouse <tim@timculverhouse.com>
* mk: detect if notmuch headers & libs are availableRobin Jarry2023-03-033-3/+39
| | | | | | | | | | Detect if notmuch is available by trying to compile a minimal go program that uses the notmuch.h header and links to libnotmuch.so. If that succeeds, set the default GOFLAGS value to -tags=notmuch. Signed-off-by: Robin Jarry <robin@jarry.cc> Tested-by: Inwit <inwit@sindominio.net>
* config: use reflection to map ini keys to struct fieldsRobin Jarry2023-03-0210-541/+465
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The default ini.Section.MapTo() function only handles basic types. Implement a more complete mapping solution that allows: * parsing templates, regexps, email addresses * defining a custom parsing method via the `parse:"MethodName"` tag * defining default values via the `default:"value"` tag * parsing rune values with the `type:"rune"` tag The field name must be specified in the `ini:"field-name"` tag as it was before. It is no longer optional. The `delim:"<separator>"` tag remains but can only be used to parse string arrays. It is now possible to override default values with "zero" values. For example: [ui] dirlist-delay = 0 Will override the default "200ms" value. Also: [statusline] status-columns = Will override the default "left<*,center>=,right>*" value. Signed-off-by: Robin Jarry <robin@jarry.cc> Tested-by: Tim Culverhouse <tim@timculverhouse.com>
* templates: allow inline user stylesTim Culverhouse2023-03-028-0/+66
| | | | | | | | | | | | | | | | | | | | Allow custom user-defined styles in a styleset. The styles can take any name, and must be under the [user] ini section. All attributes apply to user defined styles. Example: [user] red.fg=red red.bold=true Add a .Style function which accepts the name of a user-defined style and applies it to the string. {{.Style "red" "foo"}} Signed-off-by: Tim Culverhouse <tim@timculverhouse.com> Acked-by: Robin Jarry <robin@jarry.cc>
* colorize: stop parsing theme when other section startsRobin Jarry2023-03-021-1/+6
| | | | | | | | In order to allow multiple sections in a styleset, colorize must stop parsing the theme when it encounters a new section. Signed-off-by: Robin Jarry <robin@jarry.cc> Tested-by: Tim Culverhouse <tim@timculverhouse.com>
* ui: parse strings for ansi stylesTim Culverhouse2023-03-025-33/+714
| | | | | | | | | Parse UI strings for ANSI styles. If there are styles in the string, use those as the display style in tcell. This is in preparation for a template function which can apply arbitrary styles to UI elements. Signed-off-by: Tim Culverhouse <tim@timculverhouse.com> Acked-by: Robin Jarry <robin@jarry.cc>
* doc: add missing entries for :sign and :encryptwitcher2023-03-021-0/+10
| | | | | | | | Update aerc(1) with explanations about key selection priority. Fixes: https://todo.sr.ht/~rjarry/aerc/144 Signed-off-by: witcher <witcher@wiredspace.de> Acked-by: Robin Jarry <robin@jarry.cc>
* pgp: automatically determine signer from senderwitcher2023-03-021-12/+31
| | | | | | | | | | | | | | | | | | | | This patch automatically determines the signing account from the sender of the email (i.e. the From email header). To reflect this in the compose view after changing the From field, the restriction in updateCrypto() to only run the update if the signing key is empty has been lifted, so the key always gets updated when calling updateCrypt(). Additionally, a Signer() method has been added to the Composer to avoid specifying the same logic twice; once in updateCrypto() and once in WriteMessage(). If the From header is not populated for any reason the Signer() method falls back to the from field specified in accounts.conf. Signed-off-by: witcher <witcher@wiredspace.de> Acked-by: Robin Jarry <robin@jarry.cc>
* pgp: fix segfault for opportunistic encryptionwitcher2023-03-021-3/+6
| | | | | | | | | | | | | | | | | Commit 2af81a743048 ("pgp: add configurable error level for opportunistic encryption") introduced a bug where if the pgp-opportunistic-encrypt attribute is set to true and a new message is composed, aerc would crash. This is because no recipients have been specified, so checkEncryption() would fail early and *not* call updateCrypto(), which would then cause a segfault in SetEncrypt() later on. To avoid this, when the function would have returned early, it instead does *not* set c.encrypt to true and still calls updateCrypto(). Fixes: 2af81a743048 ("pgp: add configurable error level for opportunistic encryption") Signed-off-by: witcher <witcher@wiredspace.de> Acked-by: Robin Jarry <robin@jarry.cc>
* check-patches: report all errorsRobin Jarry2023-02-281-14/+18
| | | | | | | Instead of skipping to the next patch after the first error, report all issues. Change the output to make it more user friendly. Signed-off-by: Robin Jarry <robin@jarry.cc>
* Revert "ci: use alpine/latest instead of edge"Robin Jarry2023-02-261-1/+1
| | | | | | | | This reverts commit fff5e2f1bbe4e1d2afecada9b69fca0fc7cec424. Should work now. Signed-off-by: Robin Jarry <robin@jarry.cc>
generated by cgit (git 2.34.1) at 2024-10-30 07:24:34 +0000