Many hyperlinks are disabled.
Use anonymous login
to enable hyperlinks.
50 most recent check-ins
|
2025-09-30
| ||
| 15:28 | (net/a12) PQC ratcheting working with --rekey-pqc Leaf check-in: 020d77a28e user: letoram tags: trunk, master | |
|
2025-09-28
| ||
| 19:07 | (net/a12) add --rekey-bytes and --rekey-pqc option, send encapsulate on hello check-in: 52f7af2d78 user: letoram tags: trunk, master | |
| 15:35 | (a12/net) ML-KEM vendoring build system shenanigans This is rather ugly, but any consumers building a12 by source rather than relying on the public interface and libraries need to specify symbol namespace prefix (mlkem) and security level (768). check-in: a98f016764 user: letoram tags: trunk, master | |
|
2025-09-23
| ||
| 18:40 | (build) vendor mlkem reference implementation check-in: f431bb1036 user: letoram tags: trunk, master | |
| 00:39 | (a12) update with .index search and draft ML-KEM-768 rekeying. check-in: 8d3b5f7675 user: letoram tags: trunk, master | |
|
2025-09-18
| ||
| 21:12 | (net) propagate / handle _close and map to _EXIT Also pull dms in shmifsrv when enqueued. check-in: cd7da84489 user: letoram tags: trunk, master | |
| 20:41 | (net) prep a12 for vbuffer resubmission on encoder stall Also trying out _EXIT on channel_close for remote_role == SOURCE check-in: 5e2eaf44de user: letoram tags: trunk, master | |
|
2025-09-17
| ||
| 10:53 | (build) fixes to hybrid builds from displayinfo bleedover check-in: e517e3b8ee user: void tags: trunk, master | |
|
2025-09-16
| ||
| 23:41 | (net) make sure session exec resets signal masks check-in: 8e2425ce42 user: letoram tags: trunk, master | |
| 16:34 | (tui) don't forward ctrl mod with u8 field to input_utf8 handler but proceed to input_key check-in: 8c1accb6ad user: letoram tags: trunk, master | |
| 15:36 | (net) route events correctly to subsegments This is a partial fix for the issues hosting cat9 over arcan-net. The reason was simply that the wrong shmifsrv instance got the events intended for a different channel. check-in: 0e19a76013 user: void tags: trunk, master | |
| 00:25 | (core) video object rendertarget platform mapping didn't persist across sharestorage check-in: 6c5915119d user: letoram tags: trunk, master | |
|
2025-09-15
| ||
| 13:34 | (lua) stack balance in dispinfo form of video_displaydescr check-in: 727f5dab09 user: letoram tags: trunk, master | |
| 13:30 | (shmif) suggest additional CURSORHINT values for locking controls check-in: 037a94adb1 user: letoram tags: trunk, master | |
| 13:30 | (core/lua) pull in optional libdisplay-info and expose parsed EDID check-in: 03247e98cc user: letoram tags: trunk, master | |
|
2025-09-12
| ||
| 16:52 | (net) arg-parsing fix for discovery check-in: 45ad7efb50 user: letoram tags: trunk, master | |
| 16:52 | (lua) _adopt handler (last=bool) flag wasn't set correctly check-in: 9f8c6e7a90 user: letoram tags: trunk, master | |
|
2025-08-31
| ||
| 18:02 | (misc) resize welcome appl canvas on display reset event. check-in: b77d6f50f5 user: cipharius tags: trunk, master | |
|
2025-08-28
| ||
| 19:27 | (evdev) resolve utf8 correctly for ctrl combos The accompanying path in defhandler_kbd() will only be able to set translated.utf8 correctly, if the code is correctly looked up. Previously this was only hardcoded for ctrl+c with the rest assigning to 0 as per the sparse clut lookup table. The ctrl+a, ..., ctrl+z range maps continuously to 0x01, ..., 0x1a, thus we can easily expand support to most common ctrl combos. To match behaviour with the previous LUT, assign 0 outside this range. This fixes keyboard shortcuts for some native shmif clients that expect the utf8 to be present. check-in: 940c563aeb user: vimpostor tags: trunk, master | |
|
2025-08-27
| ||
| 09:39 | (net) connection resumption shouldn't initiate on client that closed properly libarcan-shmif-server didn't actually check the dms and update client status if the shmif-client exited properly. This lead to session.c thinking that the client was still alive and tried to repair / resume. check-in: e716e7728e user: letoram tags: trunk, master | |
|
2025-08-26
| ||
| 21:18 | (core) getopt_long entry for --scriptpath was missing check-in: bb263d42bd user: letoram tags: trunk, master | |
| 21:00 | (platform) namespace reporting label order wrong check-in: 2e47e43280 user: letoram tags: trunk, master | |
|
2025-08-24
| ||
| 22:32 | (lua) print Lua parser error message on failed loadfile and fix "bytecode forbidden" error message. check-in: 3afacd4a80 user: cipharius tags: trunk, master | |
|
2025-08-23
| ||
| 21:42 | (lua/tools) last round of multipart chunking fixes check-in: 1be40f93ce user: letoram tags: trunk, master | |
| 21:25 | (lua) image_access_storage: expose cursor style check-in: c697e0bd64 user: letoram tags: trunk, master | |
| 19:14 | (tui) only log screensize change when dimensions have changed check-in: 89312714a6 user: letoram tags: trunk, master | |
|
2025-08-22
| ||
| 17:50 | (tests/dnd) add ctrl+drag to announce as drag check-in: 152a24d519 user: letoram tags: trunk, master | |
| 12:22 | (platform) fix broken tilde key event utf8 attribute. check-in: 279f52c4ac user: cipharius tags: trunk, master | |
| 10:56 | (build) update nixpkgs version and add additional dependencies to flake.nix. check-in: d163b0e67c user: cipharius tags: trunk, master | |
| 08:33 | (lua) make text_surface erase the whole row on update, except when explicit x coordinate is provided. check-in: 14fbc5be58 user: cipharius tags: trunk, master | |
| 07:10 | (core/lua) make text_surface respect default font size and provide mm/pt conversion functions to avoid conversion constants being scattered around the codebase. check-in: ab4fcca9e9 user: cipharius tags: trunk, master | |
|
2025-08-21
| ||
| 20:21 | (core/lua) extend image_origo_offset with support to set local anchoring offset This is a complement to anchoring in link_image, which only defines anchoring relative to the new parent. With this you can also adjust which corner is used to position the object within the anchor without manually resolve-property/move and match transforms. check-in: 4132c275ec user: letoram tags: trunk, master | |
|
2025-08-20
| ||
| 21:41 | (core) relinking should invalidate entire hierarchy If there are no transformation scheduled on a object hierarchy and something with children gets relinked, the children retain the valid_cache status preventing recursion further down the chain. This is also a case with the early out from invalidate_cache() where an object with a valid cache would prevent children with an invalid cache to propagate since it assumed it always propagated upwards as is the case with transformations. This will impose a cost on larger hierarchies, so it might be useful to have the early-out for transform walk in regular processing and ignore it during relinking. Discovered-by: Nanderty check-in: b044778592 user: letoram tags: trunk, master | |
| 00:05 | (tui) add :message function and handler to lua bindings check-in: 7e69b6db70 user: letoram tags: trunk, master | |
|
2025-08-19
| ||
| 23:54 | (tui) add local/generic :message domain and route MESSAGE events to handler This is mainly needed for tui appl part of an a12-directory-ctrl as they interop with regular arcan appls. At the same time add a LOCAL message domain so that we can use this to signal between tui contexts. check-in: af4f4fe433 user: letoram tags: trunk, master | |
| 23:46 | (shmif) fix edge-case truncation in utf8 multipart chunking Credit-to: Bohdan check-in: 73cdd2ffa0 user: letoram tags: trunk, master | |
|
2025-08-18
| ||
| 11:41 | (net) reset multipart message offset after broadcast (ctrl- less) check-in: fa1528c364 user: letoram tags: trunk, master | |
| 02:01 | (net) add alias_appl to config.lua This function lets an appl transparently share a controller or message domain with another. The purpose of that is mainly to be able to have a tui form as a complement to a main appl. check-in: b2e030f6d3 user: letoram tags: trunk, master | |
| 01:06 | (doc) updates changelog check-in: 44a3b32510 user: letoram tags: trunk, master | |
|
2025-08-17
| ||
| 23:33 | (net) soft-link resolve swap descriptors as traversal continues check-in: 8506d9f066 user: letoram tags: trunk, master | |
| 10:06 | (platform) launch_target forward resolved system script path This caused arcan_lwa launch_target to fail for some environments check-in: 6285ca12a9 user: letoram tags: trunk, master | |
|
2025-08-15
| ||
| 00:57 | (net) forward referential link endpoint to diropen req Next step should be to just try the connection from dir_cl, fingers crossed that would work and we can move on to trying more complex paths. Otherwise a likely suspect is a kpub being flipped in dir_srv_worker. check-in: 7826d0be6a user: letoram tags: trunk, master | |
|
2025-08-12
| ||
| 15:30 | (decode) svg: respect displayhint by recalculating scale check-in: 4cbbb96106 user: letoram tags: trunk, master | |
|
2025-08-10
| ||
| 22:14 | (net) fix printing dirref as source, start respecting --path This should almost bring the feature to usable, the separate resolver gets running and the DIROPEN request gets routed. What is left is to get server side chain from dirsrv_link to provide the right connection information from its end routed through. check-in: 8dd19dfe41 user: letoram tags: trunk, master | |
| 14:40 | (net) force reference_directory to give a name for dynlist This makes the link-worker send that name back as the NETSTATE post connection. check-in: bd780cc4fc user: letoram tags: trunk, master | |
| 00:22 | (net) add dirref to permitted inbound role if local is directory check-in: c223e2b504 user: letoram tags: trunk, master | |
| 00:15 | (net) split up dir into dir and dirref Another round of prepwork for referential connections. check-in: d0ca1a0bef user: letoram tags: trunk, master | |
|
2025-07-29
| ||
| 18:56 | (net) prepare dircl for soft-link traversal This adds the (nonworking, untested) --path a/b/c argument to arcan-net. This breaks down into a connection to [a] -> waiting for discover of a referenced directory [b] and so on until the path has been fulfilled. The regular cli request is then applied to that server. check-in: b141f5730f user: letoram tags: trunk, master | |
|
2025-07-20
| ||
| 23:25 | (net) dir_cl: permit 'tui' in manifest and route through afsrv-terminal This is the first step towards letting afsrv_terminal act as a runner for tui flagged appls and was tested with directory hosted cat9. It is still dangerous in the sense that the appl can basically popen() whatever it feels like. Just like for desktop type appls that can be ok, but should go with a manifest option. The main purpose of this is to be able to provide an alternate minimalistic appl as a complement to another hosted one, both for a wider computing environment and for accessibility. check-in: 4063874561 user: letoram tags: trunk, master | |
|
2025-07-15
| ||
| 13:20 | (build) openBSD missing includes check-in: 5963fb4205 user: letoram tags: trunk, master | |