Many hyperlinks are disabled.
Use anonymous login
to enable hyperlinks.
50 most recent check-ins
|
2024-11-21
| ||
| 07:27 | (net) wire in arcan-net client side parts for get/put file Get lack creating the output file still, but the current blocker is that the server end rejects all transfers. Leaf check-in: e8096cafde user: void tags: master, trunk | |
|
2024-11-12
| ||
| 10:44 | (build) add buildtag for fossil checkout check-in: 7c9be3ab32 user: void tags: master, trunk | |
|
2024-11-04
| ||
| 12:27 | (a12) build fix, rename _ctrl permission and add _adopt entrypoint check-in: 1de6408644 user: void tags: master, trunk | |
| 03:25 | (a12) appl-ctrl update now arrives correctly in config dir check-in: 3c0a34f359 user: letoram tags: master, trunk | |
| 03:15 | (a12) cancel-stream on ctrl works, remove placeholder check-in: 3f0b038d80 user: void tags: master, trunk | |
| 03:03 | (a12) fix to push-ctrl, expose ctrl permission in .lua parsing check-in: eb96b644f1 user: void tags: master, trunk | |
|
2024-11-03
| ||
| 23:00 | (a12) add client end --push-ctrl for updating the controller slot check-in: 31e9dcc819 user: void tags: master, trunk | |
| 22:07 | (a12) patch in (untested) directory appl controller dynamic update This requires a corresponding client-side patch to --push-appl to slot in the corresponding controller update. It should allow us to get a shorter feedback loop before fixing the missing controller API/debugging features as a precond for indexing/search/directory-network parts. check-in: 754f110646 user: void tags: master, trunk | |
|
2024-11-02
| ||
| 15:01 | (a12) add --monitor-appl to show appl-group messaging This runs like arcan-net @dir applname but will not download or chain-exec but rather just forward received messages to stdout for logging, bridging and debug tools. check-in: acd8199660 user: void tags: master, trunk | |
|
2024-10-22
| ||
| 20:27 | (core/shmif) forward acquire fence with shmifext_signal This will add a poll stage on a fence (if one can be exported) before accepting a pending incoming dma-buffer frame. This should be the only remaining structural change for allowing explicit synch (though release- fences is still up for consideration, we have a similar-ish mechanism through STEPFRAME at the moment). check-in: eba9c46d5f user: void tags: master, trunk | |
| 20:24 | (shmif) event logging for cursor-leave check-in: 1cbd5e4f5f user: void tags: master, trunk | |
| 20:23 | (core) don't default mark as dirty when there is a fract timestamp shader This is a complicated remnant from the dirty tracking done before we allowed direct rendertarget mapping to an output sink. The consequence was that if there is a shader requesting fractional timestamp updates in an otherwise static scene, the worldrt or related rendertargets would be skipped but we still reported that there were updates to the platform layer. This caused lwa to swap active rendertargets when there was nothing actually drawn, effectively forwarding 'old' frames. The consequences of this change may be that static single-object 'animations in shader' would only update on ticks. The proper fix then would be to check for each rendertarget if there are active visible objects with shaders that have fractional updates and only then process it. check-in: 006451e16d user: void tags: master, trunk | |
|
2024-10-21
| ||
| 12:01 | (shmif/lua) extend target_input for analog devices with 'leave' hint X11 and Wayland clients rely on explicit 'cursor is no longer on the surface' hints, without which grab semantics and drag and drop might fail to work. This patch adds a backwards compatible form of providing that information. check-in: 1d3e8ecb61 user: void tags: master, trunk | |
| 07:58 | (core) duplicate file descriptor when replacing server-side font for a frameserver This caused a sparse re-use/close during reset when there is a tui enabled connection that gets reconfigured. On some devices this would be the first descriptor reissued when platform_event_deinit() / init() is issued. On some devices this would cause the keyboard to stop working, but the symptoms would be all over the place. check-in: cf1005ae07 user: void tags: master, trunk | |
|
2024-10-15
| ||
| 22:25 | (tui) expand readline widget with callback on current suggestion this is the simplest path for allowing tui clients to forward completion information to an accessibility window. check-in: a7e2f402bd user: void tags: master, trunk | |
| 20:19 | (tui) flag dirty partial on erase-region This caused erase updates without subsequent writes to fail to synch onwards. check-in: b1349a6469 user: void tags: master, trunk | |
|
2024-10-14
| ||
| 10:55 | (tui) synch cursor position on unpack check-in: 796aa1ae83 user: void tags: master, trunk | |
|
2024-10-10
| ||
| 13:02 | (lua) :read should not return a 1-ch str for \0 check-in: d5d3e73ebc user: void tags: master, trunk | |
|
2024-10-07
| ||
| 13:55 | (decode) tts multiline error handling bleed check-in: 86130151f6 user: void tags: master, trunk | |
|
2024-10-05
| ||
| 13:38 | (platform/egl-dri) allow passing DRM_FORMAT_MOD_LINEAR to dma buffer check-in: fb9c74fb7e user: cipharius tags: master, trunk | |
| 13:15 | (platform/headless) adds buffer passing support to headless video platform check-in: 358582e602 user: cipharius tags: master, trunk | |
| 13:13 | (platform/egl-dri) refactors out egl_errstr function check-in: ad2760431f user: cipharius tags: master, trunk | |
|
2024-10-04
| ||
| 09:49 | (build) bump version check-in: a2ab166afb user: void tags: 0.6.3.4, master, trunk | |
| 09:11 | (lua) add :translate to image_access_storage This is useful for moving from mouse coordinates to col, row when reading from the textual representation in a tui- backed surface or frameserver. check-in: 5ba3d9a865 user: void tags: master, trunk | |
|
2024-10-01
| ||
| 23:34 | (shmif/decode) cleanup / nits with multipart- message helper reuse check-in: b2902e8dd2 user: void tags: master, trunk | |
| 22:52 | (decode) switch t2s multipart buffering to shmif_interop.h one check-in: a4ddb9313c user: void tags: master, trunk | |
| 06:49 | (shmif) allow mousestate helper to handle flag for clamping check-in: 26c21bf3cc user: void tags: master, trunk | |
|
2024-09-30
| ||
| 21:37 | (lua) image_tracetag for partial update setting wrong value scope namespace collission caused the wrong alt to be set/retrieve when msg is empty. check-in: 30d9ee2281 user: void tags: master, trunk | |
| 04:05 | (a12) re-merge missed a12loop test case check-in: d4d76bcb29 user: void tags: master, trunk | |
|
2024-09-07
| ||
| 10:59 | (platform) change map_resource to handle pipe input source This was lifted from the defunct android port and isn't really used by the engine proper as nbio.c provides a better solution. check-in: fe9e0a894f user: void tags: master, trunk | |
|
2024-09-06
| ||
| 21:34 | (decode) expose proto=img fdin=0 argument for inherited work this is mainly useful for lash/tui like contexts that can control shmif client spawning environment. check-in: bf0c6ca847 user: void tags: master, trunk | |
|
2024-09-04
| ||
| 23:44 | (build) More pointless Lua 5.4 nits check-in: acdcb88cad user: void tags: master, trunk | |
| 23:33 | (build) swap luaL_typerror for luaL_argerror check-in: 858e155a51 user: void tags: master, trunk | |
| 21:25 | (net/build) fix CLIENT_LBRARIES build profile Lua + nbio deps check-in: e1d23e5488 user: void tags: master, trunk | |
|
2024-09-02
| ||
| 22:29 | (alt/tui) regression in _open on lfch defaulting to \0 The change to allow customized separator ch for the read/strip call did set it correctly for socket accept (hence why it wasn't catched until now), and cat9 explicitly sets it, but for other cases it would fail to provide the simple read() -> line format. check-in: 8d1a5f7b3d user: void tags: master, trunk | |
|
2024-08-30
| ||
| 21:19 | (encode) remove deprecated use of pts/channel-layout check-in: 1b3a475678 user: void tags: master, trunk | |
| 20:17 | (anet) committed before :w check-in: 3b863c8cfe user: void tags: master, trunk | |
| 16:46 | (anet) always ignore on SIGCHLD/SIGPIPE check-in: 1d1a662aee user: void tags: master, trunk | |
|
2024-08-27
| ||
| 08:30 | (tui) adds missing TTF_Init stub check-in: 3a8c742cba user: cipharius tags: master, trunk | |
|
2024-08-26
| ||
| 17:42 | (net) switch to 'ultrafast' preset for h264, treat x11/wayland clients as latency biased check-in: 9c297847d8 user: void tags: master, trunk | |
|
2024-08-25
| ||
| 09:18 | (platform/audio) use the correct audio playback end callback tag check-in: 528d7dadde user: cipharius tags: master, trunk | |
| 09:16 | (lua) adds missing early return for audio buffer loading check-in: c7b787fd9b user: cipharius tags: master, trunk | |
|
2024-08-04
| ||
| 23:08 | (build) update release tag check-in: fb1c0e981f user: void tags: 0.6.3.3, master, trunk | |
| 23:02 | (build) update release tag check-in: df48d9211a user: void tags: 0.6.3.3, master, trunk | |
| 13:53 | (build) update release tag check-in: b3a56bf204 user: void tags: 0.6.3.3, master, trunk | |
| 13:52 | (build) use C_STANDARD property to get the -std argument check-in: cb38ef65ef user: void tags: master, trunk | |
| 10:58 | (build) bump CMakeLists version to remove warning, switch GL to use component syntax check-in: d38f045eff user: void tags: master, trunk | |
|
2024-08-02
| ||
| 07:21 | (build) bump version check-in: 1f305bc466 user: void tags: 0.6.3.3, master, trunk | |
| 07:21 | (shmif/lua) drop VOBJ substructure, using BCHUNK-in/out is better check-in: ef8d5743e2 user: void tags: master, trunk | |
| 07:20 | (tui-lua) expose resolved symlink path for fstatus check-in: 61328078a0 user: void tags: master, trunk | |