416f870 | Tobias Ellinghaus | 08 October 2016, 12:02:18 UTC | Fix a typo in an SQL statement | 08 October 2016, 12:02:18 UTC |
5b5c954 | Roman Lebedev | 07 October 2016, 12:32:41 UTC | pixelpipe_picker_helper(): if sampling just one point, make sure to actually sample it. Unification of codepaths broke it. It previously used <=, but now < is used. So we need to actually make sure that point results in bounding box of non-zero size. Fixes e.g. color picker in color zones iop. Reported-by: houz | 07 October 2016, 12:32:48 UTC |
17047df | Tobias Ellinghaus | 07 October 2016, 10:42:31 UTC | Fix error in lua api docs This was probably a copy&paste error. Note that the XML needs to be regenerated! | 07 October 2016, 10:42:31 UTC |
ac1ad45 | Ulrich Pegelow | 07 October 2016, 04:34:30 UTC | opencl: follow-up or PR1279: this should fix color-picker for opencl | 07 October 2016, 04:34:30 UTC |
e988f75 | Roman Lebedev | 06 October 2016, 21:03:42 UTC | Merge remote-tracking branch 'upstream/pr/1279' Color picker fixes. Spot whitebalance and spot invert was broken by un-demosaicing of preview pipe. This fixes it. According to @upegelow, OpenCL part of the changes is unsufficient. Since i have no opencl to fix it, and this already is much better than what git master currently have, i'm merging it as it is. Fixes #11163. Fixes #11164. * upstream/pr/1279: Invert iop: fix deadlock Invert iop: 4Bayer changes Temperature iop: spot wb: support 4Bayer Color picker: support 4Bayer. pixelpipe_picker_cl(): roi_copy: passthrough proper x,y Color picker: support picking from X-Trans mosaic Color picker: support picking from Bayer mosaic Color picker: refactor computations into worker, properly check input type pixelpipe_picker_cl(): use dt_iop_buffer_dsc_to_bpp() Color picker: take dt_iop_buffer_dsc_t* Pixelpipe: pixelpipe_picker{,_cl}: refactor common setup code into helper Refactor actual color picking into new .[CH] | 06 October 2016, 21:03:42 UTC |
40b6a4c | Roman Lebedev | 06 October 2016, 20:44:32 UTC | Invert iop: fix deadlock | 06 October 2016, 20:45:48 UTC |
f71e2ce | Roman Lebedev | 06 October 2016, 19:56:37 UTC | Invert iop: 4Bayer changes | 06 October 2016, 20:45:48 UTC |
cd2764f | Roman Lebedev | 06 October 2016, 16:16:15 UTC | cmake/compiler-warnings: if gcc, enable -Wshadow only for gcc-5+ | 06 October 2016, 16:16:27 UTC |
a443729 | Roman Lebedev | 06 October 2016, 13:32:07 UTC | CMake: fine-tune CHECK_COMPILER_FLAG_AND_ENABLE_IT() Acc. to #11199, in order to check some flags, we need to make sure to set other flags too. | 06 October 2016, 16:16:27 UTC |
598154a | Roman Lebedev | 06 October 2016, 13:17:35 UTC | Conf.h: dt_conf_init(): avoid g_strdup_printf(), use macro. Avoids -Wformat-nonliteral warning | 06 October 2016, 16:16:26 UTC |
aa2070a | Roman Lebedev | 05 October 2016, 19:38:56 UTC | CMake: drop modules which are already distributed by cmake-3.0.0 CMake's FindThreads.cmake looks strange. | 06 October 2016, 16:16:25 UTC |
82bd738 | Roman Lebedev | 06 October 2016, 12:37:30 UTC | Temperature iop: spot wb: support 4Bayer | 06 October 2016, 15:58:04 UTC |
06098d1 | Roman Lebedev | 06 October 2016, 12:15:09 UTC | Color picker: support 4Bayer. | 06 October 2016, 15:58:04 UTC |
ee18b95 | Roman Lebedev | 06 October 2016, 11:03:42 UTC | pixelpipe_picker_cl(): roi_copy: passthrough proper x,y Should be important for mosaiced picker. | 06 October 2016, 15:58:03 UTC |
6b744f7 | Roman Lebedev | 06 October 2016, 10:52:22 UTC | Color picker: support picking from X-Trans mosaic | 06 October 2016, 15:58:03 UTC |
c1f6fff | Roman Lebedev | 25 September 2016, 18:36:02 UTC | Color picker: support picking from Bayer mosaic | 06 October 2016, 15:58:03 UTC |
e515ac1 | Roman Lebedev | 25 September 2016, 18:10:39 UTC | Color picker: refactor computations into worker, properly check input type | 06 October 2016, 15:58:03 UTC |
0e60c00 | Roman Lebedev | 25 September 2016, 16:37:34 UTC | pixelpipe_picker_cl(): use dt_iop_buffer_dsc_to_bpp() | 06 October 2016, 15:58:03 UTC |
cc91c6b | Roman Lebedev | 25 September 2016, 16:35:50 UTC | Color picker: take dt_iop_buffer_dsc_t* | 06 October 2016, 15:58:03 UTC |
469401e | Roman Lebedev | 25 September 2016, 12:51:34 UTC | Pixelpipe: pixelpipe_picker{,_cl}: refactor common setup code into helper | 06 October 2016, 15:58:03 UTC |
6e927f9 | Roman Lebedev | 25 September 2016, 12:44:48 UTC | Refactor actual color picking into new .[CH] | 06 October 2016, 15:57:58 UTC |
cd58a33 | Roman Lebedev | 05 October 2016, 18:20:17 UTC | CMake: lower stack/frame/object limits a bit | 05 October 2016, 18:34:12 UTC |
1f23822 | Roman Lebedev | 05 October 2016, 18:31:55 UTC | Rawspeed: Cr2Decoder::decodeRawInternal(): don't stack-allocate LJpegPlain Was taking ~27Kb | 05 October 2016, 18:34:12 UTC |
779c5a3 | Roman Lebedev | 05 October 2016, 18:06:55 UTC | Rawspeed: Rw2Decoder::decodeThreaded(): don't stack-allocate PanaBitpump() Was taking ~16Kb. | 05 October 2016, 18:34:12 UTC |
9491bbe | Roman Lebedev | 05 October 2016, 17:27:14 UTC | dt_opencl_init(): don't stack [over-]allocate infointtab dt_opencl_init() uses ~40Kb stack. CL_DEVICE_MAX_WORK_ITEM_SIZES will probably return 3+ entries. Pre-allocating 1024 entries is wasteful, and does not actually protect us from overflow. Let's instead do malloc() FIXME: other places that get CL_DEVICE_MAX_WORK_ITEM_SIZES, use just size_t[3] buffer... | 05 October 2016, 18:34:11 UTC |
d699c74 | Roman Lebedev | 05 October 2016, 15:46:40 UTC | Rawspeed: ArwDecoder::decodeRawInternal(): don't stack-allocate curve. Takes almost 40Kb. | 05 October 2016, 18:34:11 UTC |
6eadbde | Roman Lebedev | 05 October 2016, 15:40:59 UTC | Rawspeed: NefDecoder::decodeRawInternal(): don't stack-allocate NikonDecompressor. Takes almost 140Kb. | 05 October 2016, 18:34:10 UTC |
6eb88ee | Pascal Obry | 05 October 2016, 16:23:49 UTC | Update French translation. | 05 October 2016, 16:23:49 UTC |
e369959 | Pascal Obry | 05 October 2016, 15:48:18 UTC | Fix memory corruption and crash (same as generate-cache). We need +1 for the last NULL parameter. I did not reproduced this one but it is the same issue as generate-cache. | 05 October 2016, 15:48:18 UTC |
5e22744 | Pascal Obry | 05 October 2016, 15:44:15 UTC | Fix memory corruption and crash. We need +1 for the last NULL parameter. | 05 October 2016, 15:47:48 UTC |
2d8570f | Roman Lebedev | 05 October 2016, 14:58:34 UTC | Fix some -Wold-style-declaration warnings. It was enabled before, just wrongly | 05 October 2016, 15:07:38 UTC |
4773633 | Roman Lebedev | 05 October 2016, 14:54:58 UTC | CMake rework: compiler warning flags cleanup | 05 October 2016, 15:07:38 UTC |
7f58d40 | Roman Lebedev | 05 October 2016, 13:13:03 UTC | dt-chart: chart: parse_cht(): don't take sizeof() of malloc()'ed buf Coverity CID 1363761 | 05 October 2016, 13:16:50 UTC |
a6e8bd1 | Roman Lebedev | 05 October 2016, 13:04:06 UTC | Demosaic iop: process_vng_cl(): don't double-free ips in case of error Coverity CID 1363764 | 05 October 2016, 13:16:50 UTC |
a312826 | Roman Lebedev | 05 October 2016, 13:01:40 UTC | Rawoverexposed iop: process_cl(): actually properly free buf in case of errors Coverity CID 1363762 | 05 October 2016, 13:16:49 UTC |
246b791 | Roman Lebedev | 05 October 2016, 12:30:13 UTC | rawspeed-identify: be less C | 05 October 2016, 12:35:51 UTC |
6cd065d | Pascal Obry | 17 January 2016, 11:17:12 UTC | Rename dt_undo_clear_list to _undo_clear_list (static function). | 05 October 2016, 12:22:12 UTC |
4f1f133 | Pascal Obry | 16 January 2016, 23:55:41 UTC | Use DT_SIGNAL_DEVELOP_MODULE_REMOVE signal to invalidate history. | 05 October 2016, 12:22:12 UTC |
820546f | Pascal Obry | 16 January 2016, 23:52:54 UTC | Free memory leak in gui_cleanup. | 05 October 2016, 12:22:12 UTC |
1d889ca | Pascal Obry | 16 January 2016, 23:50:58 UTC | New routine dt_dev_invalidate_history_module. Tag all history's module to NULL (invalidated) for a given module. This is used when a module is removed. | 05 October 2016, 12:22:12 UTC |
9548ed3 | Pascal Obry | 16 January 2016, 23:16:36 UTC | Add new signal DT_SIGNAL_DEVELOP_MODULE_REMOVE. This signal is raised when a module is remove from the interface. It will be used to invalidate such module in the undo/redo history stack. | 05 October 2016, 12:22:12 UTC |
2b8bf1c | Pascal Obry | 16 January 2016, 23:09:39 UTC | Add missing field for DT_SIGNAL_VIEWMANAGER_FILMSTRIP_ACTIVATE | 05 October 2016, 12:22:12 UTC |
4734f7f | Pascal Obry | 10 January 2016, 08:23:06 UTC | Fix map undo/redo key accels. First for consistency with the darkroom, use the Ctrl-Y for redo. Also use two separate callbacks for undo and redo as the key accels can be redefined (we do not want to check explicitely for key accels). | 05 October 2016, 12:22:12 UTC |
17f9b4e | Pascal Obry | 09 January 2016, 14:25:32 UTC | Free memory where it is allocated. | 05 October 2016, 12:22:12 UTC |
e8caec5 | Pascal Obry | 09 January 2016, 13:38:18 UTC | Fix memory leak for undo items. | 05 October 2016, 12:22:12 UTC |
cfb8bcd | Pascal Obry | 05 January 2016, 17:12:29 UTC | Use malloc/free for undo data. | 05 October 2016, 12:22:12 UTC |
cb270ba | Pascal Obry | 04 January 2016, 20:41:43 UTC | Add undo/redo support for the darkroom. | 05 October 2016, 12:22:12 UTC |
13a374c | Pascal Obry | 04 January 2016, 21:21:34 UTC | Use dt_dev_free_history_item where possible. Minor code cleanup. We now have a single point where we free the history. | 05 October 2016, 12:22:12 UTC |
da400cb | Pascal Obry | 04 January 2016, 21:11:20 UTC | New function to free history params and blend_params. | 05 October 2016, 12:22:12 UTC |
8ed386d | Pascal Obry | 04 January 2016, 20:38:25 UTC | Allow a free callback for the undo clean-up. When cleaning the undo list we may want to pass a callback to free data passed recorded into the undo/redo lists. | 05 October 2016, 12:22:12 UTC |
64a40dc | Tobias Ellinghaus | 05 October 2016, 11:29:13 UTC | Fix wrong use of g_build_filename() Found by Jan Ingwer Baer. | 05 October 2016, 11:29:13 UTC |
984c3e7 | Tobias Ellinghaus | 05 October 2016, 11:09:08 UTC | Notify user when import/export is done Tell the window manager to draw the user's attention to dt when an import of a folder or an export is finished. Those two can take a long time to be processed so the user might want to do something else outside of dt in the meantime. Instead of listening to the CPU fan slowing down as an indication he will now be told by his window manager/desktop environment. On KDE for example the entry in the taskbar is highlighted. | 05 October 2016, 11:09:08 UTC |
a29c0b9 | Tobias Ellinghaus | 05 October 2016, 11:06:35 UTC | Fix a race when updating view while importing It could happen that the collection string was changed while the update was running. Minimize that by duplicating it early. This might still have a slight chance of SQL errors but those will be "fixed" by the next update. | 05 October 2016, 11:06:35 UTC |
e1d45ab | Tobias Ellinghaus | 05 October 2016, 10:19:03 UTC | [manpage] Fix some wordings Thank you Michel Leblond for spotting these. They just got lost due to being in the wrong bug tracker. Fixes #10710 | 05 October 2016, 10:19:03 UTC |
1b789fc | houz | 05 October 2016, 09:44:46 UTC | Merge pull request #1277 from michleb/master Update French translation of the user manual | 05 October 2016, 09:44:46 UTC |
e4e448e | Michel Leblond | 05 October 2016, 09:18:20 UTC | Merge branch 'master' of https://github.com/darktable-org/darktable | 05 October 2016, 09:18:20 UTC |
f0be74f | Michel Leblond | 05 October 2016, 09:10:56 UTC | French translation of some messages in the user manual after update of its English version | 05 October 2016, 09:10:56 UTC |
e59a411 | Tobias Ellinghaus | 05 October 2016, 08:41:00 UTC | Fix db update with degenerated tags+tagged_images | 05 October 2016, 08:41:00 UTC |
09a3de2 | Roman Lebedev | 05 October 2016, 07:54:54 UTC | dt_lua_init_lock(): do lock-unlock too. Coverity is stupid. Coverity CID 1363059 | 05 October 2016, 07:55:12 UTC |
f93ef9d | Roman Lebedev | 05 October 2016, 07:52:28 UTC | Lua prefs: destroy_pref_element(): add missing break after case pref_enum: Coverity CID 1363058 | 05 October 2016, 07:52:36 UTC |
7b72601 | Ulrich Pegelow | 04 October 2016, 19:42:59 UTC | opencl: highlight reconstruction with lch mode for xtrans sensors | 04 October 2016, 19:42:59 UTC |
f572a27 | Roman Lebedev | 04 October 2016, 18:17:46 UTC | Merge remote-tracking branch 'upstream/pr/1276' Dont be so stack greedy. Reduces stack pressure. Right now this only fixes unbound stack usage. No more unbound stack-based VLA's - mostly via -Wstack-usage After previous step, i looked at the -Wvla output, and almost none of them should actually be VLA, so i enabled -Wvla too and fixed the code. This is a preparatory work to support running on Musl libc. (pr #1086, refs #10541) It has default Thread stack size of 80k. There will be next steps. * upstream/pr/1276: (72 commits) CMake: make -Wvla warning a error. dt-chart: thinplate_match(): ignore VLA for now Png format: read_header(): no need for VLA. use #define Colormapping iop: kmeans(): ignore VLA for now Colortransfer iop: kmeans(): ignore VLA for now Clahe iop: process(): no need for VLA. use #define Denoiseprofile iop: process_wavelets(): no need for VLA. use #define Metadata view lib: _metadata_view_update_values(): no need for VLA. Recentcollect lib: _lib_recentcollection_updated(): no need for VLA. Focus: dt_focus_draw_clusters(): use malloc(), not VLA Cups_print: dt_get_papers(): no need for VLA. Imageio png: read_header(): no need for VLA. use #define CMake: let's also enable -Wvla CMake: make stack/object size warning a error. dt-view: don't do unbound stack-based VLA's dt-chart: thinplate: thinplate_match(): don't do unbound stack-based VLA's dt-chart: thinplate: solve(): don't do unbound stack-based VLA's dt-chart: parse_cht(): don't do unbound stack-based VLA's dt-chart: process_data(): don't do unbound stack-based VLA's dt-g-c: don't do unbound stack-based VLA's ... | 04 October 2016, 18:17:46 UTC |
735f5a5 | Roman Lebedev | 03 October 2016, 21:02:21 UTC | CMake: make -Wvla warning a error. Now that all the issues i see are fixed, this is safe to do. | 04 October 2016, 18:06:50 UTC |
2817f11 | Roman Lebedev | 03 October 2016, 20:52:42 UTC | dt-chart: thinplate_match(): ignore VLA for now | 04 October 2016, 18:06:50 UTC |
c8d3e35 | Roman Lebedev | 03 October 2016, 20:52:11 UTC | Png format: read_header(): no need for VLA. use #define | 04 October 2016, 18:06:50 UTC |
f284078 | Roman Lebedev | 03 October 2016, 20:46:56 UTC | Colormapping iop: kmeans(): ignore VLA for now | 04 October 2016, 18:06:50 UTC |
4d7a955 | Roman Lebedev | 03 October 2016, 20:45:47 UTC | Colortransfer iop: kmeans(): ignore VLA for now | 04 October 2016, 18:06:50 UTC |
518a2dd | Roman Lebedev | 03 October 2016, 20:40:15 UTC | Clahe iop: process(): no need for VLA. use #define | 04 October 2016, 18:06:50 UTC |
8088f2c | Roman Lebedev | 03 October 2016, 20:39:44 UTC | Denoiseprofile iop: process_wavelets(): no need for VLA. use #define | 04 October 2016, 18:06:50 UTC |
0bd5206 | Roman Lebedev | 03 October 2016, 20:36:17 UTC | Metadata view lib: _metadata_view_update_values(): no need for VLA. | 04 October 2016, 18:06:50 UTC |
872311a | Roman Lebedev | 03 October 2016, 20:29:42 UTC | Recentcollect lib: _lib_recentcollection_updated(): no need for VLA. | 04 October 2016, 18:06:49 UTC |
a633f04 | Roman Lebedev | 03 October 2016, 20:26:54 UTC | Focus: dt_focus_draw_clusters(): use malloc(), not VLA | 04 October 2016, 18:06:49 UTC |
a1763d0 | Roman Lebedev | 03 October 2016, 20:33:30 UTC | Cups_print: dt_get_papers(): no need for VLA. | 04 October 2016, 18:06:49 UTC |
0839c1a | Roman Lebedev | 03 October 2016, 20:23:28 UTC | Imageio png: read_header(): no need for VLA. use #define | 04 October 2016, 18:06:49 UTC |
df58e87 | Roman Lebedev | 03 October 2016, 20:22:43 UTC | CMake: let's also enable -Wvla | 04 October 2016, 18:06:49 UTC |
7a0ed95 | Roman Lebedev | 03 October 2016, 16:16:23 UTC | CMake: make stack/object size warning a error. Now that all the issues i see are fixed, this is safe to do. | 04 October 2016, 18:06:49 UTC |
c7cdbd8 | Roman Lebedev | 03 October 2016, 14:37:40 UTC | dt-view: don't do unbound stack-based VLA's | 04 October 2016, 18:06:49 UTC |
aa83ea1 | Roman Lebedev | 03 October 2016, 14:34:50 UTC | dt-chart: thinplate: thinplate_match(): don't do unbound stack-based VLA's | 04 October 2016, 18:06:49 UTC |
a17f31b | Roman Lebedev | 03 October 2016, 14:29:09 UTC | dt-chart: thinplate: solve(): don't do unbound stack-based VLA's | 04 October 2016, 18:06:48 UTC |
c0f72e8 | Roman Lebedev | 03 October 2016, 16:24:58 UTC | dt-chart: parse_cht(): don't do unbound stack-based VLA's | 04 October 2016, 18:06:48 UTC |
bd8235f | Roman Lebedev | 04 October 2016, 09:05:12 UTC | dt-chart: process_data(): don't do unbound stack-based VLA's | 04 October 2016, 18:06:48 UTC |
3dcc78c | Roman Lebedev | 03 October 2016, 13:39:32 UTC | dt-g-c: don't do unbound stack-based VLA's | 04 October 2016, 18:06:48 UTC |
a37034d | Roman Lebedev | 03 October 2016, 13:36:11 UTC | dt-cli: don't do unbound stack-based VLA's | 04 October 2016, 18:06:48 UTC |
8d01bc1 | Roman Lebedev | 03 October 2016, 13:28:43 UTC | Png format: write_image(): don't do unbound stack-based VLA's | 04 October 2016, 18:06:48 UTC |
5d1e7b1 | Roman Lebedev | 04 October 2016, 09:24:51 UTC | Jpeg format: write_image(): don't do unbound stack-based VLA's | 04 October 2016, 18:06:48 UTC |
6b7410f | Roman Lebedev | 03 October 2016, 13:26:02 UTC | Pdf format: write_image(): don't do unbound stack-based VLA's | 04 October 2016, 18:06:47 UTC |
95eb869 | Roman Lebedev | 04 October 2016, 18:06:33 UTC | Demosaic iop: process_vng_cl(): don't do unbound stack-based VLA's Tes, this was technically not VLA, but gcc is stupid. | 04 October 2016, 18:06:36 UTC |
6b1c731 | Tobias Ellinghaus | 04 October 2016, 17:07:10 UTC | Fix merged PR 793 | 04 October 2016, 17:07:10 UTC |
05ad5d3 | Roman Lebedev | 03 October 2016, 13:04:11 UTC | Colormapping iop: process(): don't do unbound stack-based VLA's | 04 October 2016, 17:07:04 UTC |
b05a06d | Roman Lebedev | 03 October 2016, 12:54:05 UTC | Colormapping iop: kmeans(): don't do unbound stack-based VLA's | 04 October 2016, 17:07:04 UTC |
d86b185 | Roman Lebedev | 03 October 2016, 12:51:11 UTC | Colortransfer iop: process(): don't do unbound stack-based VLA's | 04 October 2016, 17:07:04 UTC |
ef01814 | Roman Lebedev | 03 October 2016, 12:44:09 UTC | Colortransfer iop: kmeans(): don't do unbound stack-based VLA's | 04 October 2016, 17:07:04 UTC |
2bb5381 | Roman Lebedev | 03 October 2016, 12:35:59 UTC | Clahe iop: process(): don't do unbound stack-based VLA's | 04 October 2016, 17:07:04 UTC |
d907060 | Roman Lebedev | 03 October 2016, 12:28:43 UTC | Dither iop: process_random(): don't do unbound stack-based VLA's | 04 October 2016, 17:07:04 UTC |
f322e09 | Roman Lebedev | 03 October 2016, 12:27:13 UTC | Vignette iop: process(): don't do unbound stack-based VLA's | 04 October 2016, 17:07:03 UTC |
2822336 | Roman Lebedev | 03 October 2016, 12:25:04 UTC | Spots iop: process(): don't do unbound stack-based VLA's | 04 October 2016, 17:07:03 UTC |
f3bb208 | Roman Lebedev | 03 October 2016, 12:20:44 UTC | Denoiseprofile iop: process_wavelets_cl(): don't do unbound stack-based VLA's | 04 October 2016, 17:07:03 UTC |
6c6dcf8 | Roman Lebedev | 03 October 2016, 12:13:42 UTC | Soften iop: process_cl(): don't do unbound stack-based VLA's | 04 October 2016, 17:07:03 UTC |
51b353d | Roman Lebedev | 03 October 2016, 12:10:17 UTC | Soften iop: process_sse2(): don't do unbound stack-based VLA's | 04 October 2016, 17:07:03 UTC |
5b65a08 | Roman Lebedev | 03 October 2016, 12:07:42 UTC | Soften iop: process(): don't do unbound stack-based VLA's | 04 October 2016, 17:07:03 UTC |