| <<<Back 1 day (to 2018/10/02) | 20181003 |
Tamir_Evan | tor8: (For the logs) I built the current tor/master (with the gl fix commit), with VS Express 2013 (on Windows 7), and when I run the resulting mupdf-gl, it saves to .mupdf.history when I close the window, like it does when I press 'q'. | 02:23.50 |
sebras | tor8 (for the logs): there's a number of commits on sebras/master | 06:54.39 |
| btw, they cluster without issues. | 07:01.43 |
Robin_Watts | tor8: I'm just looking through some commits from sebras/wip, and figured we should be commenting here so you can hear my opinions and comment too. | 11:26.01 |
| "Add separations to pixmap rendering functions". I'm in favour of it, but it's API churn. | 11:26.41 |
| Should we instead use fz_new_pixmap_with_seps_from_display_list etc? | 11:27.07 |
| I think on balance not, but... | 11:27.13 |
tor8 | I would just go with the churn in this case... | 11:27.40 |
| Robin_Watts: one commit on tor/master that I'd like to get in before making 1.14.0 release | 11:28.44 |
Robin_Watts | tor8: ok, I guess. I have no strong opinions on that, but it seems plausible. | 11:30.55 |
tor8 | thanks. if I push that, can you rebuild the windows tarball and we'll tell lisa and henry the release is done? | 11:32.21 |
Robin_Watts | Sure. | 11:32.40 |
tor8 | then we can see about the 1001 commits on sebras/master | 11:32.42 |
sebras | tor8: I can squash them, but I found them iteratively. :) | 11:34.44 |
| tor8: actually I thought the release had already been done. :) | 11:34.58 |
Robin_Watts | sebras: "Fix bug where CMYKA+spots could not be converted to RGBA+spots" | 11:35.26 |
| In the original code, sd = sdata + sc | 11:35.48 |
| sc = number of composite colors = sn - ss - sa | 11:36.35 |
| where sn = number of components in total, ss = number of spots, sa = number of alpha channels. | 11:36.55 |
| (composite colors is a bad name. I mean non-spot colors) | 11:37.33 |
sebras | I wrote the commit in february, let me refresh my mind a minute. | 11:37.44 |
Robin_Watts | So for each pixel we store: <non-spot 0> <non-spot 1> ... <non-spot sc-1> <spot 0> ... <spot ss-1> <alpha> | 11:38.18 |
| So sd points to <spot 0> | 11:38.38 |
| and sd[sc] = alpha. | 11:39.10 |
sebras | that doesn't make sense. | 11:39.27 |
Robin_Watts | No, sorry. | 11:39.32 |
sebras | sdata + sc points to <spot 0> | 11:39.34 |
Robin_Watts | sd[ss] = alpha | 11:39.39 |
sebras | then sdata + sc * 2 cannot point to <alpha> | 11:39.46 |
Robin_Watts | So, the code as written is wrong. | 11:39.56 |
| but the code in your commit, which tries to use sd[-1] for alpha seems wrong too. | 11:40.10 |
sebras | right. and I made it wrong in a different way. :) | 11:40.22 |
| I assumed that sc was sn + ss | 11:40.50 |
| I think. | 11:40.55 |
Robin_Watts | It looks to me like you were thinking: <non-spots> <alpha> <spots> | 11:41.28 |
sebras | perhaps..? | 11:41.45 |
Robin_Watts | cos that would make sd[-1] correct, given your adjustment to sd. | 11:42.17 |
sebras | Robin_Watts: I'm not sure what file I used to reproduce this. | 11:46.45 |
| Robin_Watts: oh, and dinner. | 11:47.01 |
Robin_Watts | :) | 11:47.07 |
sebras | I'll be back. | 11:47.17 |
Robin_Watts | Hasta La Vista, Baby. | 11:47.35 |
sebras | Robin_Watts: the first quote here was not suitable: https://www.rottentomatoes.com/m/terminator/quotes/ | 11:53.38 |
tor8 | Robin_Watts: sebras: https://www.youtube.com/watch?v=WegeYTgV3B0 | 11:56.42 |
Robin_Watts | Those movies haven't backed up to google drive yet :) | 11:57.22 |
| "push_group_for_separations: Use convenience function 3/3." | 12:02.39 |
| You have moved clone = ... down from the top, but the first if still checks if (clone == ...) | 12:03.04 |
| "WIP: question" - good question. | 12:07.43 |
| This code probably predates colorspaces being ref counted properly. | 12:07.58 |
| If we do want to fz_keep them, then we need to correctly fz_keep/fz_drop them elsewhere in this code. It's a larger change. | 12:08.27 |
moolc | /query sebras | 12:08.42 |
| right | 12:08.45 |
Robin_Watts | moolc: Yes. | 12:09.39 |
| sebras: "Drop destination pixmap even upon error when closing draw device." - ok, by why the fz_warn? Is that debug code? | 12:10.11 |
| sebras: "drop-pixmap-upon-error" - Urm? what? | 12:11.42 |
moolc | Robin_Watts: this irc client is really unforgiving.. then again i don't have to leave the editor to use it.. so... | 12:11.45 |
Robin_Watts | In order for the fz_try to matter, the fz_copy_pixmap_area_converting_seps call would have to throw, but if it does then copy is never set. | 12:12.34 |
| sebras: "take-alpha-into-account" - again, looks like you're expecting <non-spots> <alpha> <spots> | 12:13.46 |
| sebras: "Split pixmap into components and save as PNGs." - for everything else we have fz_save_ and fz_write. And maybe an fz_bandwriter. Should we do the same here? | 12:16.06 |
| That'll teach you to ask me for reviews :) | 12:16.39 |
sebras | Robin_Watts: not at all! I want to have these things reviewed. they've been waiting since february! | 13:55.50 |
moolc | sebras, tor8: https://www.youtube.com/watch?v=-637MX1MCy0&t=258 | 14:04.44 |
| tor8: btw. the variant in the video is right up your valley (numpad location wise) | 14:09.44 |
| Forward 1 day (to 2018/10/04)>>> | |