| <<<Back 1 day (to 2016/05/16) | 20160517 |
sebras | Robin_Watts: tor8: git clone --recursive ... clones and checks out submodules too. | 07:16.02 |
Robin_Watts | sebras: Yes... ? | 08:36.17 |
sebras | Robin_Watts: have you used it? I discovered it today. | 09:35.44 |
Robin_Watts | sebras: Yeah. I often forget to use it when cloning mupdf, but when I remember, it's the command I use. | 09:39.34 |
| sebras: Are you in a position to review this for me please? http://git.ghostscript.com/?p=user/robin/mupdf.git;a=commitdiff;h=3cfdde1c1431d0ba3ca018b1dd79a384668d7d75 | 09:40.19 |
| The CMYK plotter has a cut/paste problem. I'm accessing the alpha in the wrong place. | 09:40.24 |
sebras | Robin_Watts: I'll do my worst. | 09:41.15 |
| Robin_Watts: LGTM. mistake when copying from fz_paint_span_with_mask_2() I guess. | 09:43.11 |
Robin_Watts | exactly. | 09:43.24 |
| Thanks. | 09:43.27 |
tor8 | Robin_Watts: https://support.microsoft.com/en-us/kb/899149 seen this? | 10:32.11 |
sebras | tor8: intersting that that there's a difference between wb and w+b... | 10:39.30 |
| tor8: git grep '"wb"' shows a number of likely issues. but the one I would worry the most about is the one in source/fitz/output.c of course. | 10:40.44 |
| tor8: do you mind reviewing the bmp commit on sebras/master? | 10:44.13 |
| tor8: I've fixed the bug and rerun this in bmpcmp. I only see progressions now. | 10:44.31 |
sebras | ponders if the opposite is congressions. | 10:45.20 |
kens | regressions | 10:47.52 |
tor8 | sebras: you use both info->rshift and a local variable rshift; what does the local variable do and how does it differ from info->rshift? | 10:54.00 |
| sebras: possibly call the local "rtrunc"? | 10:55.06 |
sebras | tor8: that might be a better wording, yes. | 10:57.32 |
| tor8: fixeed. | 10:59.34 |
tor8 | sebras: LGTM | 11:00.48 |
sebras | tor8: do you think I'm on the right track here? http://git.ghostscript.com/?p=user/sebras/mupdf.git;a=blobdiff;f=source/fitz/output.c;h=b31920a19b17d08c88516300a7a3b1f1064a4dc3;hp=07422955d24cedb29f8747eb57945aeb1d45cf8b;hb=b5d60e0489112ac76f00d0fc2604ece9337657dd;hpb=6bff94b4aa739eabede431320329ff5167e4884e | 11:05.57 |
| tor8: basically a stripped down fz_vsnprintf() that only accepts %[0-9]d. | 11:06.26 |
| tor8: I prefer this to manipulating the format string and then passing it to fz_vsnprintf() and having that print to a buffer. | 11:06.53 |
tor8 | sebras: z = 1 ? | 11:07.00 |
sebras | even if the code is slightly duplicated. | 11:07.05 |
| the z = 1 is for counting the number of initial zeros I believe. straight from fz_vsnprintf() | 11:07.38 |
tor8 | ah, nvm. | 11:07.44 |
sebras | tor8: I guess you wrote it. ;) | 11:07.51 |
tor8 | I didn't see where you overwrote it | 11:07.53 |
sebras | tor8: ah. | 11:07.59 |
tor8 | sebras: fz_digits is overkill here. the source used 'hex' digits, which are not consecutive hence the array for simplification | 11:09.36 |
| (count % 10) + '0' is more than enough | 11:09.50 |
sebras | tor8: sure. a trivial fix. | 11:10.08 |
tor8 | sebras: though I'd try another slightly more fancy approach | 11:10.36 |
sebras | tor8: enlighten me. | 11:10.46 |
tor8 | if no %d is found, append a number just before the file suffix | 11:10.54 |
| so "foo.pdf" becomes "foo-1.pdf" etc | 11:11.11 |
sebras | tor8: because you want to be able to be lazy and leave out -o all together? | 11:12.02 |
| tor8: right now it defaults to out.png, out.cbz, out.pdf. | 11:12.20 |
tor8 | sebras: because I don't expect all our users to be smart enough to know about %d | 11:12.56 |
| and overwriting the same file for each page is not good | 11:13.03 |
| if I just wanted to be lazy, I'd make -o default to 'out%3d.png' etc | 11:13.27 |
sebras | tor8: yeah, that's what I meant. | 11:13.52 |
| tor8: is that what you'd want? | 11:13.58 |
tor8 | mutool convert takes its cue for output format from the -o argument | 11:14.21 |
sebras | tor8: for formats that can't have multiple pages inside a single file obviously. | 11:14.23 |
tor8 | and pdf and cbz are multipage formats | 11:14.27 |
| singlepage formats should number the output files, with or without a %d, IMO | 11:14.49 |
sebras | tor8: there might be more singlepage output formats though. | 11:14.50 |
| tor8: mmm, so out-%04.ext for the singlepage formats and out.ext for the multipage ones. | 11:15.21 |
| ok. | 11:15.24 |
tor8 | sebras: yes, something like that. | 11:15.37 |
sebras | tor8: now I know where I'm heading. I'll return when I have something new for you to review. | 11:16.39 |
Robin_Watts | tor8: I don't believe that's ever going to be a limitation for us? | 12:13.20 |
tor8 | Robin_Watts: file_write calls plain fwrite | 12:15.00 |
| so any fz_write that writes >64M in one go will potentially run afoul | 12:15.14 |
| and that's certainly plausible when saving png files, and possibly other places too | 12:15.35 |
Robin_Watts | tor8: A 64Meg PNG file? | 12:18.44 |
tor8 | big yes, but not entirely impossible | 12:20.16 |
| but you're right, we're probably safe to ignore the issue | 12:22.49 |
Robin_Watts | Most of the time we fwrite we do so a line at a time, so we're unlikely to hit that. | 12:44.45 |
| If we DO hit that (say for PNG), then probably the smarter fix is to change it to not need such a mahoosive buffer in the first place. | 12:45.08 |
Robin_Watts | must grab lunch. | 12:45.22 |
tor8 | Robin_Watts: okay, I'm lost. If I run the code on tor/master with mupdf-x11 and http://ghostscript.com/~tor/test.zip the test files in that zip... | 14:46.25 |
| ...the rendered logo looks *very* bad | 14:46.35 |
| but with mupdf-gl, it only looks bad the first time it's rendered. zooming in and out will make the render look good. | 14:46.52 |
Robin_Watts | tor8: I have bits of MuPDF on the floor all around me. It'll take me a bit to get it all reassembled so I can have a look :( | 14:47.55 |
tor8 | might be something with tuning->image_scale | 14:48.08 |
| Robin_Watts: you don't have multiple checkouts? ;) | 14:48.22 |
| I'll dig some more | 14:48.29 |
| d'oh! nevermind. | 14:50.05 |
Robin_Watts | Not currently. | 14:50.11 |
tor8 | okay, so SVG support in EPUB should work now on tor/master | 14:51.07 |
sebras | tor8: cool! :) | 14:51.35 |
| tor8: does that mean that we recognize plain .svg as well? | 14:51.56 |
tor8 | sebras: we've done that for a couple of weeks already ... svg works as its own document type | 14:52.25 |
| this is hooking an svg document up to a fz_image | 14:52.38 |
sebras | tor8: ah, I have missed that. | 14:52.45 |
tor8 | where when we call fz_get_pixmap_from_image it renders the svg document (actually display list as implemented, but close enough) | 14:53.05 |
sebras | tor8: you might have told me, but I have /dev/null open. :) | 14:53.05 |
| tor8: ok, so the output from make tags has annoyed me enough: sebras/master | 15:06.54 |
sebras2 | marcosw: you intentionally rebooted casper, right? | 16:23.09 |
| marcosw: otherwise we just had an unscheduled reboot (and you should probably know about those). | 16:23.30 |
Robin_Watts | sebras: a script of mine killed it. oops. | 16:24.55 |
sebras | Robin_Watts: no worries. I just want to make sure that whomevere is responsible for casper _knows_ about it rebooting, so that we do get the uptime from the ISP that we pay for. | 16:27.50 |
StephenLynx | hey, anyway to make ghostscript automatically exit? when I run gs -q -sDEVICE=txtwrite -dNOPAUSE -sOutputFile=%stdout% it won't exit until I manually stop it (ctrl+c) | 20:09.00 |
| nvm, using -o instead of sOutputFile did what I wanted | 20:11.38 |
| Forward 1 day (to 2016/05/18)>>> | |