| <<<Back 1 day (to 2020/03/10) | Fwd 1 day (to 2020/03/12) >>> | 20200311 |
Guest45096 | Where Can I see the print statements outputs in ghostscript? | 08:57.43 |
kens | Guest45096: I have no idea what you mean, what 'print statements' ? | 08:58.06 |
Guest45096 | "printf" statements added in the ghostscript code base | 08:58.58 |
kens | You can't (or at least shouldn't) use printf in the Ghostscript C code. | 08:59.21 |
| If you try that you shoudl get a compiler warning telling you to use gs_sprintf instead | 08:59.42 |
| Actually I'm thinking of sprintf there | 09:00.22 |
| But nevertheless, you shouldn't be using printf | 09:00.30 |
Guest45096 | So if we use gs_sprintf, where can see the printed statements | 09:00.31 |
kens | Use emprintf or dmprintf | 09:00.48 |
| dmprintf won't do anything except o debug builds, emprintf sends outptu to the PostScript stderr | 09:01.10 |
| Why do you want to do this anyway ? | 09:01.26 |
Guest45096 | we have to get the code flow | 09:02.52 |
kens | Why ? | 09:03.03 |
Guest45096 | to have a better understanding of PS | 09:03.39 |
kens | Fiddling with the Ghostscript source code won't teach you very much about PostScript | 09:03.57 |
| If you insist on watching the code execute, tehn use a debugger | 09:04.25 |
Guest45096 | which is the debugger we can use for that? | 09:05.43 |
kens | Well that depends doesn't it ? You haven't told me which operating system you are using, and clearly that's going to matter. You can't use CodeWarrior on Windows for example. Perhaps if you were to explain what you are really up to we could be more helpful. | 09:06.47 |
chrisl | Reading the PLRM is likely to reap better results in understanding PS | 09:06.56 |
kens | My sentiments exactly | 09:07.07 |
Guest45096 | UNIX is the debugger I am using | 09:07.43 |
kens | UNIX is not a debugger | 09:07.52 |
Guest45096 | *OS | 09:07.58 |
kens | Guest45096: I think you need to explain further what you are doing, and on whose behalf. | 09:08.23 |
| Why do you want to mess with the Ghostscript code ? | 09:08.41 |
Guest45096 | thanks a lot for ur effort. | 09:08.56 |
zdohnal | Hi everyone, I have a question about gp_enumerate_fonts_next function | 13:00.27 |
chrisl | Well ask away! | 13:02.18 |
zdohnal | the way how it is implemented right now (together with z_fontenum), the loop in z_fontenum() will stop with the first font, for which fails one of FcPatternGet* functions | 13:03.52 |
| chrisl: this way no fonts after 'failing' font is loaded | 13:04.18 |
| chrisl: is it intended? | 13:04.29 |
| chrisl: I'm solving RHEL/Fedora bug about loading fonts from fontconfig and I backported fix from git (bug 701969), but gs still does not load all installed fonts from /usr/share/fonts | 13:06.16 |
| so I was debugging the issue and I found out that /usr/share/fonts/cantarell/Cantarell-VF.otf from abattis-cantarell-fonts package fails the FcPatternGetInteger function for FC_WEIGHT | 13:07.31 |
chrisl | Hmm, that does not look right, no | 13:08.19 |
zdohnal | I'm not sure whether if the reason of failure is in the font or in fontconfig, but gs stops iterating after that font | 13:08.33 |
| so fontconfig loads 143 fonts, but gs loads only 60, because the looping stops on 61st font - that cantarell one... | 13:09.36 |
chrisl | From the comment, I would guess this was not our code to begin with | 13:09.57 |
zdohnal | chrisl: yep, but I'm not sure how much of the code is from RHEL7 patch | 13:10.43 |
chrisl | zdohnal: I would guess the entirety of the original fontconfig integration was lifted from it, but it was before my time. | 13:11.43 |
| At a guess, we'd want the FcPattern* calls in a loop, and only return when we find a usable name, or the list is exhausted | 13:12.59 |
zdohnal | chrisl: ok, I'll check it | 13:17.14 |
chrisl | zdohnal: Something like: http://git.ghostscript.com/?p=user/chrisl/ghostpdl.git;a=commitdiff;h=55be2b0b9d94565de8ec177455e114b9ec43c2ef | 13:20.10 |
zdohnal | chrisl: ok, thank you! | 13:22.54 |
| I'll check if it works | 13:23.02 |
chrisl | zdohnal: I was going to say: my testing was very brief! If it looks like it solves your issue, then I'll test it some more, and commit it (or something like it) to our repo | 13:24.05 |
zdohnal | chrisl: the patch seems to work, I'll run it in our CI tests and if they pass, it will be okay to add the patch from our side | 14:27.00 |
chrisl | zdohnal: Cool, thanks | 14:28.18 |
zdohnal | chrisl: our CI tests pass on Fedora, so imo it is good to go from us | 15:02.07 |
chrisl | zdohnal: Thanks. I'll someone to review it, and if they are happy, I'll push it | 15:02.40 |
| zdohnal: Slight tweaj to what's getting pushed (functionally identical): http://git.ghostscript.com/?p=user/chrisl/ghostpdl.git;a=commitdiff;h=09a575f0fdc6edde7e42f325167318a1b199839f | 15:19.21 |
| s/tweaj/tweak | 15:19.44 |
zdohnal | chrisl: lgtm | 15:20.12 |
chrisl | zdohnal: I'm just running it on our regression test cluster, then I'll push it | 15:20.56 |
zdohnal | chrisl: ok | 15:21.08 |
chrisl | Although, the regression cluster doesn't actually test the fontconfig code..... | 15:21.38 |
zdohnal | chrisl: I was thinking about how to improve the debug printf message - like I would like to know which font failed, but there is no guarantee about FcPatternGetString which brings file_fc will not fail... | 15:26.12 |
| chrisl: so probably the ones which see such message will need to run debugger | 15:27.13 |
chrisl | zdohnal: I would assume that every font *must* have a FC_FILE value? | 15:28.11 |
zdohnal | chrisl: or just add it to debug messages where we know file_fc exists | 15:28.24 |
| chrisl: I hope so :) | 15:28.51 |
chrisl | zdohnal: I was thinking if we change the order so the FC_FILE came first, then we could include that string in the other debug outputs | 15:29.13 |
zdohnal | chrisl: sounds good | 15:31.38 |
chrisl | zdohnal: I'll mess with that, if you like. It shouldn't take too long. Do you know which package Cantarell-VF.otf is in? | 15:32.56 |
zdohnal | chrisl: abattis-cantarell-fonts | 15:35.02 |
| chrisl: looks like gnome fonts, git url is gitlab of gnome | 15:36.19 |
chrisl | Hmm, I have that installed, but I haven't noticed a failure | 15:36.54 |
zdohnal | chrisl: that would be nice if you change the order and add file_fc to debug msg, thank you! | 15:38.49 |
| chrisl: maybe it is a fontconfig issue... | 15:39.18 |
chrisl | zdohnal: Or different versions of the font. I just makes it a little hard for me to see the result of my fiddling! | 15:39.58 |
zdohnal | chrisl: it was in fedora rawhide, I'll check it in a minute after the image starts up | 15:42.53 |
chrisl | zdohnal: http://git.ghostscript.com/?p=user/chrisl/ghostpdl.git;a=commitdiff;h=b2d084d01a22261281fc35e990efc46dde6ea8f9 | 15:59.15 |
zdohnal | chrisl: abattis-cantarell-fonts-0.201-2 | 15:59.40 |
| chrisl: looks good, thank you! | 16:00.35 |
chrisl | zdohnal: I'm on Ubuntu, so the packaging is different <sigh> | 16:04.22 |
zdohnal | chrisl: iirc ubuntu has similar functionality as Fedora has in dnf (dnf repoquery --whatprovides /path/to/file) | 16:05.47 |
chrisl | zdohnal: apt reports: fonts-cantarell/bionic,bionic,now 0.0.25-4 | 16:08.22 |
| zdohnal: That's those two commits pushed to our main repo | 16:10.59 |
zdohnal | chrisl: I remembered - I can send you the file which triggers the issue for me | 16:17.36 |
chrisl | zdohnal: Unless there's a specific issue you want to look at, it's probably not necessary. If fontconfig can't get information from the file, it's unlikely we can use it, either. | 16:18.50 |
zdohnal | probably you can see there are message about missing fonts if you use unpatched gs | 16:21.06 |
| chrisl: and those missing fonts are installed properly in /usr/share/fonts | 16:21.53 |
chrisl | zdohnal: It's up to you, I trust your testing of these changes, and if you're happy with them, then I'm happy. | 16:22.16 |
zdohnal | chrisl: just for the case if you want it to check by yourself or add it into testing suite | 16:23.05 |
chrisl | zdohnal: The same package here: https://rpmfind.net/linux/RPM/fedora/devel/rawhide/x86_64/a/abattis-cantarell-fonts-0.201-2.fc32.noarch.html ? | 16:26.12 |
zdohnal | chrisl: seems like it :) | 16:26.43 |
chrisl | zdohnal: Okay, it looks like it's working how we want, and expect now, which is all to the good :-) | 16:32.51 |
zdohnal | chrisl: thank you! | 16:47.52 |
chrisl | zdohnal: NP, you made a good catch, thanks for bringing it up | 16:48.38 |
| <<<Back 1 day (to 2020/03/10) | Forward 1 day (to 2020/03/12)>>> | |