| <<<Back 1 day (to 2013/08/27) | 2013/08/28 |
Robin_Watts | tkamppeter: You'd need to speak to chrisl to be sure, but it wouldn't surprise me to find that that option had been removed. | 09:01.44 |
| libopenjpeg as shipped by the authors is very broken in some regards. We have been employing people to fix it. | 09:02.20 |
| hence by us using a stock libopenjpeg it creates all sorts of security issues. | 09:02.41 |
| If packagers want to put that back, then so be it, but it won't be something we support. | 09:03.00 |
| tkamppeter: Having said all that, I can see various SHARE_JPX lines in gs/Makefile.in, so maybe it still can be done. | 09:05.00 |
| Ah, it looks like the configure.ac script will never set SHARE_JPX. | 09:06.37 |
| so you'd need to extend that script if you wanted to put it back. but please don't. | 09:07.15 |
| Morning tor7. | 09:14.11 |
tor7 | hi Robin_Watts | 09:14.28 |
Robin_Watts | I've got some more RLE stuff on robin/master. | 09:14.28 |
| All in all there are 4 RLE commits. | 09:14.37 |
| I could squash them all to 1. | 09:14.51 |
| but it's up to you as to what you'd prefer to review. | 09:15.03 |
| If you haven't looked at any of them in detail, it's probably easiest if I squash them, as that way you won't have to look over me changing the same code several times. | 09:15.36 |
paulgardiner | Robin_Watts, tor7: fix for the annotation-save failure up on paul/master | 09:24.45 |
| ... one liner | 09:24.58 |
Robin_Watts | paulgardiner: Where? | 09:27.24 |
| paulgardiner: I looked last night and couldn't see it. | 09:28.10 |
paulgardiner | Bah! Might help of I push it | 09:29.45 |
| There | 09:30.07 |
Robin_Watts | "foul", not "fowl" :) | 09:30.24 |
| "all others" -> "all other" | 09:31.03 |
| but otherwise that looks good to me. | 09:31.12 |
paulgardiner | No. I actually mean a descending bird. :-) | 09:31.15 |
| Great thanks | 09:31.33 |
Robin_Watts | runs | 09:38.40 |
stanv | Hello⦠I'm looking for PS visual editor | 10:38.44 |
pipitas | I have a quick question about the Ghostscript logo as can be found here: http://commons.wikimedia.org/wiki/File:Ghostscript.svg . Is the licensing info contained there correct? | 10:53.24 |
| If so, this means that upcoming my eBook (currently work in progress at http://leanpub.com/pdfkungfoo , currently with the working title of "PDF-KungFoo with Ghostscript & Co.") will be easily able to use this logo on the title page? | 10:53.32 |
| Or should I better officially check with Artifex and Miles? | 10:53.40 |
atulagrwl | paulgardiner: I see a checkin in mupdf repository about parent object. Is this the fix for the bug? | 10:56.24 |
Robin_Watts | pipitas: I would imagine it's not a problem, but I would check with Miles and co in case. | 11:05.32 |
stanv | Robin_Watts: I can't imagine the way to create EPSâ¦. one that could hold variablesâ¦.. | 11:07.42 |
| https://pastee.org/j7emz | 11:07.43 |
| do you see {job-originating-user-name} ? | 11:07.43 |
| I want to create something like that⦠but I don't know PS language. | 11:07.43 |
Robin_Watts | stanv: You said yesterday that you wanted to create a fixed banner. | 11:08.35 |
| now you're saying you want to produce a banner that uses variables. | 11:08.49 |
stanv | InkScape/OpenDraw create EPS that I could see my text t | 11:08.57 |
Robin_Watts | For the former EPS is fine. | 11:09.02 |
| For the latter, you'll probably want to hand craft it using PS. | 11:09.14 |
stanv | Robin_Watts: how to do this ? | 11:09.28 |
Robin_Watts | or at least, you will need to speak PS in order to tweak stuff. | 11:09.36 |
| You read the Postscript Language Reference Manual. | 11:09.51 |
stanv | Robin_Watts: I want to draw bright/creative page in InkScapeâ¦. but after that add variables to some locations. | 11:11.06 |
| Is it possible ? | 11:11.12 |
Robin_Watts | stanv: Sure. But you will need to be able to speak postscript. | 11:11.23 |
| and you may find it interesting with fonts. | 11:11.42 |
stanv | EPS produced by InkScape/OpenDraw is mess for me. | 11:12.22 |
Robin_Watts | I am not going to do it for you, nor I doubt will anyone here. You WILL need to resign yourself to learning some postscript, and that will mean reading at least part of the PLRM. There is no shortcut, unless you want to hire someone. | 11:12.33 |
| stanv: quite possibly, yes. | 11:12.40 |
stanv | Robin_Watts: But, do I right understand, that is correct way to do this ? I mean draw desired page in vector editor->export EPS->add to it variables ? | 11:14.27 |
Robin_Watts | stanv: That's one way to proceed. | 11:14.52 |
| but probably you want to do the graphical elements that way, and do the textual elements all in PS. | 11:15.12 |
stanv | Robin_Watts: Are there any other ways to do this ? | 11:15.34 |
Robin_Watts | As there is no telling how the drawing package will handle the textual stuff - it might convert the text to paths, it might embed a subset font etc. | 11:15.50 |
| stanv: There are many ways to skin a cat. | 11:16.00 |
stanv | Robin_Watts: I'm looking the easiest and most simple way | 11:16.42 |
Robin_Watts | pipitas1: see my answer to pipitas above. | 11:17.03 |
| stanv: I've answered your questions as best I can. | 11:17.40 |
stanv | Robin_Watts: maybe it would be easy to write LaTex document, convert it to PS ? or not ? | 11:18.23 |
Robin_Watts | possibly, but are you sure that process won't subset the fonts,and will leave you with readable, editable PS? | 11:18.50 |
| You'd need to suck it and see. | 11:19.06 |
stanv | Robin_Watts: https://dl.dropboxusercontent.com/u/9413602/t.eps could you point where is my word ? | 11:29.09 |
| Where is first letter ? | 11:29.32 |
| "P" | 11:29.51 |
Robin_Watts | oh, god it's cairo output. give it up now. | 11:31.04 |
| [<0101>-1<020303030405>-1<06>]TJ | 11:31.34 |
| That's the line that prints the text. | 11:31.41 |
| but to understand it, you need to speak PS well enough to understand the definitions that came before it. | 11:32.04 |
| so just don't. | 11:32.09 |
stanv | sad (( | 11:33.38 |
| do you know, engine that doesn't use cario ? I have MacOC/Linux machines | 11:34.22 |
Robin_Watts | not offhand. I'd use Xara Xtreme on my windows box personally. | 11:36.47 |
zeniko | Robin_Watts: tor7: Got some patches for you to consider over on zeniko/mupdf | 11:42.30 |
Robin_Watts | zeniko: cool. let me have a look. | 11:42.45 |
zeniko | also another round of libopenjpeg fixes on zeniko/openjpeg which you might want to consider merging at some point | 11:43.18 |
Robin_Watts | xps_fixes ? pdf_fixes ? | 11:43.22 |
zeniko | from master to pdf_fixes are fixes related to PDF | 11:43.44 |
| xps_fixes are additional fixes which are only required for XPS documents | 11:43.56 |
Robin_Watts | So these are all 3 days + old ? | 11:43.59 |
| (just checking that I'm looking at the right things) | 11:44.13 |
zeniko | Yes, I've merged master last 3 days ago | 11:44.36 |
Robin_Watts | ok. | 11:45.07 |
| typo one looks good. | 11:45.25 |
| fax one looks plausible. | 11:46.31 |
| "Content streams may contain multiple inline images compressed as LZW data." | 11:47.13 |
| really? | 11:47.15 |
| I mean, the patch looks fine, I'm just struggling to think of an example of this. | 11:47.53 |
tor7 | Robin_Watts: yes. it looks sane. | 11:47.55 |
| Robin_Watts: there may be any number of inline images in a content stream | 11:48.11 |
kens | I htink pdfwrite can create ppage streams with multiple inline images compressed using LZW | 11:48.18 |
tor7 | BI raw-data EI | 11:48.21 |
zeniko | Robin_Watts: that's for http://code.google.com/p/sumatrapdf/issues/detail?id=1980 | 11:48.22 |
Robin_Watts | oh, inline images, of course, yes. | 11:48.53 |
tor7 | zeniko: wouldn't fz_dctd_set_common_tables be better as an argument to the dctd constructor function? | 11:51.18 |
Robin_Watts | All the PDF ones look good to me. | 11:52.39 |
zeniko | tor7: that would work as well | 11:53.09 |
| would you rather prefer a fz_open_dctd_with_tables or having everybody else pass NULL, 0 arguments at the end? | 11:53.31 |
tor7 | zeniko: no real preference there, the number of places where we call fz_open_dctd is small enough | 11:54.24 |
| I think passing NULL, 0 will be good enough | 11:54.57 |
Robin_Watts | All the XPS ones (with the modification you and tor7 have just discussed) look good to me too. | 11:56.16 |
tor7 | Robin_Watts: zeniko: regarding invert_cmyk_jpeg, don't we do something similar for PDF in some cases based on the ColorTransform parameter? | 11:56.21 |
zeniko | tor7: not AFAICT | 11:57.19 |
Robin_Watts | We read ColorTransform and set color_transform | 11:57.43 |
| color_transform is then used to control what transform we do based on the number of components. | 11:58.00 |
| but it's RGB <=> YCbCr and YCCK <=> CMYK | 11:58.23 |
| so no 'inversion'. | 11:58.27 |
tor7 | Robin_Watts: oh, right! | 11:58.32 |
| looks good then | 11:58.51 |
| Robin_Watts: isbigendian fix on robin/master LGTM | 12:00.06 |
| Robin_Watts: dump glyph cache size too | 12:00.29 |
zeniko | Robin_Watts: tor7: I've moved the JPEG tables patch to the end of xps_fixes and rebased on current master, so that you could merge all the fixes that looked good to you | 12:05.52 |
sebras | zeniko: did you see my message about your patches the other day? | 12:06.18 |
Robin_Watts | zeniko: Will do, ta. | 12:06.25 |
zeniko | sebras: I didn't | 12:06.37 |
sebras | zeniko: patch c875b3a should probably not refer to fz_atof() as pdf_load_version() has removed this comment. | 12:06.53 |
tor7 | Robin_Watts: the "last run'' just means one less run per line? | 12:06.56 |
sebras | zeniko: it's just a comment, I know, but it seems unnecessary to carry it around when we skipped it elsewhere. ;) | 12:07.26 |
Robin_Watts | tor7: Yes, it saves 1 byte per line. | 12:08.17 |
zeniko | sebras: I still hope that we can at some point get away from the bad assumption(s) the code currently makes | 12:08.20 |
| (i.e. that the major number has to be single-digit) | 12:08.34 |
tor7 | Robin_Watts: awful lot of code for it :) | 12:08.35 |
sebras | zeniko: it that a bad assumption according to spec? I didn't expect that. | 12:09.05 |
Robin_Watts | tor7: Typical anatomy for a line is: offs, edges, ons, edges | 12:09.08 |
tor7 | Robin_Watts: also, could you please put the lonely ; on a line of its own in the "while (foo++);" constructs? | 12:09.15 |
zeniko | sebras: at some point there might be PDF version 10.0 - no spec going to prevent that ;-) | 12:09.43 |
Robin_Watts | tor7: I'd rather use { } in that case. | 12:09.46 |
tor7 | Robin_Watts: yeah, I totally get why you want it. just making sure we were on the same page. | 12:09.46 |
| in that case, leave them as is. | 12:10.02 |
Robin_Watts | tor7: If I squash all the commits together, the eol bit code won't seem so bad :) | 12:10.27 |
tor7 | maybe just put a blank between the ) and the ; for extra clarity | 12:10.30 |
zeniko | anybody with commit access to ghostpdl.git: I could still use a hand pushing the patches for bug 694111 and bug 694121 (which both apply to MuPDF through jbig2dec as well) | 12:10.33 |
tor7 | Robin_Watts: alright, squish them together and then I'll take a second look through them | 12:10.51 |
Robin_Watts | zeniko: Who was supposed to be reviewing them for you? | 12:11.05 |
| Is jbig2dec henry these days? | 12:11.19 |
tor7 | the stuff seems okay so far, but there are a few API issues I may want to think about a bit more | 12:11.20 |
zeniko | Robin_Watts: Henry has reviewed them already | 12:11.27 |
Robin_Watts | tor7: sure. | 12:11.30 |
| zeniko: In that case he should have pushed them. | 12:11.52 |
sebras | zeniko: true enough, but it seems less likely since ISO took over. | 12:11.57 |
Robin_Watts | I think he's back later today - I'll prod him then and between us we'll get them pushed. | 12:12.10 |
tor7 | zeniko: do you use the version number for anything useful other than presentation? | 12:12.26 |
| zeniko: I'd be perfectly happy to leave it as a string in that case | 12:12.47 |
zeniko | sebras: We still have people using SumatraPDF 1.2 (even though 2.3.2 is current), so once we reach PDF 10.0, people might still use the code... | 12:12.57 |
sebras | tor7: there is a field in the root that overrides the version in the header. | 12:13.41 |
zeniko | tor7: mostly for presentation, except that we check for version 1.7 and later crypt versions to determine ExtensionLevels 3 and 8 | 12:13.55 |
sebras | zeniko: oh.. and they are still asking for bugfixes on 1.2?! wow... | 12:14.01 |
tor7 | zeniko: we used floats, until we had the "problem" with version 1.3 (or 1.299999999 as it were...) | 12:14.37 |
zeniko | sebras: They're not asking for bug fixes, but if we can prevent adding "bugs" which we already know will have to be fixed later... | 12:14.40 |
sebras | zeniko: but yeah, I see your point. I'm not sure that atof() is the correct way to do it though. | 12:14.42 |
| tor7: right, now I remember.... | 12:15.06 |
tor7 | strchr(s, '.') would be better than the current assumption either way | 12:15.18 |
zeniko | tor7: IIRC, I just added 0.01 since only a single digit after the comma was used anyway | 12:15.34 |
tor7 | zeniko: until version 1.10 of the spec :) | 12:15.54 |
paulgardiner | Robin_Watts: another little commit on paul/master: adding the logo you gave me to signature appearances | 12:15.57 |
zeniko | but we could just as well work with having access to the string instead | 12:16.00 |
sebras | tor7: that seems like correct approach yes. | 12:16.08 |
| zeniko: wouldn't that make it harder to do >= comparisons? | 12:16.28 |
tor7 | if you use the 1.7 version number to determine crypt extension levels, keeping it as a number makes some sense | 12:16.29 |
zeniko | sebras: the current format makes comparisions even trickier, since 1.10 > 1.9 > 1.1 | 12:17.20 |
| (and of course 2.0 > 1.10) | 12:17.35 |
tor7 | doc->version_major, version_minor, version_patch | 12:17.45 |
| or extlvl or whatever | 12:18.08 |
zeniko | Extension levels are vendor specific, IIRC (even though maybe only Adobe actually uses them) | 12:18.38 |
| version_major and version_minor would help however | 12:19.06 |
Robin_Watts | tor7, zeniko: so I can push all zenikos fixes except for the jpegtables/TIFF one? | 12:21.42 |
tor7 | Robin_Watts: go ahead. | 12:21.52 |
Robin_Watts | ok. | 12:21.55 |
tor7 | we'll fix the version parsing later | 12:22.05 |
Robin_Watts | pushed. Now I'll look at paulgardiners :) | 12:22.39 |
zeniko | Robin_Watts: thanks | 12:23.00 |
Robin_Watts | zeniko: np. | 12:23.13 |
| paulgardiner: pushed. | 12:24.14 |
paulgardiner | Great. Thanks Robin_Watts | 12:24.23 |
Robin_Watts | sebras: Any from you? :) | 12:24.42 |
sebras | Robin_Watts: nope, not since last time. :) | 12:25.31 |
sebras | studies too much chinese. | 12:25.47 |
pipitas1 | Robin_Watts: Thx for responding â sorry, I was away from keyboard for a while. I'll check with Miles then... | 12:33.24 |
Robin_Watts | pipitas1: I can't believe it'll be a problem, but better to have Miles rubber stamp it than for it to be a problem later. | 12:34.07 |
| oh, crap. | 12:35.35 |
| I've broken something while rebasing, so I'll need to do a force push. | 12:35.54 |
| hi henrys | 13:45.03 |
| good break? | 13:45.07 |
henrys | yes wonderful I enjoyed penticton very much | 13:45.32 |
Robin_Watts | was that the triathlon you posted pictures of? | 13:45.53 |
| henrys: have you reviewed zenikos patches for bug 694111 and bug 694121 ? Someone needs to commit them for him. | 13:46.35 |
henrys | Robin_Watts: a friend posted the pictures. | 13:48.13 |
| Robin_Watts: I'll try to look at zeniko patches today. | 13:48.41 |
Robin_Watts | s/you posted pictures of/pictures of you were posted at/ | 13:49.04 |
henrys | Robin_Watts: the race was great everything went more or less as planned some difficult course condition knocked down my time a bit, but no complaints. | 13:51.37 |
Robin_Watts | henrys: nice. I am massively fat and slow at the moment. I've been getting slower all year, and 3 weeks of not running has made it even worse. | 13:52.35 |
henrys | Robin_Watts: sometimes it helps mentally to do some intervals or hill workouts to mix it a bit or even better put a race on the calendar. | 13:54.47 |
Robin_Watts | Last year I had a challenge on the Nike website to do a half marathon a month. | 13:55.16 |
| and that focussed my attention. | 13:55.27 |
| But they broke the Nike website last year and haven't fixed it yet. | 13:55.45 |
henrys | I think the best thing is to have a race and every workout on the calendar, than it's mindless and you just do it. | 13:57.21 |
Robin_Watts | I don't generally have a problem getting out and doing my runs. I run 5 days a week, 7 miles a day (unless it's raining, or I'm ill, or I have another appointment etc) | 13:58.43 |
henrys | argh things have really piled up! | 13:58.56 |
paulgardiner | You make me feel so guilty. I went on my first run since March today. Just 5K, Felt like hell. | 14:01.24 |
Robin_Watts | paulgardiner: Yeah, but you dance 2 evenings a week, right? | 14:03.30 |
paulgardiner | Just one lately. :-( | 14:03.49 |
atulagrwl | WoW you guys are great. I am planning to run half marathon for past one year :) | 14:10.43 |
| paulgardiner: Did you submitted the fix of the problem (annotation issue on specific pdf)? | 14:12.08 |
paulgardiner | atulagrwl: Yes. That's in now | 14:12.29 |
atulagrwl | paulgardiner: Thanks again :) | 14:12.37 |
paulgardiner | No problem | 14:12.46 |
henrys | paulgardiner: how's fasting? sabrina is on a fast day, good thing I'm a good runner ;-) | 14:13.28 |
paulgardiner | henrys: It's okay. Certainly better than being on a diet every day. I don't think I'm losing weight any more. Probably too many treats on non-fast days. | 14:15.53 |
| I think it saps my energy a bit for workouts the day after a fast. | 14:16.43 |
| How's Sabrina getting on with it? Any problems sleeping? | 14:17.53 |
henrys | paulgardiner: she's really adjusted to it. | 14:18.39 |
| but I do tread lightly fast days because she can be a bit more irritable I've noticed. | 14:19.44 |
paulgardiner | That's brilliant. So no tempation to give it up, I'd guess. | 14:19.45 |
henrys | paulgardiner: no not at all. | 14:20.14 |
| paulgardiner: like you she is not losing any more weight. She got close to her goal and now uses fasting for maintenance. | 14:21.13 |
paulgardiner | :-) Yeah I think I'm maybe a bit more irritable on fast days. I usually choose days on which I don't see Linda, so no second oppinion on that. | 14:21.14 |
Robin_Watts | paulgardiner: You won't be fasting on friday, right? | 14:22.48 |
| Handing a shotgun to someone that's both irritable and hungry is possibly not a good plan :) | 14:23.28 |
paulgardiner | Ha! I know what you're thinking. Actually, I'm pretty sure I've been fasting the last few times we've been out shooting | 14:23.36 |
henrys | paulgardiner: I'm sort of looking forward to eating less, I had to eat constantly up until the race. | 14:28.38 |
paulgardiner | Now that's just weird! :-) | 14:30.50 |
Robin_Watts | tor7: ping | 14:31.24 |
| So... am I right in thinking that we currently quantise glyphs to subpixel positions only in so far as we convert from float to an unsigned char for the subpixel position? | 14:33.04 |
| How are we ever getting any cache hits? | 14:38.41 |
tor7 | Robin_Watts: the cache lookup quantizes the subpixel positions when converting to said unsigned char | 14:39.33 |
Robin_Watts | tor7: where? | 14:39.49 |
tor7 | Robin_Watts: draw-device.c | 14:41.56 |
| HSUBPIX and VSUBPIX | 14:42.03 |
Robin_Watts | Ah! | 14:42.24 |
tor7 | fz_draw_fill_text calls the QUANT macro when calculating trunc_trm | 14:42.26 |
Robin_Watts | I was looking in draw-glyph.c | 14:42.33 |
tor7 | not the most obvious place for it, I admit... | 14:42.33 |
| the glyph cache is insensitive to the number of subpix positions (it just rounds to 1/256th subpixel) | 14:43.12 |
Robin_Watts | right. | 14:43.34 |
tor7 | henrys: paulgardiner: what is this fasting thing you're always talking about? | 14:45.23 |
Robin_Watts | I have a commit on robin/master that tweaks draw-glyph.c to use sub pixel quantisation that depends on the size of the glyphs. | 14:45.24 |
| Let me move that logic out of draw-glyph into draw-device. | 14:45.47 |
tor7 | Robin_Watts: ah, right | 14:46.11 |
Robin_Watts | or maybe we should remove the logic in draw-device and leave it in draw-glyph... | 14:46.37 |
tor7 | or move it out of draw-device into draw-glyph? | 14:46.38 |
Robin_Watts | Having the logic in draw-glyph feels more sensible to me. | 14:47.00 |
paulgardiner | tor7: It's called 5/2 and involves reducing calorie intake to around 600 on two days a week | 14:47.06 |
Robin_Watts | Let me try it and see how it turns out. | 14:47.15 |
tor7 | paulgardiner: does it work well for you? | 14:48.13 |
paulgardiner | As a way of dieting, I prefer it than being constantly careful. It's supposed to have other health benefits aswell, but those are unproven. | 14:51.22 |
tor7 | paulgardiner: right. I've noticed I ought to start being more careful about what I eat, but so far I haven't had the discipline to actually do anything about it... | 14:52.10 |
paulgardiner | I find it's a constant battle. | 14:52.59 |
henrys | Robin_Watts: there's a meeting today? | 15:01.19 |
Robin_Watts | henrys: I think they have meetings all week. | 15:02.03 |
| 2 a day all week, from what scott said. | 15:02.16 |
| I was just trying to judge urgency. | 15:02.28 |
henrys | Robin_Watts: oh I thought there was a conference call. | 15:03.29 |
| bbiab | 15:03.39 |
kens | ray_laptop : thanks for checking the setpagdevice/EndPage behaviour with CPSI | 15:14.27 |
tkamppeter | Has anyone tried to build GS against the system's libopenjpeg? Is this possible? | 15:15.38 |
kens | tkamppeter : I htink you will have to check with chrisl who is not here today, but I believe this is not currently possible. | 15:16.13 |
ray_laptop | kens: sorry for thinking that Acrobat was probably broken. Hard to believe that is the way Adobe meant it to work. It totally breaks N-up printing if someone wants to change some behavior between pages (such as UseCIEColor) | 15:17.27 |
kens | ray_laptop : if I hadn't seen it I wouldnt' ahve believed it. I had got to an empty dictioanry because I was testing to see which keys caused it to deactivaye the device.... | 15:18.07 |
tkamppeter | kens, OK, will solve this with chrisl tomorrow. | 15:19.06 |
kens | tkamppeter : I blieve the problem is that 1.5 needs patches applied form us which the upstream maintainers aren't really interested in, and so does 2.0. I think we've passed all the patches upstream, but they are not all yet adopted (I could be mistaken) | 15:20.07 |
| That said, I can't see a commit which mentiosn it in the log, so I could be wrong completely | 15:20.46 |
ray_laptop | tkamppeter: I am not sure, but we used to have a set of patches to openjpeg commit ef8503414 Aug 2012 said "Reapply our patches to OpenJpeg-1.5" | 15:22.36 |
kens | I think our OpenJPEG 2.0 integration is not yet completed ? Bug 694311 | 15:23.45 |
| Shelly's comment 15 suggests all bugs we know of are fixed and passed upstream though. | 15:24.24 |
tkamppeter | kens, OK, so with 9.10 I will return to use the built-in openjpeg. | 15:27.05 |
kens | tkamppeter : I think that is best, its the only way that will not iontroduce bugs we have already fixed. | 15:27.25 |
ray_laptop | tkamppeter: so, the upshot is "use someone else's libopenjoeg with gs at your own risk". We only test with our version. You are welcome to compare the sources to what you have. | 15:27.45 |
kens | Hopefully, since we've passed out fixes upstream, at some point in the future we can use teh canonical release of OpenJPEG | 15:28.01 |
ray_laptop | right now we are in the mode of cherry picking CVE patches to openjpeg and porting them to our code. Moving to a standard release is in our interest, too. | 15:29.05 |
kens | is off for a pizza, night all | 15:53.46 |
ray_laptop | bye, kens | 15:53.55 |
Robin_Watts | night kens | 15:54.00 |
| tor7: subpixel patch on robin/master | 15:56.11 |
| 10000+ differences, as you'd expect. The bmpcmp shows 1000 of them. | 15:56.27 |
| I have a crap eye for such things, but none of the diffs offend me. | 15:56.40 |
| Stupid dev board USB interface thing took out my PC :( | 16:22.07 |
sebras | Robin_Watts: what devboard? rpi? | 16:32.58 |
Robin_Watts | sebras: No, one we had from a potential partner ages ago, that i've just been asked to dust off. | 16:33.26 |
sebras | Robin_Watts: I remember that older versions of windows had severe problems with the USB-stack and devices that might not behave according to spec. | 16:33.53 |
Robin_Watts | sebras: I suspect this is a case of a crap driver causing a blue screen though. | 16:44.33 |
sebras | Robin_Watts: oh, they have their own driver? mm, then that seems likely too. | 16:45.50 |
| Robin_Watts: as a mupdf dev one is mostly shielded from those kinds of HW annoyances. ;) | 16:46.23 |
Robin_Watts | Crap Driver does not preclude it being a Microsoft thing :) | 16:46.24 |
| tor7: ping | 18:01.25 |
sebras | Robin_Watts: FYI: he's been away for an hour on the other chat client too | 18:11.00 |
Robin_Watts | sebras: I just wanted to run an idea past him, but I've gone ahead and done it the way I think is best, and we can fight over it tomorrow if he disagrees. | 18:17.26 |
sebras | Robin_Watts: merge it and pit his sense of aesthetics against his laziness. ;) | 18:25.38 |
tor7 | sebras: oi! I heard that! | 18:27.18 |
sebras | tor7: go back to sleep... | 18:27.28 |
Robin_Watts | tor7: I've pushed all the subpixel logic inside of fz_render_glyph | 18:29.12 |
tor7 | Robin_Watts: looking at that patch right now | 18:29.24 |
Robin_Watts | we don't need the trunc_trm stuff anymore - we pass trm into the function, and trm->{e,f} are updated to allow for the results. | 18:29.32 |
tor7 | the quantisation logic eludes me | 18:29.33 |
Robin_Watts | tor7: it's stale... hold on. | 18:29.39 |
| to quantise x, I do: x += rounding; quantised_x = ((int)(x*256)) & q | 18:31.25 |
| for suitable values of rounding and q. | 18:31.45 |
| if q = 128, for instance, quantised_x will end up being 0 or 128 (0 or 0.5 as a float). rounding is therefore 0.25. | 18:32.33 |
tor7 | oh, &q is just trimming off the low bits? | 18:32.34 |
Robin_Watts | yes. | 18:32.43 |
tor7 | Robin_Watts: the trunc_trm was to ensure that the matrix to render_glyph always had 0 <= e,f < 1 | 18:37.00 |
Robin_Watts | tor7: yeah, I realise that. | 18:37.20 |
| fz_render_glyph now ensures that local_ctm.{e,f} are within 0..1 | 18:37.46 |
| effectively I've just pulled that logic inside of fz_render_glyph. | 18:38.34 |
| so it's only needed in 1 place (well, 2, cos there is fz_render_stroked_glyph too) rather than at every call site. | 18:38.58 |
| the side effect of needing less copies of matrices is just a nice side effect. | 18:39.15 |
tor7 | it's a handful of steps to follow, local_ctm.e to key.e back to local_ctm.e then to ctm->e + e | 18:39.35 |
Robin_Watts | tor7: I could comment it better, I guess. | 18:40.04 |
tor7 | or use temporaries with better names :) | 18:40.24 |
Robin_Watts | local_ctm.{e,f} ends up being just the subpixel adjustment. | 18:41.14 |
| e and f are the non-subpixel adjustment. | 18:41.33 |
| ctm->{e,f} ends up being the sum of the two (i.e. the actual translation) | 18:42.10 |
tor7 | yes. | 18:42.37 |
Robin_Watts | key.{e,f} is the same as e and f, but in int form. | 18:42.40 |
| Suggestions for better names always appreciated :) | 18:42.49 |
tor7 | subpix_e, pix_e, ctm->e = pix_e + subpix_e | 18:43.30 |
| local_ctm.e = subpix_e | 18:43.43 |
| maybe | 18:43.49 |
Robin_Watts | how about s/local_ctm/subpix_ctm/ ? | 18:46.02 |
| subpix_ctm.e += r; | 18:49.41 |
| pix_e = floorf(subpix_ctm.e); | 18:49.43 |
| subpix_ctm.e -= pix_e; | 18:49.45 |
| subpix_ctm.f += r; | 18:49.46 |
| pix_f = floorf(subpix_ctm.f); | 18:49.48 |
| subpix_ctm.f -= pix_f; | 18:49.49 |
| key.e = (int)(subpix_ctm.e * 256) & q; | 18:49.51 |
| key.f = (int)(subpix_ctm.f * 256) & q; | 18:49.52 |
| ctm->e = subpix_ctm.e + pix_e; | 18:49.54 |
| ctm->f = subpix_ctm.f + pix_f; | 18:50.00 |
| tor7: New patch up. | 19:02.40 |
| tor7: Is there any point in doing any subpixel adjustment if AA is disabled? | 19:09.29 |
| Does freetype return us different bitmaps with the different subpixel adjustments we do? | 19:09.55 |
tor7 | Robin_Watts: if AA is disabled, hinting should be enabled, and I guess we ought to round to integer coords | 19:17.50 |
| Robin_Watts: yeah, that patch looks good. remind me to look over the RLE stuff again tomorrow. | 19:19.28 |
MaDog | hi | 22:05.30 |
| i have a problem when pdf->pcl convertation doing | 22:05.55 |
Guest29405 | with -sDEVICE=djet500c | 22:06.51 |
| and the bacground is black evry row where not word or graphics the page is black | 22:07.17 |
| what is the problem? | 22:07.23 |
| i also try -dNOSUBDEVICECOLORS -NOINTERPOLATE -dNOTRANSPARENCY | 22:09.12 |
| has anybody glue what is the problem? | 22:09.30 |
| and i using ghostscript 9.09 | 22:10.39 |
Robin_Watts | Guest29405: How are you displaying the PCL? by sending it to a printer? | 22:35.35 |
Guest29405 | yes, and use PCLReader | 22:38.28 |
| the printer type Xerox Phaser | 22:39.37 |
Robin_Watts | I suspect that the PCL the dj500 device produces is just not compatible with that printer. | 22:54.44 |
Guest29405 | PCLReader? | 22:55.50 |
| my problem is convert a pdf file to pcl5 color files | 22:56.14 |
| when i using PCLReader to watch the pcl file is also has black background | 22:56.42 |
| and a HP Photosmart 5510 also print black background | 22:57.12 |
Robin_Watts | Guest29405: Right, so probably you want to use something other than -sDEVICE=djet500c | 23:03.06 |
Guest29405 | so what device is create color pcl5? | 23:03.38 |
| not pcl xl | 23:03.52 |
Robin_Watts | cljet5 maybe? | 23:06.14 |
| Failing that, try back here in about 18 hours and speak to henrys. | 23:06.56 |
Guest29405 | ok thx | 23:07.43 |
henrys | Guest29405: I'm back reading... | 23:13.28 |
Guest29405 | cool:) | 23:14.33 |
henrys | I'll have a look at the driver output it could be a bug. Did you try ghostpcl to view it? | 23:15.23 |
Guest29405 | no | 23:15.45 |
henrys | you tried a variety of files and the same result? | 23:16.15 |
Guest29405 | yes, but all same source | 23:16.53 |
henrys | the best device for modern pcl coolr is pxcolor ⦠but I'm not sure if your xerox supports that. | 23:18.06 |
Guest29405 | when i convert the pdf to postscript , the output is good | 23:18.13 |
| pcl-xl is not supported | 23:18.30 |
| can i use an older ghostscript? | 23:18.59 |
henrys | I don't think there has been a regression in dj500 if that is what you are thinking. What's the printer model number? | 23:20.13 |
Guest29405 | 6360ND | 23:20.27 |
| DN sory | 23:20.42 |
| but an older and droped device can solve my problem? | 23:21.02 |
| i try convert RGB output icc profile | 23:21.52 |
| i try convert pdf->pdf with rgb profile with compatibility level 1.4 | 23:22.41 |
| but when convert the new pdf->pcl problem still have | 23:23.09 |
henrys | xerox is not very helpful describing what languages are availablle. | 23:23.11 |
| why wouldn't you simply send postscript (ps2write device) to this printer? | 23:23.33 |
Guest29405 | the main problem is when i use the PCLReader 13.10 from pagetech the pcl has the black background | 23:24.10 |
| becouse i need use pcj username | 23:24.38 |
| djet500 not has the black background | 23:25.33 |
| but not color pcl:) | 23:25.50 |
henrys | well cljet5c will work but it will create a large unwieldy raster. | 23:25.58 |
Guest29405 | gs 9.09 not know this device | 23:26.52 |
| can i compile with something? | 23:27.16 |
henrys | djet500 is not pcl, ti appears to be custom to the device or device family. | 23:29.26 |
| Guest29405:windows linux? | 23:30.34 |
Guest29405 | windows 7 | 23:30.55 |
| i try it with gs 8.71 with djet500c | 23:32.40 |
| result is everywhere not word or image the background are black | 23:33.16 |
| in gs 9.09 only black the line where not has anything | 23:33.45 |
henrys | add cljet5c.dev in msvc.mak then rebuild clean with visual studio. | 23:34.26 |
| look at the other .dev examples in the makefile | 23:34.43 |
Guest29405 | ok thx i will try out it | 23:35.01 |
henrys | Guest29405: The commands I see in the djet500c device are not standard pcl. | 23:35.37 |
Guest29405 | where can i find this mak? | 23:44.21 |
henrys | psi/msvc.mak | 23:44.39 |
Guest29405 | thx | 23:45.32 |
| i rebuild it whit cljet5c | 23:55.52 |
| the black background goes away | 23:56.07 |
| the first page look great, but all other are wrong | 23:56.55 |
| lot of bad character not and other craps | 23:58.03 |
| Forward 1 day (to 2013/08/29)>>> | |