| <<<Back 1 day (to 2015/03/30) | 20150331 |
mvrhel_laptop | ah. looks like the final error I have is due to a missing runtime dll | 00:46.20 |
| need to have nsis insure that this is present | 00:46.34 |
| dinner time now | 00:46.52 |
henrys | Robin_Watts: do you have all those Ctrl-M's in the first file does that mess up the offsets? | 00:53.33 |
mvrhel_laptop | blah. looks like I need to package up the run time redistributable into nsis | 03:06.17 |
| to make this work seamlessly | 03:06.33 |
ncopa | hi | 06:42.50 |
ghostbot | Welcome to #ghostscript, the channel for Ghostscript and MuPDF. If you have a question, please ask it, don't ask to ask it. Do be prepared to wait for a reply as devs will check the logs and reply when they come on line. | 06:42.50 |
ncopa | how can i build mupdf as shared library? | 06:43.23 |
kens | Hmm I wonder if bug #695893 is actually a MuPDF bug report..... | 08:05.27 |
| Ah and indeed pdf_set_free_text_details is a MuPDF function. I bet that's it..... | 08:07.39 |
chrisl | He says in the last comment: "I am running on the 'Android Phone' with mupdf-1.6." | 08:08.04 |
kens | I haven't seen that yet.... | 08:08.16 |
chrisl | I think you should close the bug as "invalid"..... ;-) | 08:08.24 |
kens | We need a new category closed as 'user is an idiot' | 08:08.51 |
| That gets him off my list :-) | 08:10.35 |
Robin_Watts | henrys: The Ctrl-M's in the first file were present in the original. | 09:33.09 |
| The files differ purely in the last few bytes. | 09:33.26 |
tor8 | Robin_Watts: so... mutool has managed to include the font data files yet again :( | 09:34.12 |
| and mudraw is in serious need of a facelift | 09:34.23 |
| possible solution: make mudraw a tool in mutool's toolset and stop worrying about bloating the binary | 09:35.03 |
| and at the same time, give the command line options a working over | 09:35.17 |
| mudraw -t and -x feel wrong | 09:35.32 |
| mudraw -F txt,html,xml and mudraw -F display would be more appropriate, IMO | 09:35.54 |
| and all the -5, -m, -M, -T show-extra-information could be one flag with arguments, like "-s mem,time,md5" or "-s mt5" | 09:36.56 |
Robin_Watts | I would prefer mudraw and mutool to be separate I think, but I'm not wedded to that. | 09:38.26 |
tor8 | I'm just thinking of download size | 09:38.42 |
Robin_Watts | A sanitisation of the command line options is probably not a bad idea. | 09:38.51 |
tor8 | if we can get the font data back out of mutool, keeping them separate is also my preference | 09:39.02 |
| but if we have 3 exes with duplicated font data, merging the two command line tools makes sense to me | 09:39.24 |
| I wonder if there's a tool to query the linker dependency somehow | 09:40.38 |
| to see which files pull in another file | 09:40.50 |
Robin_Watts | tor8: comment out the data, build, see what misses it. | 09:45.38 |
| comment that out, build, see what misses that. | 09:45.48 |
| rinse and repeat :( | 09:45.52 |
tor8 | yeah... just wish there was a magic incantation I could run on libmupdf.a | 09:46.50 |
| I guess I could cook one up based on the 'nm' output | 09:46.59 |
| to build a map of which files depend on another file | 09:47.11 |
Robin_Watts | tor8: yeah... | 09:47.19 |
| tor8: I did a fix for bug 695892 last night. | 10:02.04 |
| but when I tested the fix with pdf_reference17.pdf it produces a file that Acrobat wants to repair on closing. | 10:02.36 |
| gs and mupdf are both happy with it. | 10:02.46 |
| I have no idea why it's broken. | 10:02.52 |
| The file can be found as http://ghostscript.com/~robin/WhyAmIBroken.pdf | 10:03.12 |
kens | Something to do with it being linearised ? | 10:03.19 |
Robin_Watts | kens: I really don't know. I can't see why. | 10:03.34 |
| Basically, it's the original file unchanged, with 2 objects added to the end. | 10:03.53 |
kens | Maybe Acrobat doesn't like incremental updates on a linearised file | 10:03.53 |
Robin_Watts | and it all looks plausible to me. | 10:03.57 |
kens | Oh, let me go look | 10:04.03 |
| Oh, big file :-( | 10:04.20 |
Robin_Watts | yeah, I tried to find a smaller file that it would go wrong with, but... no. | 10:04.36 |
kens | Hmm, your version of pdf_reference17.pdf differs from mine, mine is a PDF 1.4 file, yours is a PDF 1.6 file | 10:07.08 |
| The only thing I can think of is that Acrobat thinks L:inearised files should not have incremental updates and so wants to rewrite the file. I can't see anything actually wrong with the file. | 10:09.31 |
Robin_Watts | ok, then I don't care. Thanks. | 10:09.57 |
kens | NP | 10:10.06 |
| But I'll jkust quickly refresh my memory on Linearised files.... | 10:10.17 |
Robin_Watts | With linearised files, if the file length given by /L does not match the length of the file, the reader is supposed to realise that it's not linearised any more. | 10:11.35 |
kens | Yeah, and I think that is maybe why Acrobat wants to rewrite it. It claims to be linearised, but it can't really be, so Acrobat wants tot fix it | 10:12.14 |
tor8 | Robin_Watts: it's not something bizarre like mixing old-style xref sections with new-style xref streams? | 10:32.41 |
Robin_Watts | I can't see anything that says that's illegal. | 10:32.58 |
tor8 | then I don't see anything wrong with it either | 10:33.12 |
| so I guess what kens said is the likely problem | 10:33.34 |
kens | I'm not sure I'd describe that as a 'problem' :-) | 10:33.50 |
tor8 | Robin_Watts: I think I've found the problem, using a script to dump file/symbol dependencies from 'nm' | 10:36.44 |
Robin_Watts | tor8: cool... | 10:36.56 |
tor8 | doc->super.write = pdf_write_document is one of them... | 10:39.19 |
| Robin_Watts: ugh. pdfclean pulls in pdfwrite pulls in signatures pulls in annotation synthesis pulls in fonts | 11:54.55 |
Robin_Watts | Ah. Maybe it's possible to break it at the signatures point. | 13:04.43 |
tor8 | Robin_Watts: mudraw command line syntax rework on tor/master if you want to have a look and go "blech!" | 13:15.19 |
Robin_Watts | fetched | 13:20.27 |
| You removed my helpful bit of mujs code in a comment :( | 13:22.28 |
tor8 | your helpful bit of mujs code was in the wrong file! | 13:22.39 |
| or it was just plain wrong | 13:22.50 |
| there is no 'j' flag | 13:22.53 |
Robin_Watts | Why 2 \n's after -p password ? | 13:23.20 |
tor8 | it's in mjsgen.c now | 13:23.29 |
Robin_Watts | oh grouping, I see. | 13:23.30 |
tor8 | ideally, the text output prolog and trailer cruft should be in separate functions | 13:24.49 |
| and all the output should go through the fz_output we open | 13:24.59 |
Robin_Watts | Have we lost -b ? | 13:25.01 |
tor8 | I renamed it -A | 13:25.06 |
Robin_Watts | ok. | 13:25.12 |
tor8 | made the 'advanced' flags uppercase | 13:25.23 |
Robin_Watts | -d -> -D is gonna screw me up for months. | 13:25.50 |
tor8 | do you use that option often? | 13:27.19 |
Robin_Watts | When debugging the display list :) | 13:27.40 |
| Looks fine to me. | 13:27.50 |
tor8 | we could make all but -R and -F lowercase | 13:27.54 |
Robin_Watts | 'I' and 'i' collide. | 13:29.06 |
tor8 | ah, yes | 13:29.14 |
Robin_Watts | 'i' could become 'e' | 13:29.19 |
tor8 | or -n | 13:29.26 |
Robin_Watts | 'i' could become 'a' for 'aspectless' ? | 13:29.56 |
tor8 | but make uses -i for ignore-errors | 13:30.08 |
Robin_Watts | sorry, 'f' could become 'a' for 'aspectless' | 13:30.18 |
tor8 | -F for force fit (or -w could be -W for force-width) | 13:30.43 |
Robin_Watts | -w and -h set width and height. | 13:30.59 |
tor8 | -W could set width *and* fit | 13:31.11 |
Robin_Watts | ew. | 13:31.16 |
tor8 | but yeah. I rather like the current way it works. | 13:31.24 |
| a bit awkward to explain, but all the useful ways work | 13:31.40 |
Robin_Watts | I'm fine with it however. | 13:33.30 |
tor8 | then I shall rewrite the manpage | 13:33.56 |
ChenChen | Hi guys, it's us again (sorry) | 13:41.21 |
| We've managed to trace the issue we were having to one image and have produced a working and non-working pdf with the same image | 13:41.37 |
| Working: https://drive.google.com/file/d/0ByL_VDOdxc3MQjdMUms5Mnl2STg/view?usp=sharing Broken: https://drive.google.com/file/d/0ByL_VDOdxc3MVjdTcjZaczFMdGc/view?usp=sharing | 13:41.39 |
| The only difference is we used a hex-editor to remove the following from non-working one to produce the working one: /SMask 7 0 R | 13:42.14 |
Robin_Watts | Ok, so a softmask is screwing stuff up. | 13:42.30 |
ChenChen | You know more than we do :) Do you have any idea what would be generating this? I don't think it's anything we're setting through PDFSharp | 13:43.26 |
Robin_Watts | ChenChen: Can I just ask, how exactly are you using gs here? Are you using it as part of a software package you distribute? Or just as an in house tool? | 13:43.31 |
| I suspect the image you are using has an alpha plane (or at least comes from a bitmap that might have an alpha plane, such as a png). | 13:44.05 |
ChenChen | It's part of a project which will act as a virtual printer server for our customers from our web application (are we right in thinking we need to open source the printing project to comply with licensing?) | 13:44.31 |
Robin_Watts | ChenChen: You will need to open source *everything* that runs on your website. | 13:45.10 |
| The GNU AGPL basically says that any customer that wants to should be able to set up their own copy of your server. | 13:46.20 |
ChenChen | The GS application will be a separate program that users install on their computers and which receives instructions from the web server - will this still require for the website to be open sourced too? | 13:46.23 |
Robin_Watts | I am not a lawyer, so I don't want to comment on that. | 13:47.03 |
ChenChen | Ok :) | 13:47.19 |
Robin_Watts | It sounds to me that you probably ought to be talking to Scott about a commercial license though. | 13:47.27 |
| It would make your life much easier regardless. | 13:47.33 |
ChenChen | indeed, I'll bring it up in the next meeting | 13:47.46 |
chrisl | ChenChen: what version of Ghostscript are you using? | 13:47.48 |
ChenChen | 9.15 | 13:47.56 |
kens | ChenChen : I htink the only way to debug this is to put code into the mswipr2 devie that feeds back information about what its doing. | 13:48.47 |
chrisl | Hmm, strange. All we can say is to report a bug, and we'll look at it at some point. Did you establish if it went wrong with other devices? | 13:48.54 |
ChenChen | As an aside, the licensing link on the GS licensing page is broken: http://ghostscript.com/Licensing.html | 13:48.57 |
kens | Regardless of teh presence of a softmask in the PDF, what should be happening is that GS renders the PDF to a bitmap. That bitmap is a flat simple raster. It gets Blitted to the printer device context and then the WIndows print system prints it. | 13:49.38 |
ChenChen | No, we haven't been able to. It also presents an issue printing to Microsoft XPS Writer | 13:50.53 |
chrisl | Licensing page fixed..... | 13:51.04 |
ChenChen | Thanks Chris | 13:51.07 |
| Would you agree that this seems to be a bug then? We'll file it as a bug if that's the case | 13:51.34 |
chrisl | I would say it was - there's no guarantee we'll be able to reproduce it, of course, and if we can't, then we can't fix it | 13:52.45 |
ChenChen | Yep sure, we'll try to provide as much information as possible. | 13:53.13 |
kens | If it goes wrong with the MS XPS writer there's a chance I could actually debug it | 13:53.15 |
ChenChen | :) | 13:53.19 |
| Thank you for all your help guys - we'll submit the bug report now | 13:53.26 |
| what category should we submit this under? PS Writer? | 13:57.13 |
kens | No, definitely not! | 13:57.27 |
| Just go with general | 13:57.54 |
ChenChen | thanks :) | 13:58.11 |
kens | Ah no 'other driver' | 13:58.11 |
| But its not vital, general will do anyway | 13:58.32 |
chrisl | With 9.16, the Win7 XPS writer, with default settings seems to work for me, using the above file..... | 13:59.49 |
kens | I cna try 9.15 too | 14:00.06 |
chrisl | The "working" file throws a warning for me.... broken xref. But renders fine | 14:02.08 |
kens | I'll give it a try when I see the report, I'm rebullding master at the moment | 14:02.35 |
ChenChen | ok it works with 9.16 | 14:06.53 |
| I feel like such an idiot now | 14:07.01 |
| Thank you for your help | 14:07.04 |
kens | NP I'm puzzled by the bug but its nice to know its fixed | 14:07.18 |
chrisl | Well, 9.16 did only come out yesterday, so you needn't feel bad | 14:07.36 |
Robin_Watts | tor8: In the mupdf viewer, if you spin the mouse wheel with ctrl held down, does that zoom in/out? | 14:12.47 |
kens | chrisl just for interest's sake, I managed to get a valid TrueType font with 2 GIDs down to 568 bytes. The one that Tesseract was using was 3.5Kb. Using their font and a CIDToGIDMap which maps all CIDs to GID 1 added about 200 bytes to the file size. However it then produces a file which pdfwrite doesn't complain about. | 14:13.21 |
chrisl | kens: that *must* be the best solution for them | 14:14.04 |
kens | I wold have thought so, its smaller than their current result, the font is valid, and it will work even for tools like pdfwrite which actually poke the font machinery | 14:14.41 |
| Admittedly pdfwrite breaks their files that use non-Latin character codes, I'm not certain why but I don't think its unique to their files any more, I think its a general problem with pdfwrite and CIDFonts | 14:15.45 |
| I should probably look into it one day. In the meantime, pre-meeting coffee | 14:16.08 |
tor8 | Robin_Watts: I don't know ... I don't have a mouse wheel :/ | 14:19.33 |
Robin_Watts | hah! | 14:19.57 |
kens | Do you expect it to do something ? | 14:20.26 |
tor8 | you can pry my HP DY651-A out of my cold dead hands! | 14:20.28 |
henrys | good idea kens more coffee... | 14:20.32 |
Robin_Watts | kens: Shift and Ctrl are misbehaving for me under windows. Trying to see if my expectations are wrong, or whether I've coded the modifier reading badly. | 14:22.01 |
kens | Oh, I tried it on MuPDF 1.6 and it didn't do anything | 14:22.24 |
| I wasn't sure if that was expected or not | 14:22.36 |
| Ooh, it works on Miranda, I never knew that.... | 14:23.42 |
Robin_Watts | Ah. the mousewheel events are getting mapped to key events rather than mouse ones under windows. | 14:24.58 |
| Ctrl+mousewheel should zoom. Works in chrome etc. | 14:25.46 |
| and on linux mupdf. | 14:25.55 |
rayjj | morning, all | 14:28.22 |
kens | Morning | 14:28.27 |
henrys | good morning and afternoon | 14:29.30 |
fredross-perry | howdy | 14:29.45 |
ghostbot | Welcome to #ghostscript, the channel for Ghostscript and MuPDF. If you have a question, please ask it, don't ask to ask it. Do be prepared to wait for a reply as devs will check the logs and reply when they come on line. | 14:29.45 |
rayjj | Robin_Watts: I wonder what the extreme right edge pseudo-scroll wheel of my touchpad maps to | 14:30.00 |
henrys | starting with the bug report - I can clean up my 2 customer bugs soon, seems like kens ouputpage thing should be an enhancement? | 14:30.15 |
Robin_Watts | rayjj: On windows? WM_MOUSEWHEEL events I should think. | 14:30.21 |
kens | hnerys is that FirstPage and LastPage ? | 14:30.33 |
henrys | kens: yes | 14:30.39 |
kens | It probably should be an enhancement I guess yes | 14:30.51 |
henrys | then rayjj and mvrhel_laptop have bugs that look like border line enhancement to me also, rather not have them classified as defects. | 14:31.17 |
rayjj | henrys: which ones ? | 14:31.35 |
henrys | rayjj: 695568 | 14:32.10 |
rayjj | bug 695568 is one that is sort of stale because I am not able to reproduce it | 14:32.23 |
henrys | rayjj: on the bug report to "tech" | 14:32.25 |
chrisl | Several of us have been unable to reproduce Norbert's results on 695568 | 14:33.07 |
kens | In which case close as 'worksforme' ? | 14:33.18 |
marcosw | morning | 14:33.36 |
kens | Morning marcosw | 14:33.44 |
chrisl | Well, I guess the only thing is, he seems to be using a different method for his timings, so I wonder if that's influencing things?? | 14:33.50 |
henrys | rayjj: tell norbert what you found and close it. | 14:33.55 |
rayjj | maybe ask norbert to test with ootb 9.16 (without his changes) | 14:33.57 |
kens | Or ask again for a tarball which can reproduce the difference | 14:34.21 |
| But if we don't do something, it will just linger | 14:34.33 |
chrisl | We'd need two tarballs, for the comparison | 14:34.49 |
henrys | mvrhel_laptop: and we decided you were going to look at that weird one about 256 colors at the meeting? | 14:34.52 |
rayjj | henrys: I'll try once more with out of the box 9.16, then pass it back to him | 14:34.59 |
kens | chrisl : true, but its not that vast | 14:35.06 |
henrys | rayjj: thanks I just want to keep the customer bug list short, so when something is there it gets fixed. | 14:36.04 |
| Yeah chrisl for beating the release deadline! | 14:36.43 |
rayjj | henrys: I've got two "hot" issues from cust 532. One is a pdf14 blending issue that I'm going to open a bug on, although I will solve it for the customer. I talked to mvrhel about it yesterday. It exists in our HEAD code | 14:37.06 |
chrisl | It was a bit close this time :-) | 14:37.07 |
rayjj | the other one is a performance issue to "investigate" that they are also working on | 14:37.49 |
henrys | at the meeting we tagged fredross-perry with a couple of todo's but I'm not sure if he saw that? does fredross-perry have access to the regression dashboard so he can get to workflowy? | 14:37.50 |
kens | I believe so yes | 14:38.22 |
rayjj | all you need for the regression dashboard is a password from marcosw | 14:38.34 |
henrys | rayjj: well that should be a P1 customer bug, that really belongs in the list. | 14:38.37 |
kens | IIRC robin was going through regressions a few days back | 14:38.37 |
fredross-perry | Sorry I did not see the TODs. I do have dashboard access. | 14:38.37 |
mvrhel_laptop | henrys: yes about the 256 colors | 14:38.56 |
| sorry I am late | 14:38.59 |
henrys | fredross-perry: so from the dashboard go to work items, that's the agenda from live meetings - it is confidential. | 14:39.27 |
| fredross-perry: then just search for fred on that page. I just use browser seach , workflowy search collapses everything. | 14:40.33 |
mvrhel_laptop | I am also hoping to have a new windows gsview release today or tomorrow. I am hopeful that this will be good for a beta release | 14:40.42 |
| as I fixed a pile of things | 14:40.54 |
henrys | mvrhel_laptop: okay that was on my list, let's get the thing out there and see what people do with it! | 14:41.32 |
mvrhel_laptop | yes | 14:41.48 |
henrys | rayjj: can you post a font page for Postscript and PCL for your samsung printer? | 14:42.11 |
| fredross-perry: did you get to the agenda? | 14:43.05 |
fredross-perry | yes I see it now | 14:43.15 |
henrys | marcosw: are we missing a response to Grillo? | 14:43.33 |
kens | Marcos asked Michael about it, and I passed a reply last night | 14:43.59 |
| Or was it this morning.... | 14:44.16 |
henrys | kens: oh cool I must have missed it. | 14:44.19 |
kens | Might have been private mail | 14:44.28 |
| Yeha just to Michael and marcos apparently | 14:44.46 |
henrys | I didn't have much else for the meeting, on my plate today is to contact URW about getting otf CFF fonts for all our fonts. We are still in agreement we should use CFF everywhere possible? chrisl ? | 14:45.54 |
rayjj | henrys: I did send the PCL and PS font list as an attachment "Scan_20150309_121148.pdf" on my email of March 9 "Samsung CL 1860 Printer alive and well." | 14:46.24 |
henrys | rayjj: shame on me. sorry | 14:46.43 |
rayjj | np | 14:46.48 |
chrisl | henrys: if we're insisting on changing then OTF/CFF makes most sense...... | 14:47.04 |
| henrys: how much overlap is there between the Postscript 3 136 fonts and the PCL 80? | 14:47.46 |
henrys | chrisl: I believe the 80 are contained within the 136 but I'll check to make sure. | 14:49.04 |
marcosw | henrys: i'll forward kens response | 14:49.05 |
chrisl | henrys: I seemed to get the impression in Denver that we couldn't share the fonts between PS and PCL (glyph sets maybe??) | 14:49.41 |
henrys | chrisl: don't remember that. We aren't sharing symbol/encoding sets | 14:51.34 |
chrisl | henrys: I think I mentioned just getting one font dump from URW+. that being the 136 PS3 ones, and pulling out the PS2 35, and the PCL 80, and you said we'd still need a specific PCL font set from them - but I can't remember the exact reason | 14:53.00 |
henrys | chrisl: jyes that's right, I'm hoping for a superset from URW. | 14:53.45 |
chrisl | Oh, okay. Then, yes, OTF/CFF is the way to go | 14:54.11 |
henrys | anything else meeting wise? off to skype next. | 14:54.55 |
kens | henrys, can you ask Miles about hte June meeting please ? | 14:55.15 |
chrisl | I just don't want to go to the effort, and still find we need two separate font sets for the language switching setup | 14:55.15 |
kens | Its now < 10 weeks, and soon the pirces of flights will go up and availability will go down | 14:55.37 |
henrys | chrisl: the effort what? We'd want otf even if we couldn't share for compression right? | 14:56.05 |
kens | WHy ? | 14:56.32 |
| THat is, why change ? | 14:56.39 |
chrisl | henrys: we don't gain a great deal in size for Postscript because the Type 1's zlib compress better than the CFF ones | 14:56.47 |
henrys | kens: if we did ship a 136 build it would be much liter. | 14:56.53 |
| lighter | 14:56.56 |
kens | henrys, what chrisl said | 14:57.03 |
henrys | kens: oh I thought we came to the conclusion cff was superior. | 14:57.42 |
kens | No, the conclusion was that *if we have to change* then it has to be CFF< not TrueType | 14:58.02 |
henrys | I mean CFF in an otf wrapper | 14:58.28 |
kens | But I was under teh impression that the need to change was driven by having a single set of fonts for the language switch buil;d | 14:58.30 |
chrisl | As was I..... | 14:58.45 |
henrys | kens, chrisl: It was then we saw the cff did better the gzip and .... | 14:59.33 |
kens | henrys the difference in file size is not sufficient in my opinion ot warrant changing. | 15:00.09 |
| As chrisl says, the type 1 outliens compress better | 15:00.26 |
| And changing will involve numerosu small diffs in the tests, as well as me having to fix at least some cases in the pdfwrite code. | 15:00.52 |
| Sticking with type 1 fonts doesn't have those problems. | 15:01.04 |
chrisl | CFF is better than Type 1 compressed, but not hugely better. And I don't know about OTF/CFF. It's hard to judge as we can't reproduce the same CFF fonts as URW+ produce | 15:01.25 |
kens | I'd prefer to stick with type 1s for those reasons. However, if we have to have a single set of fonts, then its important that hte fonts we use are suitable for PostScript, which means that OTF fonts with CFF otulines is the only reasonable solution | 15:01.41 |
henrys | kens: okay we'll revisit it. off to skype now | 15:01.41 |
rayjj | mvrhel_laptop: here goes... I'm starting a regression test with the fix/hack to avoid double application of the transfer function. | 15:30.56 |
| marcosw: does the regression dashboard "add test files" replace an existing one (if I later improve it) ? | 15:32.09 |
kens | It puts them in Subversion, so yes | 15:32.30 |
Robin_Watts | rayjj: That's my script. | 15:32.42 |
| and no. | 15:32.44 |
kens | Really ? | 15:32.48 |
Robin_Watts | It does an svn import, so it will complain if the two files are the same. | 15:33.00 |
| You can always manually commit a fixed version. | 15:33.10 |
| but the script will not overwrite. | 15:33.24 |
chrisl | I wonder if we can get rid of the old, long obsolete Windows DLL stuff....... | 15:35.59 |
rayjj | Robin_Watts: but then I have to have the entire svn locally, right ? | 15:36.05 |
kens | chrisl what WIndows DLL stuff ? | 15:36.21 |
Robin_Watts | rayjj: Like you do on peeves? and peeved? | 15:36.33 |
chrisl | kens: The stuff in gsdll.c etc | 15:36.53 |
rayjj | Robin_Watts: well, the "regression" user has it, so I can sudo su regression I suppose | 15:37.11 |
kens | Hmm, is that obselete ? I've never looked at it | 15:37.12 |
Robin_Watts | rayjj: yeah. | 15:37.42 |
chrisl | kens: Yeh, I think the current DLL interface is in dwdll.c/h | 15:37.59 |
kens | Oh yes, that makes sense. Well if we aren't using the other stuff we should kill it. It probably doesn't work anyway :-) | 15:38.23 |
chrisl | Well, DLL.htm was changed in 2001(!!) to say: "WARNING: The API described in this document is obsolete and will be removed in the future" | 15:39.00 |
kens | I think 14 years is long enough warning, even for us | 15:39.19 |
rayjj | Robin_Watts: actually, the user for the svn is marcos. Interestingly, on peeved the jeita dir and 5 other files show up as '?' from svn st | 15:40.43 |
Robin_Watts | rayjj: svn is another country, I'm afraid. | 15:41.30 |
| My svn foo is not strong. | 15:41.35 |
rayjj | I've managed to forget most of it as well :-) | 15:42.08 |
chrisl | kens: of course, just removing the old DLL stuff won't work, as it still has hooks into one of the devices (which is pretty frightening, in itself!) | 15:44.28 |
kens | That sounds so badly wrong..... | 15:44.47 |
rayjj | I think that implies that marcos has some uncommited files lying around on the cluster master (casper) that the cluster machines rsync from. I checked and peeves has the same files tagged with '?' | 15:44.53 |
chrisl | kens: I cannot disagree with that assessment! | 15:45.14 |
| I guess this was an early punt at something like the display device | 15:45.49 |
Robin_Watts | rayjj: hmm. That's possible. | 15:45.50 |
| I have an svn checkout here. | 15:45.55 |
| What path? | 15:46.01 |
kens | chrisl tht is possible I guess, like I say, I've never actually looked at it | 15:46.09 |
Robin_Watts | I'll see if they show up in my copy. | 15:46.14 |
rayjj | Robin_Watts: sorry... test_private/jeita for one, tests_private/customer_tests/F041.bin.xl and tests_private/comparefiles/Bug690379.pdf are a few | 15:48.08 |
| that's tests_private/jeita | 15:48.27 |
Robin_Watts | My tests_private has no jeita. | 15:48.59 |
chrisl | kens: I have doubts about the efficacy of the mswindll device on modern Windows boxes | 15:49.18 |
Robin_Watts | So I think marcosw needs to check in the jeita dir. | 15:49.18 |
kens | Let me see.... | 15:49.29 |
rayjj | Robin_Watts: right. So they aren't in svn even though they are laying around in tests_private | 15:49.30 |
Robin_Watts | rayjj: Yeah. | 15:49.36 |
chrisl | kens: devices/gdevwdib.c | 15:49.43 |
Robin_Watts | rayjj: Does peeves/peeved have a pdf/uploads directory in tests_private ? | 15:49.56 |
rayjj | I'll leave that up to marcosw | 15:50.05 |
henrys | tor8, Robin_Watts pointed out we forgot to talk about the mupdf release. | 15:50.43 |
rayjj | Robin_Watts: nope | 15:50.47 |
kens | chrisl it probably works I'd have thought, I don;t think the DIB stuff has changed in decades | 15:50.49 |
rayjj | Robin_Watts: oops. Sorry. Yes, it is there | 15:51.07 |
Robin_Watts | rayjj: Ah, cool. | 15:51.17 |
henrys | tor8: I do have a concern that the release is going to come out right on the heels of you taking off for sabatical which is not great ... | 15:51.24 |
rayjj | I looked for tests_private/pdf_uploads instead of tests_private/pdf/uploads | 15:51.36 |
chrisl | kens: I suspect the Windows end works, I'm less convinced whether the device itself will still work, and whether we need it for anything | 15:51.52 |
rayjj | Robin_Watts: it has: 2016 bug695863.pdf 36 tiger.pdf in it | 15:52.01 |
kens | chrisl Well it doesn't crash..... | 15:52.07 |
Robin_Watts | rayjj: yeah, those are the files I added. | 15:52.23 |
kens | chrisl otoh it doesn't seem to write any output anywhere | 15:52.46 |
chrisl | kens: no, you'd need to add code to grab the bitmap via the DLL/API and do something with it | 15:53.12 |
kens | Seems like another candidate for the scaffold | 15:53.26 |
chrisl | I don't believe it serves a purpose, when we have the display device | 15:54.04 |
kens | No, it seems less than useful to me, even as an example | 15:54.19 |
chrisl | I'll raise it for discussion in June | 15:54.45 |
kens | Sounds good to me | 15:54.57 |
henrys | Robin_Watts: perhaps tor8 left ... what did you want to fix before the release, what is blocking? | 15:55.10 |
Robin_Watts | henrys: All my big stuff is in (the display list rewrite, the path optimisations). | 15:55.45 |
rayjj | mvrhel_laptop: well, my "fix" for the transparency+transfer_function problem showed up 1137+203 dffs (non-pdfwrite + pdfwrite) | 15:56.18 |
Robin_Watts | There are are a couple of smaller things that aren't (the ability to handle files >2Gig), which tor has said we should keep to after the release. | 15:56.19 |
| and I'm happy to go along with that. | 15:56.29 |
mvrhel_laptop | rayjj :( | 15:56.36 |
Robin_Watts | The stuff that's left to do is to run through bugzilla tidying up smaller bugs. | 15:57.01 |
| I've been doing that for the past few days. | 15:57.27 |
rayjj | mvrhel_laptop: running bmpcmp to have a look. I'm actually surprised that suppressing the xfer function has that much of an effect | 15:57.31 |
Robin_Watts | I like to think we're in pretty solid shape - if the path stuff or the display list stuff was going to cause problems, our testing would have shown it up. | 15:58.38 |
kens | rayjj lots of Quark files (mis) use the transfer functions and we have some PDF files with Tr entries also | 15:59.20 |
Robin_Watts | I don't know about the state of tor8's html stuff. It's committed, but I've not played with it myself. | 15:59.40 |
henrys | Robin_Watts: so if you think it can be a lot earlier than May I can hold off on doing the newsletter. | 15:59.40 |
chrisl | kens: so *old* versions of GSView used the gswindll device....... | 16:00.11 |
rayjj | mvrhel_laptop: interestingly there are only a few files that show a problem with pgmraw, none with ppmraw. The problems are all with pbmraw, pkmraw and cups | 16:00.17 |
kens | I believe so chrisl yes | 16:00.20 |
Robin_Watts | henrys: If we get an rc at the end of the week, we could have a release in Mid april. | 16:00.22 |
kens | chrisl but I don't believe that's a reason to maintain it any further. | 16:00.38 |
Robin_Watts | but that assumes we don't hit any problems. | 16:00.44 |
kens | If you update GS and GSView stops working then update GSView too | 16:00.48 |
henrys | Robin_Watts: okay. | 16:00.54 |
chrisl | kens: but not since 2001 (at the latest) - since then, it's used the display device | 16:01.00 |
kens | OK so defintiely no reason to keep it | 16:01.16 |
| s/no/not a/ | 16:01.25 |
chrisl | Right, noted for the next meeting | 16:02.02 |
kens | Right, wnough for today, goodnight all | 16:03.02 |
Robin_Watts | fredross-perry: In your gsview, do you call mupdf as a library, or as an exe ? | 16:16.23 |
fredross-perry | as a library. But gs is called as an exe. | 16:16.45 |
Robin_Watts | fredross-perry: Ok, then tor8's latest set of changes to the command line flags won't affect you :) | 16:17.04 |
fredross-perry | all righty | 16:17.26 |
Robin_Watts | Ah, I wanted to change the pdfclean stuff so that there was a nicer interface for mvrhel, didn't I.... | 16:17.36 |
mvrhel_laptop | Robin_Watts: yes | 16:20.02 |
| oh, you are not using the API with gs | 16:20.52 |
Robin_Watts | I have to step afk for a bit. back soon. | 16:21.33 |
fredross-perry | not using the API. | 16:21.44 |
| gs will output text when it completes pages, which I use to drive the progress bar. | 16:22.25 |
tor8 | Robin_Watts: henrys: aiming for mid-april sounds doable | 16:28.24 |
| gives us next week to bash out small bugs then do an RC | 16:28.38 |
henrys | tor8: great, that gives us some time to have it out there before you take off too. | 16:29.01 |
| I'm going to renew my driver's license, I've heard really bad things about this, rivaling the post office in inefficiency and many people in Colorado are armed. | 16:45.34 |
pedro_mac | henrys: good luck! | 16:50.13 |
mvrhel_laptop | grrrrr. the stupid windows xps writer is crashing for some reason on the windows 7 32 bit machine. | 17:07.24 |
Robin_Watts | tor8: 2 fixes on robin/master | 17:28.56 |
| the last one works on windows. I'm just going to test it on linux now. | 17:29.12 |
| Yeah, that works nicely on linux. | 17:36.31 |
| tor8: Can you test MinGW building? Or if you have done so, close bug 695854 :) | 17:38.41 |
| Earth calling Marcos. Earth calling Marcos. Come in Marcos. | 17:46.53 |
henrys | Robin_Watts: he does follow his email regularly AFAICT | 18:00.32 |
mvrhel_laptop | ok so there is something clearly wrong with the adobe pdf printer driver on my old machine. at least now I catch the error and inform the user instead of crashing | 19:05.19 |
| printing to the microsoft xps writer as well as a real printer works fine | 19:05.54 |
| so, on that note I think I am ready to try another gsview beta | 19:06.31 |
inkbottle | Hi, I'm trying to analyse a pdf document, | 19:42.37 |
| for that purpose I've extracted one page with: | 19:43.28 |
| qpdf --empty out66.pdf --pages myfile.pdf 66 -- --stream-data=uncompress | 19:43.52 |
| my question is: can I then extract an image using a simple text editor | 19:44.43 |
| and view it with an image viewer | 19:44.52 |
mvrhel_laptop | fredross-perry you there? | 20:03.16 |
fredross-perry | yessir | 20:03.23 |
mvrhel_laptop | I *think* I have gsview for windows in reasonable shape. do you feel the mac and linux versions are good enough for a beta release to the wild? | 20:03.55 |
| been beating on in on a very old 32 bit windows 7 computer | 20:04.21 |
| painful but it seems to be working | 20:04.26 |
| hard to believe I used to work on this computer | 20:04.35 |
| it is so darn slow | 20:04.42 |
| trying to get it all updated. hopefully it will be better behaved then | 20:06.10 |
| bbiaw | 20:10.13 |
| Robin_Watts: if you want to try the new gsview for windows, it is in ~mvrhel/gsview | 20:10.41 |
| or anyone else for that matter | 20:10.59 |
fredross-perry | I think so. Hoping others have tried the builds and chimed in with issues by now. Some have. | 20:14.43 |
mvrhel_laptop | ok. good deal. I have only had ken beat on the windows version. What would be nice is for everyone to set these as their default viewers for awhile to catch any issues. | 20:16.07 |
fredross-perry | mvrhel_laptop: Iâll try yours right now. | 20:17.09 |
mvrhel_laptop | ok cool. thanks. | 20:17.25 |
| I will grab you linux version | 20:17.36 |
fredross-perry | there are two, one is 64 andthe other is 32 | 20:17.54 |
mvrhel_laptop | ok. crap. I found a bug already playing around on the 32 bit system | 20:28.13 |
| hmmm | 20:28.14 |
| I will need to look at this a bit later | 20:28.23 |
| bbiaw | 20:28.24 |
| not on yours fredross-perry on mine | 20:28.39 |
fredross-perry | gotcha | 20:28.54 |
mvrhel_laptop | so everyone hold off one more day.... | 20:33.40 |
fredross-perry | iâve made one more update to the mac/linux installers, so you can open docs from the command line. | 22:43.50 |
Aldo | Hi there, is there a way to silence errors printed through gs_warn1 | 22:46.12 |
| ? | 22:46.16 |
| I'm trying to convert a pdf that falls into this case http://bugs.ghostscript.com/show_bug.cgi?id=692343 but php imagick always breaks if "gs" prints messages | 22:47.06 |
henrys | Aldo:lots of file print messages why not fix php "imagick" is that what it's called? | 22:52.34 |
Aldo | so there is no switch on GS to turn off warnings? besides dQUIET -q? | 22:55.03 |
henrys | Aldo: just those, how is gs being invoked in this php system? api or process? | 22:57.10 |
Aldo | well not sure how the php lib calls gs but I'm doing this on the command line to test | 23:04.02 |
| "gs" -q -dQUIET -dSAFER -dBATCH -dNOPAUSE -dNOPROMPT -dPARANOIDSAFER -dMaxBitmap=500000000 -dAlignToPixels=0 -dGridFitTT=2 "-sDEVICE=eps2write" "-sOutputFile=test.eps" "-ftest.pdf" | 23:04.06 |
| and that gives me this output " ./base/saes.c:156: s_aes_process(): invalid aes padding byte (0x93)" | 23:04.33 |
Robin_Watts | Aldo: So, modify the invocation to do: the same thing, but with > /dev/null 2>&1 | 23:05.04 |
henrys | Robin_Watts: he doesn't know how gs is called. | 23:05.46 |
Robin_Watts | Ok, so rename gs to be gsbin and make a shell script called gs that passes the invocation on with the redirect. | 23:06.38 |
| Or find out how it's called. | 23:06.49 |
henrys | Aldo: really wish you'd report this to the imagick folks and get this fixed. That defect is quite serious... | 23:07.03 |
Aldo | its called like that Imagick is using this | 23:07.06 |
| "gs" -q -dQUIET -dSAFER -dBATCH -dNOPAUSE -dNOPROMPT -dMaxBitmap=500000000 -dAlignToPixels=0 -dGridFitTT=2 "-sDEVICE=eps2write" "-sOutputFile=%o" "-f%i" | 23:07.08 |
| ok will find out how to report it to them | 23:07.37 |
Robin_Watts | By imagick do you mean imagemagick? | 23:07.44 |
Aldo | thanks for your help | 23:07.45 |
| correct | 23:08.12 |
| Version: ImageMagick 6.9.0-10 | 23:08.12 |
Robin_Watts | imagemagick is a set of binaries. | 23:08.47 |
| there may be php wrappers that call it. | 23:08.58 |
| I suspect that what you're seeing is php calling imagemagicks 'convert' utility, and that calls gs. | 23:09.26 |
henrys | I think he means: https://github.com/mkoppanen/imagick/ which uses imagemagick | 23:09.32 |
Aldo | exactly | 23:09.33 |
Robin_Watts | So it's the 'convert' utility that needs to be fixed. | 23:09.36 |
Aldo | php has a wrapper lib that calls convert | 23:09.47 |
Robin_Watts | henrys: Right, that would fit with what I just said. | 23:09.50 |
Aldo | which then calls this "gs" -q -dQUIET -dSAFER -dBATCH -dNOPAUSE -dNOPROMPT -dPARANOIDSAFER -dMaxBitmap=500000000 -dAlignToPixels=0 -dGridFitTT=2 "-sDEVICE=eps2write" "-sOutputFile=test.eps" "-ftest.pdf" | 23:10.00 |
henrys | Aldo can you post your file somewhat want to see if my convert works? | 23:10.08 |
| Robin_Watts: yeah but I though convert was using the api? hmph | 23:10.36 |
Aldo | sure how may I send it to you, I dont want this to be public | 23:10.38 |
Robin_Watts | henrys: No. convert calls gs direct, aiui. | 23:10.51 |
henrys | maybe I'm mistaken | 23:10.52 |
| Aldo: I don't really want anything proprietary | 23:11.11 |
Robin_Watts | We often get people complaining of imagemagick bugs and they can fairly easily get us the command line. | 23:11.16 |
henrys | Aldo: there is a later version of the imagemagick library that the github site is recommending. FWIW | 23:12.03 |
| seems to me convert could just output our warnings, at least that's what I'd expect. | 23:13.26 |
| off to dinner. I'll check the logs later. | 23:14.06 |
Robin_Watts | http://trac.imagemagick.org/browser/ImageMagick/trunk/MagickCore/delegate.c | 23:14.14 |
Aldo | right convert doesnt really break | 23:14.18 |
| its the php wrapper that when it sees output coming from convert it just stops executing | 23:14.41 |
| I'll keep digging | 23:14.51 |
| thanks for your time | 23:15.04 |
Robin_Watts | Aldo, it would be worth seeing if calling convert directly on that file works or not. | 23:15.34 |
| i.e. confirm that this is a problem with convert itself, not with the php wrapper for convert. | 23:15.51 |
| D'Oh. I should read what you write :) | 23:16.08 |
Aldo | yeah it works | 23:16.21 |
Robin_Watts | OK, so 2 possibilities... | 23:16.31 |
Aldo | it spits the warnings from gs, but still it creates the files successfully | 23:16.34 |
Robin_Watts | 1) call convert directly. | 23:16.38 |
| (or call gs directly) | 23:16.45 |
| 2) fix the imagick php code. | 23:16.54 |
mvrhel_laptop | ok. so I am running into a GenericResourceDir doesn't point to a valid resource directory on this other machine with gsview. I thought I building gs with compile inits true. | 23:18.07 |
| by default | 23:18.11 |
| and yes it is. | 23:20.00 |
| so why the issue | 23:22.41 |
Robin_Watts | mvrhel_laptop: Does that machine already have gs installed? | 23:23.04 |
mvrhel_laptop | no | 23:23.12 |
Robin_Watts | Is there an environment variable hanging around from a previous installation? | 23:23.15 |
mvrhel_laptop | it is calling into the dll that is in the same dir as the application def. due to the release number | 23:24.07 |
| this machine had not been used in a couple years and the release says 9.16 | 23:24.26 |
| and it includes the romfs | 23:24.37 |
Robin_Watts | mvrhel_laptop: Right, I'm not suggesting it's calling the wrong version. | 23:24.45 |
mvrhel_laptop | which is the first place gs is supposed to look for stuff | 23:24.46 |
| gs should first look in the romfs and if it finds the resources there then it should stop looking | 23:25.18 |
| at least that is what the documentation says ;) | 23:25.29 |
Robin_Watts | I thought, and I could easily be wrong on this, that if you built with compile_inits=1 then it would look in romfs by default. BUT if you set an environment variable (or maybe a registry value) it would look there instead. | 23:25.50 |
| hence if there was a stray env var or reg value, even from several years ago, that could screw up the behaviour of your new build. | 23:26.25 |
mvrhel_laptop | hmm let me check my env. variables | 23:26.26 |
| yes | 23:26.28 |
| there is a GS_LIB.... | 23:27.21 |
Robin_Watts | As a test, can you nobble that before calling GS? | 23:32.28 |
Robin_Watts | heads to bed. | 23:38.07 |
mvrhel_laptop | ray got me straight | 23:38.17 |
| apparently I need to include -I%romesource/Init/ | 23:38.35 |
| oops | 23:38.39 |
| I guess % R is a special thing | 23:39.07 |
Robin_Watts | %'s are special on irc, yeah :( | 23:39.19 |
mvrhel_laptop | anyway, gs needs this or it will NOT look in the romfs might might be using some env. variable like you said | 23:39.35 |
Robin_Watts | I see. You include the magic runes, and it looks there before looking in GS_LIB ? | 23:39.40 |
mvrhel_laptop | so different than what I thought was the case | 23:39.42 |
Robin_Watts | gotcha. | 23:39.45 |
mvrhel_laptop | yes | 23:39.47 |
Robin_Watts | fab. | 23:39.47 |
mvrhel_laptop | will fix this up tonight after dinner | 23:40.01 |
| have a good night Robin_Watts | 23:40.04 |
Robin_Watts | you too. If you get a new install build up, tell me where, and I'll finally get around to trying it :) | 23:40.23 |
mvrhel_laptop | awesome thanks | 23:40.36 |
| Forward 1 day (to 2015/04/01)>>> | |