| <<<Back 1 day (to 2013/05/22) | 2013/05/23 |
michaells | hi | 02:13.05 |
ray_laptop | test | 05:03.54 |
sebras | tor8: http://lwn.net/Articles/551477/rss | 06:52.53 |
| i guess this means you will move mupdf elsewhere..? | 06:53.26 |
| morning kens! | 06:53.38 |
kens | Good morning | 06:53.44 |
chrisl | sebras: and we were planning to keep the ghostscript downloads available on google code! That was a bit of a surprise | 07:00.20 |
| Parenthetically, I did like the quote: "Ask a programmer to review 10 lines of code, he'll find 10 issues. Ask him to do 500 lines and he'll say it looks good." :-) | 07:01.51 |
sebras | chrisl: yes, isn't it annoying. :p | 07:07.34 |
| chrisl: and re: quote, but it is so true! this is why I always try to make small patches, because then people will _actually_ review them. | 07:08.07 |
chrisl | sebras: TBH, it's why I prefer big patches! I get bored of endlessly debating the trivial | 07:16.02 |
sebras | chrisl: we might be interpreting the quote differently. I feel that devs can't be bothered to actually look at big patches, so I might as well don't do a code review at all. | 07:17.33 |
| chrisl: with smaller patches the actually look at the changes because it seems less imposing. | 07:17.52 |
| chrisl: but yeah, I'm really annoyed by pointless formatting changes. it is rare that anyone reviews the actual design of the monstrosities that I build. ;) | 07:18.39 |
| (at work, I should probably point out) | 07:19.00 |
chrisl | sebras: yes, I get that. But so many times the the review process seems to descend into long debates about formatting, or naming, rather than the actual function of the code - or having to explain stuff that's covered in the spec because the reviewer(s) don't know it, and can't be bothered to read it..... that stuff frustrates me! | 07:20.58 |
sebras | chrisl: naming matters in interfaces a whole lot, but less so in internal code imho. at work one has to explain relevant RFCs to reviewers of ones code quite a bit sometimes. not to mention to project managers to explain why we need to do something. ;) | 07:27.52 |
chrisl | sebras: naming is certainly important, but I think too often it becomes the focus of the review process, over whether the code does what it's supposed to. I also don't think code reviews are the right venue for explaining specifications, *except* if there is something in the code that's not obviously connected to the spec. | 07:30.53 |
| I'm not arguing that reviews aren't useful, I'm saying that they seem (to me) to often focus over much on the wrong aspects. | 07:31.38 |
sebras | chrisl: yes, I sincerely agree with you. :) | 08:19.05 |
chrisl | sebras: In a (totally unconnected) book I read recently, there was a nice phrase to sum it up: "prolonged discussion of the inessentials" | 08:21.54 |
| kens: have you read any of Charles Stross's "Merchant Princes" series? | 08:46.25 |
kens | Ye, all of them | 08:46.34 |
chrisl | Worthwhile, then? | 08:46.47 |
kens | Hmm.... | 08:46.53 |
| Not his best work IMO | 08:46.59 |
| Not bad, but I found them ultimately unsatisfying | 08:47.19 |
| Would you like to borrow one ? | 08:47.29 |
chrisl | Sure, that would be good, thanks. Given your review, probably not worth adding to an amazon order, right now | 08:48.13 |
kens | I would try them first and see what you think. I take it you haven't read any ? SO the first one would be the place to start | 08:48.41 |
| The idea is good, the writing is of his usual standard, but somehow I found them hard going. | 08:49.11 |
chrisl | Yes, first one, please. I haven't read any - they came up on an Amazon recommended list | 08:49.36 |
kens | OK I'll bring the first one along for the next trip, its a regular format paperback, so no problem | 08:50.01 |
chrisl | Thanks | 08:50.31 |
kens | Hmm, I wonder why my new device provokes the PDF interpreter into an error.... | 08:51.22 |
| ah, undefinedresult | 08:51.48 |
| That's got to be my hacky text handling 'work around' | 08:52.10 |
| More coffee required | 08:52.27 |
tor8 | Robin_Watts: I hated the reboot, "red matter" is even worse than the old science make believe, and oh dear god the plot holes! | 09:45.52 |
| but then I shan't spoil "into darkness" for you | 09:46.10 |
| chrisl, kens: I haven't read them myself, but stross has re-released new editions of the merchant princes recently (http://www.antipope.org/charlie/blog-static/2013/05/a-message-from-our-uk-sponsors-2.html) | 10:01.07 |
kens | tor8 yes I knew but since I wasn't completely sold on teh series, I wasn't hugely interested.... | 10:03.21 |
chrisl | tor8: that may be why amazon decided to recommend it to me | 10:28.59 |
paulgardiner | tor8: what was the plan with openssl? Are we going to handle it in the same way as v8? | 10:38.00 |
tor8 | paulgardiner: for now I think we should do something similar. | 10:38.51 |
| add some makefile rules to invoke configure && make on it if it's checked out in thirdparty, else use the system library, else use no-op implementations that always fail | 10:39.25 |
| how do you feel about putting the code in fitz/crypt_pkcs7_openssl.c and have that #ifdef HAVE_OPENSSL guarded with some null implementations in the #else ? | 10:40.22 |
| or just crypt_pkcs7.c | 10:40.32 |
paulgardiner | Yes. Makes sense. With v8, I believe we have a prebuilt version that people can download, and which is also transfered to the cluster machines during testing. | 10:41.58 |
tor8 | paulgardiner: yeah, but v8 is even bigger and more difficult to build than openssl :) | 10:42.20 |
paulgardiner | We may run people into problems if we don't pick a particular version of openssl. | 10:43.19 |
Robin_Watts | tor8, paulgardiner: I am attempting to build armeabi and x86 versions of v8 now. | 10:43.58 |
paulgardiner | Robin_Watts: excellent | 10:44.32 |
tor8 | paulgardiner: are there version incompatibilities? I was under the impression the openssl project was very conservative about change. | 10:54.03 |
paulgardiner | Maybe not. Just wary there might be. The only one I've noticed is that they used to not check "purpose" on self-signed signatures, but now do. | 11:07.47 |
| Also, we may be using APIs they consider to be internal | 11:08.33 |
Yoshi47 | anyone know if its possible to tell mupdf to keep using one window, so anytime a new pdf is opened it uses an existing window if opened already? | 12:28.32 |
Robin_Watts | Yoshi47: It's not possible, currently. | 12:58.17 |
noxy | hey there, i'm really confused about GS licensing, in short, i have a windows app. (i'm creating a virtual printer that is using "Ghostscript PDF" driver distributed with ghostscript, while installing.) which is closed-source commercial, this app is basically using virtual printer for converting files to pdf., so am i abusing GS licence ? | 13:16.18 |
kens | You won't find any lawyers here I'm afraid. | 13:16.55 |
| You are certainly abusing the spirit if not the terms of the GPL, and we won't like you for doing it. | 13:17.30 |
Robin_Watts | noxy: There are some rules of thumb we can offer, but don't take this as gospel legal advice. | 13:17.42 |
| 1) If you call into gs using the DLL api, then you need to GPL your code. | 13:18.12 |
noxy | Robin_Watts: first thing, can i create virtual printer which is using Ghostscript PDF" distributed with ghostscript in ghostpdf.inf/ghostpdf.ppd without abusing GPL ? | 13:19.28 |
Robin_Watts | noxy: I am not personally familiar with using ppd's. | 13:20.02 |
kens | noxy; we will not give you a definitive legally binding answer. 'It depends' is the best you can expect, ecause you ahevn't said exactly *how* you are using GS | 13:20.07 |
Robin_Watts | noxy: When you install your app, if you install gs as part of that installation silently, then that looks bad for you. | 13:20.45 |
kens | If all you are doing is using our PPD then you should just go and create your own, its trivial to do. | 13:20.53 |
Robin_Watts | better to have a separate installer for gs and the ppd. | 13:21.04 |
sebras | chrisl: "Change all if (ptr) and assert(ptr); to if (ptr != NULL and assert (ptr != NULL) and change if (...)\n\t x; to if (...) {\nx;\n}". review comments like these are fun. sorry for the escaped whitespace, I'm not sure irssi can do multiline messages. | 13:21.31 |
kens | PPD = PostScript Printer Definition, its a relatively simple text file. | 13:21.36 |
Robin_Watts | That way users can uninstall gs or upgrade it without touching the rest of your app. | 13:21.58 |
| Or they could even replace gs with a different virtual printer that behaved the same. | 13:22.18 |
noxy | Robin_Watts: okidoki i got it | 13:23.04 |
| i'm gonna separate virtual printer setup | 13:23.19 |
| but still can i distribute that setup with my setup file ? | 13:23.39 |
Robin_Watts | noxy: Including GPL stuff on distribution media with non GPL stuff is fine, as long as you follow the terms of the GPL. | 13:24.38 |
| You should read the GPL for exact details, but from memory you need to provide pointers to where people can get the source that was used to build that GPL stuff, and include the license etc. | 13:25.12 |
| The problems with the GPL come when you 'link' GPL and non-GPL stuff together. And the possible interpretations of 'linking'. | 13:26.23 |
noxy | oke second step, i'm calling "gswin32c.exe" as a process in my application and i'm using its output did i abuse GPL ? | 13:26.25 |
kens | abuse, or violate ? :-) | 13:26.49 |
noxy | still my setup contains Ghostscript :S | 13:27.01 |
| i mean violate ^^ | 13:27.19 |
kens | YOu are permitted to call the open-source application. FOr details you need to read the GPL in detail | 13:27.21 |
| As far as I know (and I am not a lawyer) you are permitted to call Ghostscript from your application, even if it is closed-source/commcercial | 13:28.00 |
Robin_Watts | noxy: AIUI, If you're calling out to a system installed gswin32c.exe, then we can't complain. | 13:28.06 |
| If you hide that exe within your app, it makes it harder for people to replace/upgrade it, and that gets you into shadier territory. | 13:28.47 |
noxy | oke my plan is that let me explain ; | 13:29.09 |
tor8 | sebras: now you know why gofmt exists ... to nip those kinds of review comments in the bud! | 13:30.03 |
sebras | tor8: of course I know. we discussed having a tool enforce this at git commit-time, but it was deemed too intrusive because of ridiculously unformatted code already existing. | 13:31.47 |
noxy | I have MySetup which is contains GSSetup, during installation MySetup prompt the user for installing GSSetup then GSSetup is starting to installation it shows the user about GS user can install it or cancel it but in this step if user cancel it can't use my app :) anyway, after complete GSSetup installation MySetup resume installation | 13:32.41 |
kens | noxy, this is reasonable, yes. You should probably also check for an existing installation of Ghostscript and use it if present. | 13:33.17 |
noxy | kens: last question | 13:33.36 |
kens | Or at least offer this as an optin. People will be understandably upset if you downgrade their installation. | 13:33.43 |
noxy | my application 100% depends to GS is it problem ? | 13:34.10 |
kens | noxy, depends what you mean by 'is it a problem'. If you plan to charge for something which is 90+% based on our application, which you distribute for free, you can't expect us to be totally happy about it. Is it legel ? Probably, there are several other otufits doing this already. However, that is not an answer (form me) that you can rely on in court. | 13:35.24 |
sebras | kens: noxy: or even upgrading their version unless you let them know. | 13:35.25 |
kens | My legal opinion is wortth what you pay for it :-) | 13:36.00 |
Robin_Watts | My legal opinion is worth less :) | 13:36.16 |
kens | and sebras is right, you should also not upgrade their installation without prompting | 13:36.33 |
sebras | noxy: just because I'm curious. is it an option for you to make your application GPL? that would also solve your problems of course. | 13:39.59 |
noxy | sebras: i didn't get what you mean ? | 13:40.07 |
sebras | noxy: I mean that if the computer already has a ghostscript installation, and a user installs your application and it installs a newer ghostscript installation then the user might not be happy because they wanted to keep the older version. | 13:41.17 |
| ...of ghostscript. | 13:41.27 |
noxy | sebras: oh got it | 13:41.45 |
sebras | noxy: and the other question was whether it is an option for you to make your application open source..? | 13:42.07 |
noxy | sebras: it's not an option, actually GS is small part of my application, we're working on IWB (Interactive Whiteboard Software) i'll use GS only for importing files to my application such as (powerpoint, word, excel, etc ) and my company doesn't wanna make that application Opoen source only for this feature beacuse there is tons of new feature for IWB market | 13:45.35 |
kens | noxy an alternative, of course, is to buy a commercial licence. | 13:46.02 |
| Artifex licences Ghostscript commerciaslly, which absolves you form any GPL problems. | 13:46.31 |
Robin_Watts | noxy: How is gs going to import powerpoint/word/excel ? | 13:47.15 |
kens | Robin_Watts : form what he syas, its going to use the app to 'print' to PDF, then handle the PDF. | 13:47.39 |
noxy | kens: as i see there is runtime licensing model, our application gonna work on 500k~600k computer and we gonna sell it really uber cheap , probably we cannot pay for 500k runtime licence :S | 13:48.20 |
Robin_Watts | noxy: Our licensing plans are tailored. | 13:49.11 |
noxy | Robin_Watts: as kens says, i'm gonna convert anby printable file to pdf, then rendering that pdf to our application | 13:49.14 |
kens | noxy, well I'm not into the commercial end, but.... It gets you out of GPL problems, and it gets you support, which you don't get otherwise. | 13:49.19 |
| I woudl suggest at least discussing it with our sales VP | 13:49.38 |
Robin_Watts | If you sell a piece of software for $x and we provide y% of the functionality, then clearly we can't ask for more than $x * y/100 :) | 13:50.02 |
| It's definitely worth a conversation with Scott. | 13:50.37 |
| Woo Hoo! MuPDF+V8 running in the emulator. | 13:51.05 |
sebras | Robin_Watts: didn't his mail address use to be on your website? I have been browsing, strugglign to find it..?! | 13:51.07 |
Robin_Watts | sebras: sales at artifex.com | 13:51.21 |
kens | scott.sackett@artifex.com | 13:51.25 |
| sales@artifex.com works too :-) | 13:51.35 |
tor8 | paulgardiner: how did you build openssl to test signatures? for windows? | 13:52.05 |
sebras | kens: ah. I sincerely wonder why that address is not availabled on http://artifex.com/indexlicense.htm | 13:52.08 |
noxy | kens: what kinds of out of GPL problem , i don't see any technical difficulties ? | 13:52.45 |
kens | well info@ gets there too | 13:52.45 |
paulgardiner | tor8: I just followed the instructions in INSTALL.W32 | 13:53.07 |
kens | noxy you are obviously concerned about GPL implications, if you have a commercial licence, there are no GPL implications. | 13:53.07 |
tor8 | paulgardiner: you have perl on your windows box? | 13:53.19 |
Robin_Watts | noxy: Yeah, with a commercial license, you can do what you want. | 13:53.34 |
| noxy: In particular you can make installation much simpler (no need for 2 different installers) | 13:53.49 |
paulgardiner | tor8: Yes. I've required it in the past for other projects | 13:54.26 |
Robin_Watts | noxy: And you can avoid the possibility of users breaking your app by uninstalling gs or installing new versions. | 13:54.27 |
sebras | noxy: send a mail to salex@artifex.com, describe your situation and ask what commerical license options you can get. it doesn't hurt to ask. :) | 13:55.06 |
noxy | Robin_Watts: tbh, i'm just a developer, (only developer it's hard to work in startup :S ) i'll tell about commercial licensing to my boss after this step he can give his decision, but thanks for your help | 13:56.24 |
| at least, it's more clear | 13:56.58 |
sebras | tor8: btw, I knew why gofmt existed before. where did you get the impression that I didn't..? | 14:00.08 |
tor8 | sebras: just sayin' you should be using go instead of gobject ;) | 14:02.26 |
sebras | tor8: vala has been contemplated. | 14:04.09 |
| tor8: not sure if there are tools for that. | 14:04.20 |
chrisl | Hmmm, cluster dead?? | 14:06.33 |
Robin_Watts | looks like network outages to me. | 14:09.09 |
chrisl | When I do a push, nothing happens | 14:09.33 |
Robin_Watts | looks like the clustermaster.pl script has died. | 14:10.30 |
chrisl | Ah, possibly due to network outages? | 14:10.55 |
| Actually, I thought clustermaster.pl was triggered by cron | 14:11.25 |
Robin_Watts | chrisl: It should be running pretty much all the time. | 14:12.45 |
| It's retriggered by cron every 15 mins, when it checks to see if it's crashed or not. | 14:13.05 |
| so it should restart automatically. | 14:13.08 |
chrisl | Yeh, I've tried a few times in the last hour or so | 14:13.32 |
Yoshi47 | Robin_Watts, thanks for the reply, any idea if its in the plans for a feature? | 14:13.52 |
Robin_Watts | Yoshi47: It's not planned. | 14:14.07 |
Yoshi47 | Robin_Watts, would it be alright if we did plan it? or is that not in the scope of MuPDF? | 14:14.39 |
Robin_Watts | Yoshi47: We'll consider patches, sure, but we aren't likely to put much effort into the existing mupdf viewers on windows or linux. | 14:15.21 |
| we are working on a new GTk based one for linux. | 14:15.39 |
Yoshi47 | Robin_Watts, so its not planned for that one either i guess. | 14:16.33 |
Robin_Watts | no. Open an enhancement bug on bugs.ghostscript.com and make your case. | 14:16.56 |
| chrisl: I can't see why it's not working. | 14:18.21 |
chrisl | Robin_Watts: me neither - hopefully marcosw will kick it at some point | 14:19.02 |
Robin_Watts | chrisl: Try to push something now. | 14:22.55 |
chrisl | trying now | 14:23.38 |
| Nothing coming up on the dashboard :-( | 14:24.18 |
Robin_Watts | Nothing seems to make it as far as the queue. | 14:24.45 |
| chrisl: I've texted marcosw. | 14:29.32 |
sebras | Yoshi47: if you are on linux there is one option to trigger the x11-based viewer to reload the current pdf. would that solve your problem. | 14:29.39 |
Yoshi47 | for those who are on linux yes, but those people on Windows are still out of luck | 14:30.13 |
| for interest sake, what is that option? | 14:30.27 |
sebras | Yoshi47: there are two ways. either you reload the pdf by pressing 'r' (ought to work in windows as well). | 14:32.31 |
| Yoshi47: or, in linux, you can send SIGHUP to the process. the idea being that you can have this in a makefile that generates a pdf and having rebuilding trigger reloading. | 14:33.16 |
Yoshi47 | its a different pdf file | 14:33.37 |
sebras | Yoshi47: then you're out of luck currently. :) | 14:34.15 |
Yoshi47 | yep, thanks, Robin_Watts already made that known | 14:34.33 |
kens | You can always modify the source | 14:34.53 |
Yoshi47 | Robin_Watts, http://bugs.ghostscript.com/show_bug.cgi?id=694098 | 14:40.01 |
| i suck at making bugs | 14:40.04 |
| Robin_Watts, is there a spec for the new GTK viewer? | 14:44.38 |
| I would love to collaborate on it if im allowed | 14:44.52 |
sebras | Yoshi47: no, not really. | 14:44.52 |
Robin_Watts | Yoshi47: I don't think so, unless it's in tor8's head :) | 14:45.04 |
Yoshi47 | nice | 14:45.09 |
| just use ghostscript to print it out, should work. | 14:45.24 |
| is it in GTK 3? | 14:46.38 |
henrys | kens:I stepped through pdfwrite text processing yesterday to get some ideas about how to do it in xps ⦠at each turn I thought: my goodness it can't be this involved. | 14:46.52 |
Robin_Watts | henrys: It's a bitch :( | 14:49.33 |
| Morning marcosw_ | 15:09.05 |
marcosw_ | morning. | 15:09.13 |
Robin_Watts | What was up with the cluster? | 15:13.08 |
marcosw_ | I broke the code that builds the commands last night. I violated the rule to always at least make sure that programs run before going to bed. | 15:15.39 |
| I'll go through the logs and put the skipped jobs into the queue | 15:16.02 |
Yoshi47 | marcosw_, you should get a beating increment | 15:16.42 |
chrisl | marcosw_: I don't think we've missed any commits, and my pushes can be zapped (if they're still there): I did those tests locally | 15:17.10 |
Robin_Watts | Ypshit47: marcosw_.chastisement++; | 15:17.35 |
| ahem, apologies for that typo :( | 15:18.02 |
Yoshi47 | np | 15:18.06 |
marcosw_ | chrisl: it does appear that it was your jobs that were skipped | 15:19.13 |
| Robin_Watts: and one of robins | 15:19.27 |
chrisl | marcosw_: yeh, like I say, mine can be binned | 15:19.43 |
Robin_Watts | marcosw_: My user push was just a test to see if the cluster was working. | 15:19.58 |
marcosw_ | Robin_Watts: okay, so I'll ignore that job as well. | 15:20.22 |
Yoshi47 | what cluster is this? a build cluster? | 15:22.13 |
Robin_Watts | Yoshi47: A test cluster. See http://ghostscript.com/regression/ | 15:22.41 |
Yoshi47 | huh, cool | 15:23.05 |
kens | OK I've had enough goodnight all | 16:27.32 |
sebras | and I just can't get enough, I just can't get enough... | 16:46.46 |
nordmike | Hello. I have a problem with pdf. I can see text. But when I copy and paste it to other text field I get this "ââââ¬â¡ââ£". So I can't use search or copy anything. | 17:30.59 |
Robin_Watts | nordmike: Well, that's probably down to the PDF you are using. | 17:32.35 |
nordmike | Robin_Watts: Yes. I've seen such method to fix encoding: | 17:35.31 |
| gs -sDEVICE=pdfwrite -dNOPAUSE -dBATCH -dSAFER -sOutputFile=output.pdf input.pdf | 17:35.33 |
| But this didn't work. | 17:35.44 |
vtorri | hey | 17:36.13 |
ghostbot | bonjour | 17:36.13 |
vtorri | in an outline, are all kind of link available ? | 17:36.31 |
| or only GOTO is possible ? | 17:36.54 |
nordmike | Maybe ghostscript can fix encoding by other way? | 17:38.34 |
Robin_Watts | nordmike: The information may simply not be there. | 17:49.00 |
nordmike | Robin_Watts: which information? | 17:49.32 |
Robin_Watts | The text can be encoded so that it uses a glyph ordering that is completely unrelated to anything else. | 17:49.35 |
| So the PDF might say "show glyphs 0,1,2,3,4 in font X" | 17:49.55 |
| and those might happen to spell out "HELLO" | 17:50.12 |
| but there is not necessarily any information in the PDF to tell us that 0 is H, etc. | 17:50.36 |
| Can you cut/paste the text from acrobat successfully? | 17:50.50 |
| vtorri: All link types are possible, AIUI. | 17:51.14 |
nordmike | Robin_Watts: âââââââââ â ââââ¡â¡ ââââ¬ââ¢â ââ ââ¢ââ«â â⣠âââââ ⢠âââ â ââ¡â¡âââ âââ ââ ââ | 17:52.26 |
| Robin_Watts: These should be cyrilic symbols. | 17:52.54 |
Robin_Watts | right, so the answer is no :) | 17:53.10 |
| The information is simply not there. Nothing you can do can restore it. | 17:53.23 |
nordmike | Robin_Watts: Maybe I can make codepage manualy? | 17:54.44 |
sebras | nordmike: if Robin_Watts is correct setting the codepage probably doesn't help. it could be the case that the font used in your PDF only contains a subset of the characters in the original font, and the PDF creator might have squashed the character numbers together so that they can not be mapped to letters, except visually. | 18:20.55 |
tor8 | nordmike: rendering to an image and using an OCR package would be your best bet... | 18:21.25 |
sebras | nordmike: if you manage to copy the text correctly in another PDF viewer but it does not work correctly in mupdf e.g. then you have a PDF that we might find interesting. | 18:21.44 |
| tor8: seems like PDF users rarely fance using OCR to make their PDFs searchable... | 18:33.02 |
| s/fance/fancy/ | 18:33.13 |
vtorri | hey | 21:50.12 |
| can someone try to open the following file with mupdf, please : | 21:50.30 |
| http://www.ratp.fr/informer/picts/plans/pdf/reseaux/metro.pdf | 21:50.48 |
| Forward 1 day (to 2013/05/24)>>> | |