IRC Logs

Log of #ghostscript at irc.freenode.net.

Search:
 <<<Back 1 day (to 2013/02/13)2013/02/14 
Sling having issues with ghostscript in chrooted environment, which doesnt happen outside of the chroot.. ive quickpkg'ed fontconfig and ghostscript-gpl and untarred them in chroot, also copied /etc/fonts into chroot, no help ;( this is the error I get when running 'convert "/file.pdf"[0] -thumbnail 200x300 "/file.png"' http://paste2.org/p/286509908:46.17 
  all paths in gs -h which exist outside the chroot, also exist inside the chroot08:46.34 
  at the point now where i'm comparing strace -f output of both convert runs, but not getting much wiser08:47.00 
kens Firts suggestion, try using a recent version of Ghostscript08:47.19 
  THIs 'looks' like your input is a PDF file, using a TrueType font called Verdana which is embedded in the document. If so, paths are not going to be an issue.08:48.08 
Sling gs versions are the same inside and outside the chroot though08:48.40 
kens THe action of switches like -dSAFER has changed08:48.54 
  And you don't give us the Ghostscript command line08:49.10 
  I don't believe your problem can be font paths, unless the font is not embedded, and I can't tell that without seeing teh PDF file.08:49.40 
Sling well this is /usr/bin/convert I'm calling, i suppose that in turn calls gs08:50.18 
kens I believe it calls ImageMagick which calls GS to do the heavy lifting yes08:50.39 
Sling http://paste2.org/p/286549208:53.05 
  the output of 'strings file.pdf | grep TrueType'08:53.14 
  so I suppose it is embedded08:53.30 
kens Unfortunaterly, that's not the thing to look for08:53.36 
  The FontDescriptor contains the FontFile object and its presence/absence determines whether the font is embedded08:54.06 
Sling http://paste2.org/p/286550208:54.20 
  pdffonts file.pdf then :)08:54.25 
kens Right, then Verdana is not embedded, so you have to go to the system.08:54.52 
  My guessis that ImageMagick is invoking GS with -dSAFER or something, and so we can't get to the font path08:55.19 
  You would need to look at the fontmap to see where Verdana is installed (indeed all the fonts bar Arial and Symbol)08:56.15 
Sling I'd say /usr/share/fonts/corefonts/verdana[zbi].ttf, since those files exist08:59.39 
  but they are also there in the chroot, meh08:59.50 
kens Well I am not a Linux expert.09:00.02 
Sling yeah, bit offtopic here09:00.10 
  I'm going to ask the customer if he can easily embed the fonts09:00.26 
  if so, thats a quicker solution for now09:00.33 
Robin_Watts move the files away in the non-chroot. Does the non-chroot then give the same results as the chroot ?09:00.35 
kens embedding fonts is always to be preferred09:00.38 
marvin_ Hi guys! Not sure if i'm talking to the right people, but i'm a little confused about the licensing of GhostScript. I'm thinking about using the executables in a commercial file conversion program, and i'm not sure if I may do that or not. Does anyone here have a clue about these things?11:20.18 
kens If you are concerned you should probably seek professional legal advice, we are not lawyers....11:20.52 
  Provided you are within the terms of the Affero GOPL licence then you do not need a licence from Artifex11:21.28 
  GPL*11:21.44 
marvin_ Yeah, thats what I'm unsure about :)11:23.43 
  just thought that this gets asked a lot, so you have a canned answer you can pull out of the drawer11:24.35 
  if thats not the case i won't bug you any further11:25.03 
Robin_Watts marvin_: The 'gold standard' way for using gs in a file converter program under the GNU GPL is (as far as Artifex is concerned, as I understand it)...11:33.37 
  1) Have gs installed as a separate installer.11:33.50 
  2) Install it into a standard place where people can come and replace it with newer versions independent of what version of your code they are using.11:34.27 
  3) call it by invoking the executables using standard flags.11:34.49 
  4) Don't modify the gs code at all.11:34.59 
  5) Make sure that the license files are all supplied with gs, and that you acknowledge in your documentation etc that you are calling out to gs which is supplied under the GNU GPL.11:35.54 
  I am not a lawyer, not can I speak authoratatively for Artifex on this matter.11:36.15 
  but I believe that if you comply with all of those, it would be very hard for anyone to have a go at you.11:36.34 
marvin_ thanks robin, that is very helpful advise!11:37.12 
Robin_Watts marvin_: no worries.11:37.19 
  tor8: bug 692639 - lots of stuff from zeniko. at least some of which probably justifies going into the release.11:42.15 
  and I think bug 693290 can be closed so we can give him the bounty?11:43.10 
tor8 Robin_Watts: okay. I'll take a look through the lots of stuff, and I think we're good on the bounty one.11:58.04 
  Robin_Watts: damn. I keep forgetting the magic git incantations I use to update jbig2dec git from ghostpdl...12:22.46 
chrisl tor8: I thought you were going to cron it?12:24.19 
tor8 chrisl: me too :(12:25.10 
  I know I had something up and running with git subtree12:25.36 
Robin_Watts tor8: I can't help with that, sorry. Even if I ever knew such runes, unless I stuck 'em in a shell script somewhere they would have fallen out of cache by now.12:29.45 
tor8 Robin_Watts: git subtree split --prefix gs/jbig2dec in ghostpdl gives a sha1 you can pull into jbig2dec.git12:31.12 
oznt hi everyone12:35.59 
  a quick question about mupdf12:36.08 
  can I create annotations with it?12:36.19 
Robin_Watts oznt: In the current release, no.12:36.39 
  In the release that's about to come out, we have started to add support.12:37.05 
oznt Robin_Watts, in the technology preview?12:37.23 
Robin_Watts but I'm not sure how far we've got - maybe only strikeout annotations so far.12:37.27 
  Right. It's not in the technology preview.12:37.43 
  1.2 is due out within a week or so.12:38.00 
  A release candidate went up yesterday.12:38.12 
oznt so release 1.2 will have initial support for it?12:38.22 
Robin_Watts Yes.12:38.29 
oznt YAY!12:38.32 
  you guys are great12:38.38 
Robin_Watts And it should be much more complete in 1.312:38.38 
  Or you can follow us on git, and get it sooner.12:38.49 
oznt i still need some time to tinker with the gui12:39.00 
Robin_Watts What platform are you doing the gui on ?12:39.12 
oznt i am trying to write a simple GTK gui around mupdf12:39.17 
  should be cross platfrom12:39.23 
  :-)12:39.25 
Robin_Watts Ah. We'd love to see it when you have something to show.12:39.34 
oznt don't wait so much :-) i am a complete noob... 12:39.53 
  i am just really frustrated about having no decent pdf reader in linux12:40.09 
  (well, there is okular, but too many kde deps ...)12:40.22 
  Robin_Watts, are you working for Artifex ?12:40.57 
Robin_Watts I do.12:41.03 
oznt i envy ... :-) looks like a very interesting place to work for12:42.24 
  well, anyway i got to go back to my java pleasures.... I just needed a break to see if my plan to use mupdf for my pdf reader is viable. it would have been dumb to start working on it, just to find out there is no annotations support.12:43.44 
tor8 Robin_Watts: chrisl: okay, now it's updated and I have a crontab script to do it daily12:43.49 
Robin_Watts oznt: If you're about to start work, then I recommend following us on git.12:48.00 
  We made a few API changes in 1.2, so it would make sense for you to start using the new API rather than the old one.12:48.20 
  And you can get annotation updates as we produce them.12:48.34 
oznt Robin_Watts, ok, i will do12:48.42 
  Robin_Watts, if i plan to produce binding to another language, the pdf functionality is only in fitz right?12:51.52 
Robin_Watts oznt: The pdf functionality is in the "pdf" module.12:53.18 
oznt what is the difference between the functionallity in the directory pdf and fitz ?12:53.23 
Robin_Watts but fitz is both our underlying renderer, and provides the document interface with which you generally drive things.12:54.01 
  In short, you'll talk to fitz, and fitz will talk to pdf.12:54.14 
  There may be times when you want to talk to pdf directly (if you want to do manipulations of low level pdf objects), but the plan is that we should offer fitz level interfaces for all that stuff.12:55.02 
tor8 fitz is both the highest and the lowest level interface, pdf is basically just a plug-in module to provide pdf parsing, same with the xps module.12:55.52 
  so unless you're doing something very PDF specific, you'll only need to talk to fitz12:56.21 
oznt ok, so the binding should be to "libfitz" e.g. if i am writing a pygtk app, I should probably used python-fitz instead of python-poppler12:59.21 
  there is a github project that made a wrapper to fitz using swing, but i don't know how up-to-date it is12:59.52 
Robin_Watts oznt: Urm... I am not sure now many libs we make.13:00.12 
  on Android, I know we build everything into libmupdf.so and bind to that.13:00.29 
tor8 Robin_Watts: with the regular makefile, be build libfitz.a and the third party libs as separate static libs13:01.02 
Robin_Watts tor8: So is there a libpdf and a libxps ? or are they in libfitz?13:03.04 
tor8 Robin_Watts: all in libfitz.a13:03.14 
  it's a bit of a shame the way the names have turned out13:03.36 
  nobody knows or cares about "fitz" anymore, and mupdf does more than just pdf...13:03.50 
oznt https://github.com/rk700/python-fitz13:04.24 
  tor8, i just build the git sources of mupdf, in debug i only see libfitz, what do you mean with "nobody cares about fitz" ?13:06.09 
tor8 oznt: ten years ago, Fitz was the project intended to replace the aging ghostscript graphics library13:06.46 
  it kind of took on a life of its own with mupdf though13:06.59 
oznt ok, but does it mean i should not use it in my project? and instead use "libmupdf" if exists ? or what? 13:07.44 
tor8 oznt: no, that's not what I mean. I'm just griping about the naming.13:08.17 
oznt the README file by the way states that the git repo is http://mupdf.com/repos/mupdf.git, which seems to be wrong ...i could not clone from there, but the repo stated in google code worked13:08.56 
  tor8, ok, i get you13:09.01 
Robin_Watts tor8: We could separate the low and high level interfaces into fitz and 'mu' if you'd prefer ;)13:20.10 
tor8 mu_page and mu_document taking fz_contexts ... urgh.13:20.41 
Robin_Watts yeah. And it's a needless naming change. I think we're stuck with fitz.13:21.06 
tor8 we are. maybe rename the header and library, but then it's confusing why all the functions start with fz_13:21.39 
  Robin_Watts: I am happy with most of zenikos patches from the bug report.13:23.30 
  one thing that's wrong (fz_translate_rect should take the offsets as floats)13:24.12 
  one thing I don't understand (the utf-8 BOM one)13:24.21 
  and then some reindentation that needs to be done, I see he's keeping the patches minimal there by not reindenting after changing the try/catch block structure13:24.48 
  I'd say this set could be worth a bounty13:25.12 
Robin_Watts tor8: fair enough. Looking at the volume of fixes that occurred to me too.14:01.57 
kens Release done,congrats chrisl14:43.42 
chrisl Ta, took an age to upload to google and sourceforge today.....14:44.15 
Robin_Watts indeed.14:44.18 
chrisl I haven't changed the mupdf download page on ghostscript.com to list the Affero license - I'll do that when we actually put an Affero licensed release out.14:50.10 
sebras tor8: how about renaming it liball.a? ;)14:55.50 
tor8 libdontyoudaremakemeshared.a14:56.13 
  chrisl: thanks for the reminder.14:59.11 
chrisl tor8: haven't you done the license change yet?14:59.54 
tor8 chrisl: nope. we warned the sumatra folks about doing it, but haven't committed yet.15:00.19 
chrisl I can't imagine how it would affect sumatra.....15:00.59 
tor8 chrisl: they'd have to change license as well15:01.14 
chrisl Oh, okay15:01.53 
henrys chrisl:curious if you've measure wireless bandwidth with your new router - using something like iperf - I am wondering if dd-wrt squeezes more performance out of the router than the installed software.15:31.29 
chrisl henrys: I never have, no. Anything I have that is bandwidth critical, I use wires.....15:32.05 
henrys I got like 20 mbits which is a lot less than promised with the router yet I measured at a fairly long distance15:34.07 
Robin_Watts henrys: Do you have many wifi networks in your area?15:34.58 
henrys I do and I'm running 2 frequency bands on my spot15:35.20 
Robin_Watts For channel 'n', it broadcasts on a range of frequencies from n-2 to n+2.15:35.28 
  hence you will get interference from any other wifi spots +/- 4 of you.15:35.51 
henrys yes I have a tool to measure for the best channel and I seems to be okay or at least as well as I can do.15:36.30 
chrisl henrys: apparently I'm getting ~26.8 Mbits/sec - I also have a lot of other wifi networks nearby.....15:38.17 
  Compared with 627 Mbits/sec wired - which is lower than I expected15:39.48 
henrys where do these number up to 300 mbs come from? Advertised with the router - how inflated do these numbers have to get before somebody says false advertising.15:41.06 
  chrisl:yea I get a 1 gig wired as expected.15:41.35 
chrisl I think the 300mbs is with proprietary extensions, so only if you have suitable hardware all from the same vender15:42.35 
Robin_Watts and you have to remember that for a company so desperate to make people shun wires, apples wifi really sucks.15:49.37 
henrys Robin_Watts: some price and pert improvements on the retina FWIW http://www.anandtech.com/show/6744/apple-cuts-pricing-on-macbook-pro-with-retina-display-and-ssd-upgrades16:02.00 
Robin_Watts yeah, that was what prompted my rant yesterday :)16:02.35 
  RAM isn't upgradable, so better buy the 16 gig one to start with :(16:03.29 
kens Looks like Apple is trying to move to the commodity market and have nothing be upgradeable16:04.00 
  ore replaceable16:04.06 
Robin_Watts kens: indeed. buy it now. throw it away and buy the newer shiny next year.16:04.42 
tor8 Robin_Watts: shun wires? you must mean desperate to make people crave their drm-chipped special usb-but-not-really cables...16:04.52 
kens Yep, more lucre for Apple16:04.52 
henrys i'd love to get that 15 upgraded but it's really too much expensive16:05.27 
Robin_Watts tor8: I mean, if you want ethernet on your shiny new $3599 apple... that'll be a $20 adapter.16:05.40 
  IF you want DVI on your apple... that'll be a $30 cable.16:06.01 
  etc.16:06.10 
kens Standards -16:06.24 
  Microsoft "embrace and extend"16:06.32 
  Apple "embrace and charge for as an extra"16:06.44 
henrys and optical drive but I guess that is getting common16:07.10 
  s/and/and no16:07.18 
kens Yeah, no slimline laptop has optical media any more16:07.25 
tor8 nobody I know uses optical media anymore, so good riddance there16:07.39 
kens I do16:07.53 
tor8 whatever for?16:08.00 
Robin_Watts I am reminded of Top Gear reviewing a BMW. "When you buy the car, they give you all the air in it for free. Everything else is an option."16:08.04 
kens backup, long term storage, playing DvDs on my TV16:08.14 
henrys I don't know my kids buy text books and they all have CD's.16:08.14 
Robin_Watts tor8: what kens said.16:08.29 
henrys students aren't customers for laptops?16:08.36 
kens tablets :-)16:08.46 
tor8 kens: external hard drives, usb thumb drives (for installing/upgrading o.s.), and the pirate "cloud" are all solutions to that :)16:09.09 
kens Yes, but I already have a lot of stuff on optical media16:09.32 
henrys I guess they don't read the books anyway16:09.38 
tor8 kens: I've migrated all my optical media to hard drives16:10.03 
Robin_Watts Buying downloaded music/films is a mugs game.16:10.06 
  tor8: I don't have enough harddrives for that.16:10.22 
kens tor8 yes, I do that too, but I also have the optical as a backup for when the drive fails16:10.27 
tor8 buying games that come on discs are the only reason I still have an optical drive. but only to make disk images that I backup to hard drives.16:10.36 
kens Also making CDs of images to give to people (giving away USB sticks is still too expensive)16:11.15 
  and Videos too16:11.23 
  (eg our daughter's polo party)16:11.33 
tor8 kens: hmm. okay. so I do know someone who uses dvd:s after all :)16:11.49 
kens I'm not saying its a major use, but I wouldn't like to give it up yet16:12.14 
  OK time to go night all17:16.20 
chrisl Off to squash - will check mail and IRC logs when I get back......17:41.44 
ray_laptop Robin_Watts and kens: (for the logs): I saw the comments to marvin_ earrlier. We should mention to people using Ghostscript that want to conform to the GPL that the BEST way to conform is to release their product under the GPL as well. We want to encourage people to develop open source apps !18:27.01 
Robin_Watts ray_laptop: Well, yes, but he'd said that he was producing a commercial app.18:27.45 
  Given that he'd said he wanted to supply/use gs with a commercial app, I described what (in my view) would be the gold standard way of doing that.18:28.36 
  which I believe conforms with Miles' interpretation of the GPL.18:28.58 
ray_laptop Robin_Watts: GPL can be sold -- as long as you offer to provide source18:31.06 
  that's how Redhat and Suse (are they still around), etc. sell their linux18:31.41 
Robin_Watts Yes, you make money off the "convenience" factor.18:32.17 
  but frankly that's unrealistic for most things, I believe.18:33.02 
ray_laptop miles' (Artifex's) interpretation of the GPL is only that gs used invisibly by a non-GPL app is not conforming since the viral nature of GPL is violated.18:33.04 
  Robin_Watts: in fact, you can charge people for GPL ghostscript (as long as you identify it as GPL and make source available)18:33.54 
Robin_Watts right. And if he followed my steps, the use of gs would be clear. (Noted in the docs, and as a separate install step)18:34.09 
  yes, I am aware of this.18:34.16 
ray_laptop I thought about doing that, but I think it falls under my employment terms..18:34.33 
Robin_Watts If you're tempted by the thought of paying for GPL ghostscript, then I have some fresh air I'd like to sell you...18:35.00 
ray_laptop Robin_Watts: even a separate install step isn't OK. For instance, if someone installs gsdll32.dll and lnks to it, then clearly their app is virally infected. Miles has simply expanded the definition of 'linking' to include invocation via a process18:36.36 
  Robin_Watts: people _do_ market cans of air !18:37.01 
Robin_Watts ray_laptop: Right. I also said "invoke it using a separate process with standard flags"18:37.27 
  ray_laptop: Yes, but those are pressurised, with chemicals added.18:37.49 
  I believe that if he followed all my steps, even Miles couldn't complain.18:38.15 
ray_laptop Robin_Watts: Miles _could_ complain. But ... 18:39.00 
Robin_Watts http://www.teslamotors.com/blog/most-peculiar-test-drive18:39.23 
  It was on here that the NYT article was discussed the other day, right?18:39.39 
henrys the community is split on linking with a shared library that is gpl - it is unclear how a court case would come out.18:56.09 
  static linking is definitely not allowed18:56.46 
Robin_Watts henrys: Right, but at no point did I suggest shared or static linking.18:57.27 
henrys I was responding to ray_laptop's statement18:58.42 
Robin_Watts right, sorry.18:58.58 
henrys I should have written his name18:59.17 
ray_laptop henrys: I don't know if linking to a shared library has ever been tested.18:59.52 
Robin_Watts tested in court?19:00.15 
ray_laptop Robin_Watts: tested in court, right (sorry)19:00.38 
Robin_Watts Yeah, that's certainly less clear. I'm surprised the GNU GPL doesn't explicitly mention that, actually.19:01.30 
  Or have variants that specifically allow/preclude it.19:01.50 
ray_laptop too bad there isn't a way to test (in our gswin*.exe) if it is being invoked by an app "system" call vs. by Windows or a shell.19:02.27 
henrys arg[0]?19:02.56 
ray_laptop I suppose we _could_ pop up a window (that timed out and went away) that would irritate apps that want to invoke us silently.19:03.38 
Robin_Watts henrys: argv[0] tells you what name you were invoked under.19:03.55 
  It doesn't tell you *who* invoked you.19:04.05 
ray_laptop "GPL Ghostscript, developed and maintained by Artifex Software Inc."19:04.28 
henrys I am assuming a library user won't name their app gswin3219:04.29 
  or whatever19:04.39 
Robin_Watts henrys: I think ray was suggesting that he wanted our gswin32c to behave differently if it was being called from a shell or from another process.19:05.22 
ray_laptop henrys: why not -- most "cheaters" just use our .exe unmodified (otherwise they would have to provide source under GPL for their .exe)19:05.23 
Robin_Watts but fundamentally shells are just processes.19:05.33 
  I have: cmd.exe, bash.exe, sh.exe on this windows machine.19:06.00 
ray_laptop Robin_Watts: right -- shells are just a certain type of app themselves19:06.03 
Robin_Watts I'm sure powershell is another one.19:06.09 
  and anyone could write a new one.19:06.18 
ray_laptop that spawn processes19:06.23 
henrys oh I didn't understand sorry19:06.35 
Robin_Watts The FSF apparently takes the position that dynamic linkage == derivative work.19:09.06 
  And Miles takes the view that silent invocation == dynamic linkage.19:11.07 
henrys that's quite a leap19:11.31 
ray_laptop if the default gswin32c required a a <cr> on stdin in order to exit, it would effectively hang the app since most don't invoke with a way to provide stdin. And since most users never invoke gswin32 directly, it won't irritate them much. Of course, our customer version could have that disabled.19:11.52 
  apps that wanted to avoid it would have to provide their own modified source, or call the DLL19:13.10 
Robin_Watts ray_laptop: Eh? I run gswin32c.exe from scripts all the time.19:13.40 
ray_laptop Robin_Watts: so, you could run the customer one19:13.58 
henrys I don't think we should fool with people invoking gs as a process. There are probably plenty of real GPL cheaters that we can chase after19:13.59 
Robin_Watts I'd hate to have to hit return after each one.19:14.00 
ray_laptop have to go. have a haircut appt. ...19:14.22 
Robin_Watts ray_laptop: If you do that, you break EVERYONE that uses gs.19:14.26 
  and someone would just take our releases, remove that and release it as GNU Ghostscript again.19:14.44 
malc tor8: hi, what's this reflow thing commit messages keep telling about?19:16.47 
Robin_Watts malc: The android app has a reflow mode.19:17.10 
  You load a pdf document, and then hit the 'reflow mode' button.19:17.22 
  That feeds each page into our text extraction code, and throws out HTML. That gets fed into a WebView.19:17.52 
  If you zoom in and out on the webview, the width of the page stays fixed, but the fonts scale.19:18.17 
  hence you see the text from your PDF in a reflowable form.19:18.26 
  Does that make sense ?19:18.41 
henrys I do wonder how Affero affects gnu ghostscript - seems like they shouldn't be able to take our code and remove the clause.19:22.03 
Robin_Watts henrys: Indeed not.19:22.25 
  We're releasing our code under a more restrictive license. They don't have the right to use any of our changes since the last vanilla GNU GPL release except under Aferro.19:23.24 
  Afferro.19:23.29 
  Affero.19:23.33 
  bah.19:23.35 
henrys AGPL19:23.46 
  much easier19:24.03 
Robin_Watts yeah :)19:24.08 
malc Robin_Watts: it does, but.. geee...19:24.23 
  thanks19:24.26 
Robin_Watts geee ?19:24.35 
malc complicated19:24.42 
Robin_Watts It's not complicated for the user (they hit the button, they get reflowable text)19:25.02 
malc Robin_Watts: couldn't care less about the user :)19:25.15 
Robin_Watts It's complicated to code, but arguably feeding to HTML is less work that doing our own scalable view.19:25.37 
malc Robin_Watts: btw. perhaps you have a link to some pdf document with embedded launch actions?19:25.41 
Robin_Watts malc: I think calc.pdf might have one.19:25.56 
malc Robin_Watts: nope19:26.46 
  it has remote ones, not launch19:27.02 
Robin_Watts Ah, right. No, sorry, not offhand.19:27.14 
malc thanks nonetheless19:28.11 
henrys so I can write the gnu people now or let them screw up first. The latter seems more appealing ;-)20:05.18 
Robin_Watts Is the GNU Ghostscript thing still going?20:06.16 
  hmm, yes. 9.06 appeared on 1/1/2013.20:08.40 
henrys I'm having a look at the photoshop 1.0 source code in pascal! http://computerhistory.org/atchm/adobe-photoshop-source-code/22:15.25 
 Forward 1 day (to 2013/02/15)>>> 
ghostscript.com
Search: