IRC Logs

Log of #ghostscript at irc.freenode.net.

Search:
 <<<Back 1 day (to 2013/09/22)2013/09/23 
mvrhel_laptop henrys: is the regression dashboard working for you?03:59.47 
  nevermind04:00.06 
  there it goes04:00.08 
sebras2 good morning!05:08.49 
  kens: it's much easier to be up before you now that I'm in a timezone sufficiently east of you. :)05:09.17 
  kens: ^^06:47.29 
kens Yes ?06:47.37 
sebras2 kens: there is an irclog for you.06:47.50 
kens OK I'll get to it in a bit, still doing email06:48.00 
  So where are ou this morning Sebras ?06:58.08 
sebras2 kens: in shanghai! I'm not sure I can get to much more east of you. :)07:00.10 
kens I guess there's not far to go form there, no07:00.47 
  Maybe New Zealand ?07:01.26 
  or is it CHristmas Island...07:01.34 
  Apparently Tokyo is +907:02.43 
  Tahiti is +1007:03.11 
  Oh Midway is +1107:03.43 
  Ah and I was rioght kiritimati (Christmas island0 is UTC+1407:04.49 
kens is unsure how you cna have UTC+14 and UTC-12 at the same time.....07:05.28 
sebras2 well, I guess you can have UTC+x and UTC-y at the same time, but I don't think UTC+14 and UTC-12 is correct. those ought not to be at the same place since 14+12 > 24, no..?07:11.35 
kens Exactly :-)07:11.51 
  Reading wikipedia it seems to be some kind of geopolitical thing.07:12.11 
sebras2 though I have no immediate plans to go to Midway. :)07:12.17 
kens I meant that I don't see how UTC+14 and UTC-12 can both exist in a world with a 24-hour rotation period.07:13.59 
sebras2 wouldn't UTC+14 be the same as UTC-10 except that the consistently get the date wrong..?07:20.41 
  s/the/they/07:20.49 
kens Yes, apparently the islands used to be -10 and -11 and changed to +13 and +1407:21.07 
  I guess to keep the same date as their neighbours in the same country07:21.26 
jhabjan hi07:26.06 
ghostbot hi, jhabjan07:26.06 
jhabjan kens, chrisl, i saw you spoke about Ghostscript.NET and Ghostscript Studio couple of days ago07:26.51 
kens Briefly yes07:26.59 
jhabjan to introduce myself, i'm the author of both07:27.18 
kens Aha, nice to meet you07:27.29 
jhabjan if you have couple of minutes to share with me, i would like to advise about licensing issue with you07:28.22 
chrisl jhabjan: hi, nice to have you here07:28.30 
kens Oh dear, I'm not a legal expert (as my meanderings about licences the other day should handily prove)07:28.53 
jhabjan no worries07:29.11 
chrisl Yeh, our discussions here tend to revolve around "do we think we need to refer this 'upwards'?"07:29.57 
jhabjan i just want to say why i published my source code under MPL v.2.0.07:29.58 
  and to see what options do i have07:30.11 
kens Feel free to chat, but opinions are totally our own (however half baked)07:31.08 
jhabjan the thing is that my library has some code i ported from c to C#, and that original code was released under MPL07:31.38 
  the code i speak about is Memory Module07:32.07 
  it allows to load a library from a memory instead from the disk07:32.30 
  a same way LoadLibrary works07:32.40 
  gsdll api calls allows only single instance of the ghostscript per process07:34.13 
chrisl The potential problem is how MPL interacts with (A)GPL3, and I'm not at all clear on that.... but also, MPL, as posted on your site is, I believe, incompatible with the GPL, anyway07:34.16 
  Admittedly, that's based on wikipedia, which isn't exactly totally reliable!07:35.16 
jhabjan by loading gs from the memory, i was able to make a workaround for that problem07:35.19 
chrisl By default, gs is neither thread safe nor reentrant, so multiple instances probably isn't a great idea......07:35.56 
  Unless that code allows you to load DLLs into separate address spaces, in some fashion07:36.36 
jhabjan chrisl, i'm not an expert in legal / licensing also, that's why i'm here... try to find out what i can and can't do about the license in my source code07:37.09 
  my workaround allows me to load DLL's in separate address space07:37.45 
chrisl jhabjan: the problem, as I see it, is how GPL3 restricts linking to code of another license - I *know* you can't do that with closed sources, I am totally unclear on how that relates to other open source licenses07:38.42 
jhabjan yep, i get it07:40.11 
chrisl I'm rereading the wikipedia page on MPL, and by my reckoning, you are fine - but my opinion may not reflect that of the Ghostscript copyright holder (spot the disclaimer!)07:41.12 
  jhabjan: I'm basing my thinking on this: http://en.wikipedia.org/wiki/Mozilla_Public_License#Compatibility_with_other_licenses07:42.24 
jhabjan chrisl, do you have someone you can pass this issue / question to?07:42.32 
chrisl jhabjan: you can get in touch with Artifex ( http://www.artifex.com/ )07:43.12 
jhabjan i don't want get into the trouble with this legal stuff07:43.44 
  just want to give my contribution to the developer community :)07:44.09 
  will try to contact artifex07:44.25 
chrisl Well, I can't promise, but I think you are, at the very least, within the spirit of the open source licensing, so likely you'll be fine.07:44.45 
jhabjan let's hope so :)07:45.16 
  other thing i did not think about when releasing my source code was the name07:45.45 
chrisl jhabjan: if you just explain, as clearly as you can, what you need to clarify - we're not generally out to stomp on projects like yours07:46.17 
jhabjan now i realized that using a 'Ghostscrip' name was not a good idea07:46.18 
kens I don't think Artifex would hit you with a law suit as an opening gambit anyway, a poliote letter would be the first point of contact I think.07:46.29 
jhabjan yep07:47.21 
chrisl jhabjan: the name might be a bigger issue, *but* if you make contact, explain the situation, etc, that's far better than being "found out"....07:47.24 
jhabjan i will send them an email explaining my point of view07:49.12 
  well, thank you both07:51.14 
  will leave you to work now :)07:51.19 
chrisl jhabjan: np, and do come back - it's nice to see people doing new stuff with Ghostscript!07:52.03 
jhabjan it's not a new stuff :) i saw what's missing in .NET community so i took some free time to create a wrapper that will provide all the features of the Ghostscript library. Most of .NET wrapper that can be found on the internet are half done, messed up, not working... and neither one had ability to rasterize pdf or ps directly to the screen...07:57.02 
  will keep in touch :) my email address: habjan (at) gmail.com08:00.05 
chrisl jhabjan: oh, channel is logged, so plain e-mail addresses are a bit of an invitation to a spamplosion..... I'll see if I can edit the logs08:02.36 
jhabjan my bad08:03.24 
  thanks chrisl08:03.43 
chrisl It's probably not a big deal with gmail's filters, but still08:03.48 
jhabjan yep08:03.56 
ray_laptop mvrhel_laptop: did you assign it to me ? (the fuzzing bug)16:30.44 
Robin_Watts ray_laptop: I have various bugs that I think should be yours.16:30.59 
  bug 69374116:31.25 
  bug 69374216:31.29 
  bug 69374316:31.30 
  bug 69374516:31.35 
  bug 69375016:31.39 
  bug 69375116:31.41 
  bug 69375216:31.44 
  bug 69375316:31.45 
  bug 69375416:31.49 
  bug 69375516:31.56 
  for starters :)16:31.59 
mvrhel_laptop ha16:32.26 
  Robin_Watts: that is fine16:32.31 
  pile them on16:32.35 
  oh I thought those were for me16:32.47 
  not ray_laptop16:32.53 
Robin_Watts mvrhel_laptop: I may have some for you too :)16:33.11 
  bug 693749 ?16:33.38 
mvrhel_laptop oh good i dont want to be left out16:33.46 
ray_laptop mvrhel_laptop: here's one that I think is yours: (part of bug 694133). Let me know if I should open one a new bug, or if you just want to cc on the bug: 3792.pdf.SIGSEGV.c5.3104 SEGFAULT in pdf14_pop_transparency_group16:33.47 
  mvrhel_laptop: it fails on Windows 32-bit as welll with -sDEVICE=pgmraw -r300 -dMaxBitmap=1000016:34.29 
Robin_Watts mvrhel_laptop: I'm sure that when people start to actually look at the valgrind ones there will be lots of passing the hot potatoes.16:34.30 
  mvrhel_laptop: bug 693811 has the magic "pdf14" word in it that makes me want to pass it on asap.16:35.26 
ray_laptop I made the mistake of starting with one that had a lot of files, thinking I could fix a lot of things. Wasted a lot of time finding that most were already fixed.16:35.34 
Robin_Watts ray_laptop: Yeah, I think step 1 for these bugs is always to try cut/pasting the lines from the log on peeves to verify that they do still occur.16:36.40 
  I've learnt that to my cost :(16:36.49 
ray_laptop Robin_Watts: how did the clist ones get assigned to you ?16:37.36 
Robin_Watts ray_laptop: I think I was the last person to touch the file when marcosw ran the tests.16:37.59 
  which serves me right for touching things that don't belong to me :)16:38.26 
ray_laptop Robin_Watts: may be. I thought the arrangement was that marcos would leave them assigned to support, then we would review and 'take' our bugs16:38.43 
Robin_Watts These were in the previous batch.16:39.51 
mvrhel_laptop back. sorry I was gone for sec16:40.11 
ray_laptop Robin_Watts: I'll take the ones that I think are mine.16:40.26 
  mvrhel_laptop: why did you thing those were yours ?16:40.39 
  s/thing/think/16:40.47 
Robin_Watts I think he just misread irc.16:41.09 
mvrhel_laptop I saw a bit list from Robin after I just got done handing a bug to someone so I figure I was geting my just reward16:41.18 
  s/bit/big/16:41.23 
  yes. I misread16:41.26 
Robin_Watts tor7, paulgardiner: Another simple commit on robin/master16:41.43 
mvrhel_laptop ray_laptop: if you can open a new bug for the one, I would appreciate it16:41.53 
  ray_laptop: you can give me 69381116:42.25 
  and I will take bug 693749 from you Robin_Watts 16:42.46 
ray_laptop mvrhel_laptop: 693811 was robin's, not mine. I see you took it on16:46.58 
Robin_Watts ray_laptop: ssshh! He might give it back!16:47.24 
ray_laptop mvrhel_laptop: so what about the 3792.pdf.SIGSEGV.c5.3104 SEGFAULT in pdf14_pop_transparency_group one ?16:47.39 
  mvrhel_laptop oops. I miissed your reply. Sorry. I'll open a new one for you16:50.16 
mvrhel_laptop ray_laptop: ok thanks16:50.23 
Robin_Watts tor7, paulgardiner: And another one.16:52.15 
mvrhel_laptop ray_laptop: so do you want 694190?16:54.04 
  it is a very odd one16:54.19 
  that I feel needs to be solved in the interpreter16:54.28 
ray_laptop mvrhel_laptop: I thought you were going to add a new compositor action to abort out of transparency with something other than .poppdf14devicefilter16:59.37 
mvrhel_laptop I did17:00.00 
  and it solved several problems17:00.06 
  but not this one17:00.09 
  this one is very odd17:00.17 
  the interpreter pushes a group17:00.32 
  then it pushes a mask17:00.37 
  then it pops the group17:00.40 
  not a good idea17:00.46 
ray_laptop so it doesn't get an error ?17:00.53 
mvrhel_laptop well there is an error prior to the group pop17:01.13 
  but not from the pdf14 device17:01.32 
  and I don't know what the interpreter is doing with the error that it did get17:01.44 
Robin_Watts mvrhel_laptop: What's the upshot of this? bad rendering? or a crash?17:01.50 
mvrhel_laptop but it is not shutting down the pdf14 device17:01.52 
  crash17:01.53 
ray_laptop mvrhel_laptop: no, I meant: does the error get back to the interpreter ?17:02.02 
mvrhel_laptop yes17:02.08 
ray_laptop Robin_Watts: it's a segfault17:02.14 
Robin_Watts To stick my nose in where it's probably not warranted... nothing the interpreter does should really be able to crash the core.17:02.30 
mvrhel_laptop right17:02.38 
  but it is 17:02.41 
Robin_Watts Cos any user could write similar postscript and cause a crash.17:02.51 
ray_laptop Robin_Watts: agreed. But the pdf14 actions are not documented so users don't use them17:03.22 
Robin_Watts so the core should probably be fixed so that it doesn't crash. And the interpreter should be fixed so it doesn't do silly things.17:03.34 
  ray_laptop: Security holes generally aren't documented, but people make use of them too :/17:03.55 
ray_laptop but it might be better if the pdf14 device had better handling so it didn't depend on the interpreter to avoid crashing17:04.15 
Robin_Watts Being able to crash ghostscript with any input file is a security hole, I feel.17:04.41 
mvrhel_laptop ok. I can add in special checking to make sure that we have a group to pop within a mask but I don't see why the interpreter would ever be doing this17:05.03 
ray_laptop mvrhel_laptop: so, if i agree with Robin_Watts (and I tend to), then it stays with you to fix the device so that sequence doesn't cause a crash :-)17:05.08 
mvrhel_laptop the pop can only come from the interpreter17:05.10 
  to me, there is a flaw in the interpreter since it is doing this17:05.43 
  it makes no sense17:06.02 
Robin_Watts The interpreter may well be doing something wrong, but the core should NEVER crash.17:06.09 
mvrhel_laptop the interpreter is def doing something wrong17:06.20 
  there is no maybe about it17:06.24 
ray_laptop mvrhel_laptop: maybe, but as Robin_Watts points out the PDF interpreter is just PS that exposes a way to force a SEGFAULT17:06.25 
Robin_Watts So there are 2 bugs here. One in the interpreter, one in the core.17:07.01 
ray_laptop mvrhel_laptop: so after it no longer segfaults, if you want one of us (like me) to change the interpreter as well, that's fine17:07.15 
mvrhel_laptop ok. fair enough17:08.03 
Robin_Watts feels a stabbing pain in his side as mvrhel finds his robin voodoo doll.17:08.05 
mvrhel_laptop yes, where did I stash that17:08.17 
Robin_Watts Sorry.17:08.24 
mvrhel_laptop I understand your point. but I do like to catch issues early and not deep down later17:08.49 
ray_laptop mvrhel_laptop must have fumbled with the pin. He usually doesn't hit mine in the side (more like the backside)17:08.51 
Robin_Watts paulgardiner, tor7, sebras: 3 there now.17:13.07 
mvrhel_laptop this one is just a mess17:14.23 
  it really bothers me that the interpreter is doing this17:14.53 
  hmm actually I think I see what I can do with this17:22.57 
Robin_Watts mvrhel_laptop: Would a cutdown version of the file help?17:23.49 
mvrhel_laptop Robin_Watts: no I think I see a problem now in the pdf14 device code17:24.08 
  with a buffer resizing/color conversion17:24.24 
  fixed the one issue, but not the file merrily makes its way all the way to the pdf14 put image and there is no way for me to tell that we are out of sync in our pushes and pops 17:46.12 
  s/not/now/17:46.23 
  I guess I will need to add in some sort of push and pop counter to make sure everything has evened out before we do the pdf14 put image17:46.54 
  actually, I can just look at the saved value in the context stack17:48.16 
  and if there is something there I will return an error17:48.31 
ray_laptop mvrhel_laptop: sounds like a reasonable approach17:55.32 
mvrhel_laptop yes. it seems to work too17:55.46 
  the file no longer segvs17:55.59 
ray_laptop mvrhel_laptop: great. I have to run an errand. bbiab17:56.41 
mvrhel_laptop ttyl17:56.51 
Robin_Watts mvrhel_laptop: Nice one.18:11.51 
mvrhel_laptop Robin_Watts: so 694158 is really bad18:12.29 
  I had to reboot 18:12.33 
  It complains about some j2K issues and my machine just goes to a crawl18:13.13 
  are you familiar with that part of the code?18:13.35 
Robin_Watts mvrhel_laptop: Oh, does it massively malloc or something?18:13.38 
mvrhel_laptop likely18:13.42 
Robin_Watts No, not at all.18:13.44 
mvrhel_laptop ok18:13.52 
Robin_Watts but if you want to pass it to me, I might be able to work some memento magic on it.18:14.08 
mvrhel_laptop ok. I will do that18:14.14 
  sigh I think that is if for the segv's the rest look for me look like valgrid issues to be dealt with18:52.48 
  bbiaw19:02.50 
ray_laptop 3TB external WD drive for $95. I may have to get one just for backups :-)19:57.19 
mvrhel_laptop time to beat on non-isolated knockout groups. see if I can get this thing off my list20:27.10 
Robin_Watts ray_laptop: For the logs, I got a 4TB Hitachi external drive a while ago, and broke it open to take the drive out. It's now the internal spinning disc in my main machine next to my SSD.22:54.27 
  ooh, rays customer back in touch.22:59.28 
  ray_laptop: (For the logs) Can you grab their simulator etc from the Ftp site (or mail me details of how to get it) please? Thanks.23:00.04 
 Forward 1 day (to 2013/09/24)>>> 
ghostscript.com
Search: