| <<<Back 1 day (to 2013/10/03) | 2013/10/04 |
kens | thinks I'mgoing to set all the -dPDFSETTINGS to do nothing at all | 07:25.20 |
Robin_Watts | tor7: so I have a laptop to mail. | 10:54.52 |
kens | That w\s quick | 10:55.18 |
Robin_Watts | tor7:ping me when you're around and we can talk about parcels and insurance. | 10:58.35 |
kens | Don't forget Tor has moved | 10:58.52 |
| Joann's address is (I think) incorrect | 10:59.02 |
Robin_Watts | kens: Yeah. | 10:59.19 |
tor7 | Robin_Watts: ping! | 10:59.20 |
Robin_Watts | via private tab. | 10:59.31 |
tor7 | kens: I have sent my new address to Joann | 10:59.47 |
kens | Fair enough :-) | 10:59.53 |
Robin_Watts | tor7: Laptop won't be sent until monday. Sorry. | 13:21.11 |
| I just tried to drop it at the local post office and they won't take it cos it's "Global priority". | 13:21.37 |
| Stupid bloody red tape. | 13:21.46 |
| tor7: Android fix on robin/master | 13:54.26 |
| Hmm. It looks to me that we let jbig2dec call malloc/free too :( | 14:36.26 |
chrisl | Robin_Watts: yes, we do. I have an open bug about it | 14:37.08 |
Robin_Watts | At least it looks like jbig2dec CAN be told to use other allocators sanely. | 14:37.29 |
| So one of my valgrind bugs comes down to the fact that jbig2dec is not initialising all the pixels in an image. | 14:40.21 |
| presumably it's hitting an error and baling out part way through leaving the rest unset. | 14:40.39 |
| I can fix it by memsetting the data block to zero before starting, but that feels wrong. | 14:41.00 |
kens | Yes, because its expensive and (should be) pointless | 14:41.17 |
chrisl | Hmm, IIRC, there was a CCITT decode bug that was leaving bytes uninitialised in the middle of the image | 14:41.37 |
Robin_Watts | I could wrap it in #ifdef PACIFY_VALGRIND, but that's nasty too. | 14:41.52 |
| Let me see if I can pin it down some more. | 14:42.07 |
kens | Are the pixels being accessed ? If there's an error they really shoudn't be | 14:42.14 |
Robin_Watts | No error is reported to the user. | 14:42.31 |
kens | smells more like a real problme then | 14:43.07 |
chrisl | Robin_Watts: FWIW, fixing jbig2dec's memory manager will mean an API change - it doesn't provide for context to be passed around | 14:43.57 |
tor7 | Robin_Watts: LGTM. | 14:44.18 |
Robin_Watts | chrisl: Malloc/Free etc are passed a Jbig2Allocator *... | 14:44.51 |
chrisl | Which doesn't contain a context pointer | 14:45.11 |
Robin_Watts | chrisl: It doesn't need to does it? | 14:45.48 |
chrisl | How do we get a gs_memory pointer? | 14:46.05 |
Robin_Watts | All it needs is for there to be a mechanism for us to replace the JBig2Allocator * with a GsJbig2Allocator * | 14:46.26 |
| Where a GsJbig2Allocator starts with a Jbig2Allocator, and continues with a gs_memory_t *. | 14:46.45 |
chrisl | Possibly. Personally, I'd prefer to fix the API | 14:47.26 |
Robin_Watts | chrisl: The API is fine. | 14:47.37 |
chrisl | The API specifies a Jbig2Allocator * which isn't sufficient | 14:48.07 |
Robin_Watts | chrisl: Right. And we pass in a Jbig2Allocator *. | 14:48.22 |
| However, the function pointers in there, "know" that they can cast that back to a super class and get the required extra field. | 14:48.52 |
| It's a perfectly respectable Object Oriented way of working. | 14:49.09 |
chrisl | Given that we own the code, why not do it right? | 14:49.14 |
Robin_Watts | because it's right the way it is. | 14:49.23 |
| Or at least, it's not wrong :) | 14:49.37 |
| We could just add a void * to the Jbig2Allocator struct if you want. | 14:49.50 |
chrisl | That's what I would do, yes | 14:50.03 |
Robin_Watts | I don't count that as a change to the API. | 14:50.07 |
| as no ones code will be broken by that. | 14:50.18 |
| openjpeg needs new args to be added to functions etc. That is a mess. | 14:50.40 |
chrisl | Yes, that why I thought we should leave that to Shelly | 14:51.00 |
Robin_Watts | yeah, I opened a bug, and mailed him about it, and he got back to me. He's talking to the openjpeg guys now. | 14:51.56 |
| So I can see that jbig2dec is detecting a problem in the image and baling out of decompressing. | 15:06.28 |
| but it uses the image anyway. | 15:06.33 |
| I could change it to not use the image. | 15:06.52 |
| but that might break some files that were relying on the fact that we were at least seeing the image as much as was possible. | 15:07.23 |
| (i.e. if the error is on the last scanline of an image, should we really throw away everything we've got so far?) | 15:07.52 |
ray_laptop | kens: we could hack the PS code so that when -dPDSETTINGS=printer is processed (in gs_pdfwr.ps) we could define UseCIEColor, right ? | 15:46.54 |
kens | ray_laptop : possibly, but its code that I dno't intend to pursue, because there is new colour code that doesn't need it | 15:53.58 |
Robin_Watts | ray_laptop: Just to be clear - we think that we're done with Lens panic for now? | 15:54.32 |
ray_laptop | Robin_Watts: I'm going to call him and make sure, but that's what I think is the case | 15:55.51 |
Robin_Watts | cool. | 15:56.04 |
ray_laptop | kens: oh, right. You mentioned that. Makes sense not to patch it in the meantime | 15:56.26 |
| Robin_Watts: Just talked with Len. They backed out the gxblend? change for compose_inner_group because it caused some crashes and some bands missing. He will be sending email with the details. | 16:11.37 |
| Robin_Watts: but even with that out, the performance is acceptable | 16:11.53 |
Robin_Watts | really? | 16:11.53 |
| how strange. | 16:12.09 |
ray_laptop | Robin_Watts: yes, he's really sending an email | 16:12.17 |
| Robin_Watts: what's strange ? | 16:13.40 |
| that the performance is still OK, or that the gxblend change caused a problem ? | 16:14.05 |
Robin_Watts | the latter. | 16:14.40 |
ray_laptop | Robin_Watts: he's going to give us the files that were problems (probably FTP site). | 16:16.40 |
| Robin_Watts: while it's not a fire drill, it would be nice to figure out what's going on with the problem files. | 16:17.15 |
| and it's probably better done now while the changes are not to stale in the synapses | 16:17.53 |
| but since it is not urgent for them, it's up to you whether or not to do it now | 16:18.42 |
Robin_Watts | I will look as soon as the email arrives. | 16:18.56 |
ray_laptop | bbiab (about 5 min) | 16:19.00 |
Robin_Watts | ray_laptop: I found one crash. stupid mistake on my part. | 16:42.36 |
ray_laptop | Robin_Watts: sorry -- I was distracted (by other work). Great! | 17:04.30 |
Robin_Watts | ray_laptop: no worries. | 17:04.39 |
| I am constructing a changelog for Len now. | 17:04.48 |
ray_laptop | Robin_Watts: are you able to get the ATS files ? | 17:05.00 |
Robin_Watts | I'll pastebin it to you in a mo for you comments. | 17:05.03 |
| ray_laptop: I have tested the first ATS file. | 17:05.17 |
| I will test the second in a mo. | 17:05.30 |
| ray_laptop: ping: See private tab | 17:14.26 |
| Second ATS file looks fine too. | 17:16.20 |
ray_laptop | OK. I read the email. It has everything in it. Thanks. | 17:41.30 |
Robin_Watts | Gah. Whoever thought that fuzzing encrypted PDF files was a good idea? :( | 19:19.51 |
ray_laptop | oh, great! My windows explorer crashed and after reboot, msysgit won't start (it pops up a window and immediately exits) | 19:20.34 |
| Robin_Watts: umm, wasn't it the G folks ? | 19:20.55 |
Robin_Watts | yeah. | 19:21.16 |
| Just makes it MUCH harder to cut the file down to find out which bit is upsetting it :( | 19:21.44 |
klyed2 | Hey there, I'm using a windows machine and am trying to get a custom font to be used. I am getting a "Error: /invalidaccess in /findfont" message. I tried using a default windows font and receive the same error. I also tried running gswin32c.exe as administrator. Any hints? | 19:22.16 |
| I should mentioned I am on version 8.71 | 19:22.45 |
ray_laptop | klyed2: are you running with -dSAFER ? | 19:23.01 |
Robin_Watts | So only 4 years out of date? | 19:23.03 |
ray_laptop | Robin_Watts: yeah, like cust 532 | 19:23.18 |
klyed2 | ray_laptop: yes that command line is in there. I'm using image magick to convert a pdf to an image. | 19:24.04 |
| I can remove it though. | 19:24.13 |
Robin_Watts | What's the voodoo to show the PDF operators as they are run in gs? | 19:24.22 |
ray_laptop | klyed2: invalidaccess is detected BEFORE trying to open the file | 19:24.25 |
| klyed2: you can either make sure that the folder (path) for the font is given with -I______ or get rid of -dSAFER | 19:25.09 |
| klyed2: any paths listed with -I are automatically added to folders that can be accessed in SAFER mode. Otherwise, invalidaccess | 19:25.43 |
klyed2 | ray_laptop: thanks, that fixed that error message. received another message for a similar font now. | 19:32.40 |
| erm similar error message, will look into it | 19:32.56 |
ray_laptop | klyed2: removing -dSAFFR should allow access for any path | 19:35.06 |
| Robin_Watts: Looks like I need to reinstall msysgit -- what's a good version to use (I see things like 1.8.4preview and a bunch of other Beta versions, but am not sure about that) | 19:44.43 |
Robin_Watts | ray_laptop: I'd use the latest stable version personally. | 19:46.06 |
ray_laptop | google isn't finding anything not labelled "Beta". I tried 1.7.4 (what I downloaded last) and it also dies. :-( Windows is sick :-( | 19:54.20 |
| hmm... the problem is Avast (my anti virus) "Behavior monitor" apparently it decided that bash.exe was a bad thing | 20:10.17 |
| I made c:\msys\bin\bash.exe a "trusted process" and it runs, although I _do_ see a message when it starts that's a little worrisome: | 20:13.18 |
| m.AllocationBase 0x0, m.BaseAddress 0x68540000, m.RegionSize 0x120000, m.State 0x10000 | 20:13.19 |
| C:\msys\bin\sed.exe: *** Couldn't reserve space for cygwin's heap (0x68540000 <0xA70000>) in child, Win32 error 0 | 20:13.21 |
| That wasn't there before. | 20:13.22 |
| oh, wonderful :-( 'ls' worked the first time, but now it errors out (with a similar error) as does git or anything except the shell. Guess avast has to be disabled for me to get work done :-( | 20:15.24 |
| maybe time to switch AV progreams | 20:15.36 |
| it probably doesn't like the cygwin use of "shared memory" | 20:17.29 |
| oops. does ghostbot need to be kicked ? | 20:19.01 |
| ghostbot: hello | 20:19.05 |
ghostbot | que tal | 20:19.05 |
klyed2 | ray_laptop: huh, i took another look at the problem, this time with 9.x version of post script and everything works fine which is odd. | 21:03.38 |
| I did not have to muck around with custom font mapping. | 21:03.57 |
| Thanks for your help anyways =) | 21:04.11 |
| (The other error complained about some Adobe-Identity CDIFont thing) | 21:04.29 |
| Forward 1 day (to 2013/10/05)>>> | |