IRC Logs

Log of #ghostscript at irc.freenode.net.

Search:
 <<<Back 1 day (to 2014/08/17)2014/08/18 
kens Great, my changes to the JPX code in the PDF interpreter cause a segfault on the cluster when running an EPS file.....07:49.46 
chrisl Erm.......07:51.08 
kens Yeah, I know, but I ran ti twice, same result07:51.18 
  Bug688845.eps to pkmraw at 300 dpi07:51.46 
  And of course, it works just fine here on Windows07:54.33 
chrisl If you push the changes to your repo, I can try it on Linux07:56.19 
kens I'm just about to try it here on my VM, if that fails I'll stick it up#07:56.39 
  Or rather, if it works07:56.59 
  Yeah, works fine :-(08:01.02 
  chrisl I believe I've pushed the changes to my repo08:16.58 
  Opps, no, git complained08:17.09 
chrisl Okay, just a sec08:17.13 
kens OK --force seems to work08:17.58 
chrisl What's the command line?08:18.16 
kens let me pull it from the cluster log08:18.32 
  " ./gs/bin/gs -sOutputFile='| md5sum >>./temp/tests_private__comparefiles__Bug688845.eps.pkmraw.300.1.md5' -dMaxBitmap=10000 -sDEVICE=pkmraw -r300 -Z: -sDEFAULTPAPERSIZE=letter -dNOPAUSE -dBATCH -K1000000 -dClusterJob -dJOBSERVER - < ./... "08:18.58 
  Obviously I used a file to output to, not md508:19.10 
chrisl Doesn't crash for me08:20.06 
kens It doesn't seem to crash on all the cluster machines either, some of tehm give an 'error reading input file'08:20.25 
  It can't be the change, it must be some kind of memory thing08:20.55 
chrisl I'll try it with valgrind, but I wouldn't hold your breath......08:24.28 
kens Well its kind of difficult to fix if it won't go wrong for me :-(08:24.57 
  I'm inclined to commit it anyway, we've been ignoring the other seg fault for long enough, we can ignore 2 as easily I guess.08:25.28 
chrisl It may be compiler or library related08:26.22 
kens Umm, yes I guess. It does seem to happen on angstroms mostly I think08:26.42 
chrisl I'd commit it - it *clearly* can't be your change. Maybe mention it to Marcos when he gets back08:27.34 
kens OK, well thanks for trying it anyway....08:27.46 
chrisl valgrind reports "Use of unintialised value of size 8" but it's never going to cause a crash08:29.27 
kens Sigh....08:29.40 
rayjj IRC net seems fkaly today. And the IRC logs seem unusually quiet. Consider this a test to see if ghostbot is working19:01.25 
  nothing yet on the logs19:04.49 
Robin_Watts rayjj: No one has said anything on here since kens and chrisl this morning,19:10.35 
rayjj the garbage error message for bug 695420 is caused by a dictionary having a pointer to the OutputICCProfile (string) in it that gets freed by pdfwrite when it rc_decrements the icc_struct->device_profile[0] during converting from RGB to Gray due to ColorConversionStrategy 19:10.41 
  Robin_Watts: i see that my comments showed up on the logs. Thanks19:11.14 
  Robin_Watts: I hope you are feeling better19:11.35 
Robin_Watts Much better now (though still not right).19:11.55 
  I was properly awful on saturday.19:12.05 
  I think I've had ebola :)19:12.16 
rayjj Robin_Watts: that doesn't sound proper at all ;-)19:12.33 
  dengue was enough for me -- I don't want a different hemorrhagic disease19:13.45 
  mvrhel_laptop: I think I found something. Have time for a quick question on the icc_struct ?22:20.23 
mvrhel_laptop rayjj: hi there22:21.36 
  yer22:21.38 
  yes22:21.47 
rayjj sorry -- was starting at code.22:24.06 
  mvrhel_laptop: in gsicc_manage.c gsicc_new_device_profile_array gets passed dev->memory, and then uses the non_gc_memory to do the allocations.22:25.49 
  mvrhel_laptop: but in gsicc_init_device_profile_struct, it uses dev->memory in the gs_alloc_bytes call for the profile_name22:27.22 
mvrhel_laptop rayjj: let me open up ghostscript22:28.19 
  hmm I see it allocated and then freed shortly after22:31.37 
  the default name is copied into the allocation22:31.56 
  and then the call is made to set the profile22:32.09 
  which does an allocation with non-gc memory22:32.59 
rayjj mvrhel_laptop: Oh, I see doen in gsicc_profile_new it allocates a name and copies it22:33.01 
mvrhel_laptop then when we return we free22:33.04 
  I think it is ok22:33.06 
  yes22:33.13 
  probably not the most straight forward thing22:33.33 
  but I think it is OK memory wise22:33.39 
  back to the SOT salt mine22:33.57 
  rayjj: by the way you sent me a 10 minute skype message this weekend22:34.11 
rayjj mvrhel_laptop: OK. I still have somebody keeping a stale pointer22:34.19 
mvrhel_laptop I think someone was playing on your laptop and hit a record button22:34.27 
rayjj mvrhel_laptop: I did ?22:34.33 
  probably my youngest son, Austen22:34.51 
mvrhel_laptop the tv was playing in the background. something about a german pork roast of some sort22:35.01 
rayjj anything interesting (or embarrassing) ?22:35.12 
mvrhel_laptop I didn't listen to it after a minute22:35.23 
  I assumed it was a butt-dial22:35.32 
rayjj he uses skype to talk to his friend while playing online games (usually minecraft)22:36.15 
mvrhel_laptop sounds like a typical 8 year old (is that how old he is?)22:36.39 
rayjj although I don't recall him being on my laptop this weekend22:36.47 
  mvrhel_laptop: yes, he'll be 9 10/2622:37.03 
mvrhel_laptop ok. I thought he was 1 to 2 years younger than alden22:37.20 
  alden just turned 10 last june22:37.37 
rayjj right -- so starting 5th grade ?22:37.52 
mvrhel_laptop yes. last year of elementary22:38.02 
  ok well back to the you know where salt mine22:38.40 
  rayjj: if you find some other issue you need me to glance at, let me know22:38.57 
rayjj we aren't sure if we can get a transfer authorized at 6th to the middle school that starts in 6th grade. The "home" district elementary includes 6th grade22:39.22 
mvrhel_laptop oh ok22:39.58 
rayjj but Austen wants to start middle school where Eric and Michelle went for 6th grade22:40.25 
mvrhel_laptop of course22:43.03 
 Forward 1 day (to 2014/08/19)>>> 
ghostscript.com
Search: