| <<<Back 1 day (to 2011/08/08) | 2011/08/09 |
alexcher | cryptopsy: I've read a book about this, probably, by Agatha Christie. | 03:04.51 |
robin_watts_mac | tor8: You here? | 14:40.44 |
tor8 | robin_watts_mac: yes. | 14:42.34 |
robin_watts_mac | http://wss.co.uk/pinknoise/except/ | 14:43.32 |
| I wonder if that would be useful with mupdf. | 14:43.51 |
| It's exceptions done using setjmp/longjmp. | 14:44.07 |
| I have a local branch here with a 'memento' build option. | 14:45.35 |
tor8 | robin_watts_mac: possibly. it would be a lot of upheaval, and i'm not sure how it would take care of the nested cleanup case with reference counted objects. | 14:48.14 |
| the alloced case is easy, because free(NULL) is safe, but drop(something-i-may-or-may-not-have-kept-depending-on-where-the-throw-occurred) | 14:48.47 |
| ...is not | 14:48.54 |
robin_watts_mac | tor8: That's why I was thinking about that lib. | 14:49.06 |
| It lets you nest Except { } Catch { } things. | 14:49.20 |
tor8 | if you go back far enough in the git history, you'll see if (erro) goto cleanup; a lot | 14:49.25 |
| with possibly many cleanup-labels, to get the reference counting and resource cleanup nesting right | 14:49.59 |
robin_watts_mac | If we always 'keep' things by writing them into structures, and destroy them by keeping them in locals, it's easier. | 14:50.58 |
| There is no magic bullet though of course. | 14:51.10 |
tor8 | no, or we'd all be using that lib (or something similar) :) | 14:51.31 |
| I'm really fond of the way Go handles resource cleanups in the absence of exceptions with the 'defer' statement | 14:51.50 |
| f = open("filename"); defer f.Close(); â¦do stuff... | 14:52.16 |
| I doubt it'd be easy to add the same with macros in C though :( | 14:53.35 |
robin_watts_mac | indeed. | 14:53.45 |
| hey henrys | 14:56.10 |
henrys | where are you now? | 14:59.42 |
robin_watts_mac | Hanoi. | 14:59.49 |
henrys | gxdownscale.c isn't tested with the cluster right? I assume not - Robin_Watts_Mac has adjust_width where I'm sure he intends adjust_width_proc on line 742 gxdownscale.c | 15:51.42 |
robin_watts_mac | It is used. | 15:53.55 |
henrys | in the overnight? | 15:54.48 |
robin_watts_mac | But you're right, that's wrong. | 15:54.57 |
| I have both an adjust width and an adjust_width_proc | 15:55.13 |
| and currently adjust_width is only !=0 when adjust_with_proc != NULL, so it's equivalent, but nasty. | 15:55.32 |
henrys | I have a patch just wondering if it should be cluster untested or not. | 15:55.47 |
| or you can fix it if you want. | 15:55.54 |
robin_watts_mac | just fixing now. | 15:56.02 |
henrys | thanks | 15:56.17 |
| our first fix from Hanoi | 15:57.35 |
mvrhel2 | Hows the temp and humidity there? | 15:58.00 |
robin_watts_mac | ask me tomorrow - we arrived only recently (late at night) and it was warm and humid. | 15:58.30 |
| The real test will be in theday tomorrow. | 15:58.36 |
mvrhel2 | have fun! | 15:59.37 |
henrys | so I have another meeting at 9:30 so I'll be ducking out then. | 16:00.03 |
robin_watts_mac | Anything in the meeting for me, cos I'm about to go to bed. | 16:00.26 |
henrys | robin_watts_mac:not that I know of. | 16:00.40 |
| you should be on vacation! | 16:00.54 |
robin_watts_mac | mvrhel2: You can use the freenode webchat if irc is messing you around. | 16:00.58 |
mvrhel2 | that is what I am doing now :) | 16:01.13 |
| I suspect that the router is wacky at this place | 16:01.27 |
henrys | mvrhel2:I wanted to ask if you could clean up the warnings in the icc stuff. marcosw waring email has a link to the unix warnings - maybe msvc isn't as bad and you don't see them. | 16:02.28 |
mvrhel2 | henrys: sure I can do that. I will check into today | 16:03.05 |
henrys | alexcher:did you have any luck producing a simple shading example so we can understand the memory issues. | 16:03.08 |
| ? | 16:03.09 |
alexcher | henrys: not yet. | 16:03.40 |
henrys | I did see a customer complaint about 9.04 hopefully marcos will process it quickly. | 16:04.38 |
mvrhel2 | I am starting to suspect that bug 692378 is an interpreter issue | 16:04.52 |
henrys | other than that the release looks good, yea! | 16:05.00 |
mvrhel2 | It would seem that the set color is getting messed up | 16:05.03 |
| the handling of the fill for sep with colorant value of all is correct in the c code | 16:05.33 |
henrys | alexcher? | 16:05.53 |
mvrhel2 | alexcher: would you be willing to dig into this one a bit? | 16:05.53 |
| I can assign it to you... | 16:06.14 |
chrisl | henrys: that's the PDF/A thing? It seems to be a bit of a moving target - validation changes with the wind, it seems :-( | 16:06.23 |
alexcher | mvrhel: yes, I'll trache the color values at different levels. | 16:06.39 |
henrys | chrisl:yes | 16:06.45 |
mvrhel2 | ok. great. thanks alexcher | 16:06.51 |
| I will add one more comment | 16:07.11 |
henrys | also looking at alexcher customer bug list it seems there are a few color problems - were these problems introduced with the color changes alexcher? | 16:07.44 |
| I'm looking at marcos' bug aging report. | 16:08.49 |
alexcher | henrys: what are the bug numbers? | 16:08.55 |
henrys | well we can go through all of them in the report. | 16:09.26 |
| 689290 - soft mask? | 16:09.40 |
| how much of that is interpreter vs. michael stuff? | 16:09.59 |
mvrhel2 | uhoh. does that mean I hand alexcher one and he hands me back 10 | 16:11.18 |
alexcher | henrys: I don't have anything to say about 682290 now. | 16:12.04 |
henrys | well sitting on the bug aging list for for eons seems unattractive. | 16:12.15 |
| and the memory allocation and high res bugs, maybe you can simplify but I doubt they are language issues: 689323 and 689933 <- alexcher. | 16:14.11 |
mvrhel2 | there is a patch for 689290 | 16:16.21 |
alexcher | henrys: I don't have anything to say about 689290 now. | 16:16.35 |
| henrys: please disregard my previous message | 16:16.59 |
mvrhel2 | does the clusterpush come back fine with it? | 16:17.06 |
henrys | tor8:your lone xps bug on the list can probably be moved along also. | 16:18.33 |
| I realize some bugs will stay on the list but it does seem like a few of these could be worked on. | 16:19.21 |
| anything else for "skeleton crew meeting" | 16:20.23 |
| ? | 16:20.25 |
alexcher | henrys: What to do with OpenJPEG library? | 16:21.03 |
henrys | marcosw_:did you have anthing for the meeting. | 16:21.13 |
| ? | 16:21.15 |
marcosw_ | no, I'm good. | 16:21.30 |
henrys | alexcher:I said my vote was to treat it like casper. | 16:21.38 |
| jasper. | 16:21.43 |
tor8 | henrys: the more memory hungry xps bug? I'd be inclined to close that as wontfix. | 16:21.46 |
henrys | I always do that. | 16:21.50 |
| tor8:incline away! | 16:22.02 |
tor8 | it's the transparency buffer use, would be my guess | 16:22.17 |
marcosw_ | I've asked Miles to reboot the two cluster nodes at his office but it didn't seem to work. I may have to make a trip up there (I was going to deliver one of the new cluster nodes next week anyway). | 16:23.01 |
henrys | alexcher:does that answer the question - chrisl may have a view on it. | 16:23.04 |
alexcher | henrys: Yes, does anybody has a different approach? | 16:23.08 |
henrys | you may have to call it's configure from our makefile like tiff. | 16:23.50 |
chrisl | henrys, alexcher: my preference is to pull in third party libs fully, and as close to their canonical release possible. | 16:24.08 |
tor8 | alexcher: I made the thirdparty/openjpeg.git a subset, since they started doing what we do -- include all external dependencies in their source repository. and none of them are needed for the library, all of them used for the various command line tools. | 16:24.44 |
henrys | maybe chrisl can set up the directory and alexcher can get it working once it is in the tree. I don't know if that helps alexcher much. | 16:26.03 |
| oh god openjpeg has third party libraries? | 16:26.55 |
alexcher | I have a revision that compiles but doesn't work yet. | 16:27.49 |
henrys | I am expecting my other meeting to start in 3 minutes so I have to go. | 16:27.56 |
| please continue without me. | 16:28.05 |
tor8 | henrys: yup. libpng, libtiff, libz, liblcms2. only used for the command line tools, not for the library thankfully. but it makes pulling/merging their source tree more troublesome. | 16:28.48 |
alexcher | makefiles are mostly done but I need some help with autoconf. | 16:28.50 |
mvrhel2 | ok. I am going to go and work on cleaning up my warning... | 16:29.40 |
| s | 16:29.47 |
chrisl | alexcher: if you stick the changes on a branch on casper, I can sort out the autoconf stuff. | 16:29.53 |
alexcher | mvrhel3: -W3 generates some useful warnings and a lot ow junk. | 16:30.17 |
| chrisl: OK, that's what git is for. | 16:30.55 |
chrisl | alexcher: and given we don't need/want the third party stuff in openjpeg, we could do without their "libs" directory. | 16:31.08 |
mvrhel2 | alexcher: ok thanks | 16:31.52 |
| hmm lots of broken const... | 16:34.11 |
alexcher | chrisl: current OpenJPEG repository works for me. I just don't know a proper way to add it to gs tree. | 16:35.06 |
mvrhel2 | other than than it is not too bad | 16:35.18 |
| bbiab | 16:35.23 |
chrisl | alexcher: Okay, leave it with me - I've got a couple of things to finish up that I postponed for the release, then I'll get a build to include openjpeg. | 16:37.37 |
| alexcher: just to clarify: do you mean the openjpeg repos at git.ghostscript.com? Or openjpeg's own source repos? | 16:39.18 |
alexcher | chrisl: Yes, our mirror. | 16:39.53 |
chrisl | alexcher: okay, thanks. I'll try to get to it soon. | 16:40.44 |
henrys | wow long meetings suck | 17:25.48 |
| it is not clear to me how autoconf is supposed to set -L when libraries are in different directories, it is clear what we do on my mac is wrong - we have no -L option until autoconf searches for fontconfig which is found in /opt/local so that becomes the official -L for the entire build. | 19:00.05 |
| I wonder how many man years of development open source has lost to working to autoconf problems. | 19:06.14 |
| s/working to/working on/ | 19:06.34 |
mvrhel2 | alexcher: are you around? | 20:10.16 |
alexcher | yes | 20:10.24 |
mvrhel2 | so a quick question from you that we got from a customer | 20:10.35 |
| about setting his own halftone screen in ghostscript | 20:10.47 |
| did you see than email? it went out to support too | 20:11.06 |
| so, what is the easist way for him to make sure that his screen is used and not the default ones in gs? | 20:11.50 |
alexcher | No, I didn't get any new email in the last 30 min. | 20:12.11 |
mvrhel2 | gs is set up embedded in their device I belive | 20:12.12 |
| hmm. ok. let me forward it to you | 20:12.27 |
| I am thinking he is going to want to do changes in the init files | 20:13.14 |
alexcher | I've answered a similar question in the comment #4 to the bug 692390. | 20:13.46 |
| The email is received. | 20:15.05 |
mvrhel2 | oh. thanks alexcher | 20:15.11 |
| I will let him know about this | 20:15.18 |
| if you have any other thoughts after reading the email, let me know. Like I said they are using gs embedded, but I think for now what you answered on the bug will work. this should go in an FAQ | 20:16.31 |
alexcher | It is also possible to add the halftone to Halftone directory. | 20:20.14 |
mvrhel2 | alexcher: what would that involve? | 20:20.32 |
alexcher | It we don't check for /Default halftone on start-up, it's an easy to fix bug. | 20:21.05 |
| mkdir Resource/Halftone; cp FOO Resource/Halftone/Default | 20:22.01 |
mvrhel2 | alexcher: so if there is a halftone in this directory it is used as the default? | 20:24.11 |
alexcher | It should be used. gs defines default halftone in memory. I'm not sure that it checks the resources first. | 20:25.38 |
mvrhel2 | alexcher: ok I will check this out in a bit. dinner time now.... | 20:27.38 |
| alexcher: thanks for the help | 20:27.47 |
| Forward 1 day (to 2011/08/10)>>> | |