| <<<Back 1 day (to 2012/08/23) | 2012/08/24 |
tor8 | chrisl: ping | 11:02.11 |
Robin_Watts | How would people feel about me changing the dashboard so that following links on it opens them in new tabs ? | 11:03.18 |
kens | I don't mind either way Robin_Watts | 11:03.34 |
chrisl | tor8: pong | 11:03.38 |
Robin_Watts | At the moment, if you follow a link (to look at a report say), then go 'back' to get back to the dashboard, you have to reload the dashboard otherwise it doesn't refresh etc as it should. | 11:04.13 |
kens | Robin_Watts : I'm usually not interested in teh dashboard by then ;-) | 11:04.42 |
Robin_Watts | I keep a dashboard tab up all day, and so am easily confused when I go back to it and find it hasn't updated for hours. | 11:05.10 |
kens | Oh, I only look at it when I need it | 11:05.26 |
chrisl | Robin_Watts: I almost always open the cluster related links in a new tab (sometimes window) so it won't make much difference to me.... | 11:05.55 |
tor8 | Robin_Watts: isn't that what the middle button is for? | 11:06.17 |
Robin_Watts | tor8: Yeah, but I always forget. | 11:06.29 |
tor8 | chrisl: I'm playing with "git-subtree" on the jbig2dec and ghostpdl repos | 11:06.47 |
| trying to see if I can get future merge/split operations between the repos done with less manual labor | 11:07.08 |
chrisl | How's it going? I haven't had a chance to look into it | 11:07.19 |
tor8 | chrisl: at the latest merge point (may 14) there are still a few derived files diffing | 11:07.44 |
| from may 14 on all development has gone into the jbig2dec repo | 11:07.54 |
| chrisl: http://ghostscript.com/~tor/stuff/jbig2dec.txt shows the diffstat | 11:08.56 |
chrisl | I opted to keep the derived files in the ghostpdl version because I wasn't happy with the robustness of "subcalling" automake etc | 11:09.33 |
tor8 | that may be problematic if we want to automatically split out the jbig2dec directory though | 11:10.12 |
chrisl | OKay, I'll take another look, and see if I can get it to work reliably | 11:11.26 |
tor8 | I was thinking we could 'git-subtree merge' in the external jbig2dec repo, or just delete gs/jbig2dec first, then bring it back in from a git-subtreee. then we can move patches automatically with git-subtree merge and git-subtree split | 11:11.26 |
| anyway, the sources modulo these derived autocrap files are identical on the may 14 date so that could be a good starting commit for pulling in the jbig2dec.git | 11:12.31 |
| hm, either my graphics card is fried or the driver is having hickups... | 11:13.08 |
chrisl | Hmm, for Ghostscript, we run the jbig2dec configure script, IIRC, in order to get one header file - I wonder if a simpler solution would be to have Ghostscript's configure create that header...... | 11:13.47 |
| oh, crap, can't do that :-( | 11:17.03 |
Robin_Watts | tor8: What are you seeing? Momentary losses of sync ? | 11:20.09 |
tor8 | scratch that, loose dvi connector causing really odd behaviour | 11:20.33 |
| Robin_Watts: http://ghostscript.com/~tor/stuff/ati-glitch.jpeg | 11:20.46 |
| horizontal flickering lines following the green fields around | 11:21.17 |
Robin_Watts | ooh, nasty. That really looks like a dying card. | 11:22.22 |
| On the positive side; you get to buy a faster one now :) | 11:22.38 |
tor8 | yeah. but it reminded me of when I was using a too long dvi cable with a too high resolution, got the same kind of horizontal glitching lines then too. | 11:22.55 |
| Robin_Watts: there's only one model that's faster than what I have and still fanless | 11:23.20 |
| and that's a fairly minor upgrade | 11:23.27 |
Robin_Watts | oh, right, I was forgetting your aversion to moving parts. | 11:23.44 |
tor8 | noisy parts more than moving, actually | 11:23.57 |
| but they generally coincide | 11:24.01 |
Robin_Watts | I have an MSI N560GTX-Ti in mine with a "Twin Frozr II" fan setup, and I don't register the fan noise from that really. | 11:26.08 |
| but your perceptions may be different. | 11:26.33 |
tor8 | Robin_Watts: the only noise my computer makes is from the two slow spinning case fans. can't hear it unless I put my ear to the case in the dead of night when all else is silent. | 11:33.57 |
| maybe a bit overkill, but I like it that way | 11:34.14 |
chrisl | tor8: okay, I've got an idea how we can get rid of the jbig2dec derived files in the ghostpdl repository and have it work "nicely" - but I need to feed the cat, and grab some lunch before I look at it properly. | 11:38.37 |
tor8 | chrisl: cool. ping me when you're back. | 11:39.10 |
chrisl | tor8: we have a small problem: current building Ghostscript from git only depends on autoconf - in order to remove the jbig2dec derived files, it adds dependencies on automake, autoheader, libtoolize :-( | 12:28.27 |
tor8 | chrisl: uh oh. I wonder though, do we really need an automake + libtool based build system for jbig2dec? | 12:37.46 |
chrisl | Certainly not as far as GS is concerned - I suspect we'd get grief from the distros if we rip that stuff out completely, though :-( | 12:38.33 |
tor8 | so leave the intermediates in both gits then? | 12:38.57 |
chrisl | I suppose so - it's a bit icky, though :-( | 12:39.32 |
tor8 | yeah... | 12:44.05 |
chrisl | I am baffled - I can't see where/how config.h.in gets converted into config.h. This is even more f*cked up and opaque than the GS build...... | 12:47.46 |
sebras | chrisl: http://www.lrde.epita.fr/~adl/dl/autotools.pdf search in this one. this is my best source of info on autotools. | 12:48.50 |
| reading the docs or the shell scripts only drives you insane. I have tried that. | 12:49.36 |
chrisl | sebras: thanks - but doesn't help :-( | 12:51.01 |
| That tells me that config.h.in is derived from configure.ac using autoheader, and config.h is derived from config.h.in by configure - but it doesn't really explain *how* config.h is derived (it's not the "normal" autoconf way) | 12:54.10 |
| Hmm, and even worse, to my way of thinking, this breaks one of the primary "benefits" of autoconf - that the configure script is supposed to be essentially standalone | 13:06.00 |
| tor8: I'm sorry, I really don't know what the best answer for this is :-( | 13:08.11 |
tor8 | chrisl: rats. also, Makefile.am and configure.ac now has diffs between jbig2dec.git and ghostpdl.git as well as snprintf.c (probably due to the 'header copyright' patch going missing) | 13:09.56 |
chrisl | err, Makefile.am and configure.ac should be the same - let me check.... | 13:10.40 |
| Oh, all those memento changes missing from ghostpdl...... | 13:15.22 |
| tor8: I'm not getting any diffs in snprintf.c | 13:17.42 |
tor8 | http://ghostscript.com/~tor/stuff/jbig2dec.txt where jbig2dec is from jbig2dec.git and jbig2dec-split2 is from "git subtree split -P gs/jbig2dec" on the latest ghostpdl.git | 13:19.34 |
chrisl | Well, "diff ~/artifex/ghostpdl/gs/jbig2dec/snprintf.c ~/artifex/jbig2dec/snprintf.c" gives me no output | 13:20.52 |
kens | chrisl which version of FreeType do we have in 9.06 ? | 13:25.10 |
chrisl | kens: I think it's 2.4.9 + a couple of patches (patches which are in 2.4.10) | 13:25.55 |
kens | Thanks, I'lll send a mail to ft-devel | 13:26.36 |
chrisl | tor8: I can put Robin's memento related changes to ghostpdl now, if you want, but I still can't see any differences in snprintf.c | 13:45.11 |
Robin_Watts | chrisl: The memento changes that went into mupdf and into jbig2dec CAME from gs. | 13:50.17 |
chrisl | Huh, so why do they not appear in my ghostpdl repos, I wonder...... | 13:55.00 |
| Robin_Watts: Makefile.am is different in jbig2dec and ghostpdl/gs/jbig2dec | 14:01.45 |
Robin_Watts | because of something I did ? | 14:03.32 |
chrisl | I guess so: http://git.ghostscript.com/?p=ghostpdl.git;a=blob_plain;f=gs/jbig2dec/Makefile.am;hb=ebe054e3 | 14:04.06 |
| vs http://git.ghostscript.com/?p=jbig2dec.git;a=blob_plain;f=Makefile.am;hb=07abb8a5 | 14:04.36 |
| Unless it was me - but I thought I'd done that correctly..... | 14:05.26 |
Robin_Watts | Blame would appear to blame me. | 14:06.31 |
| back from 8th june ? | 14:06.42 |
| http://git.ghostscript.com/?p=jbig2dec.git;a=commit;f=Makefile.am;h=536620333aa7b752506626dce651ede08e2cdbc9 | 14:07.02 |
| Sorry, I thought you were talking about the memento changes I did yesterday. | 14:07.34 |
chrisl | That may have been me - I'm getting confused :-( | 14:07.47 |
| What I see is the "memento.c memento.h" addition missing in ghostpdl | 14:09.02 |
Robin_Watts | Right. I suspect I added that on the 8th june to the jbig2dec, but missed the corresponding change in gs. | 14:09.32 |
chrisl | Hmm, but I committed that patch to jbig2dec..... | 14:10.30 |
Robin_Watts | oh. | 14:10.45 |
| wow. | 14:10.47 |
| Then I'm confused too :) | 14:10.55 |
chrisl | So, I pull the patch from ghostpdl/gs/jbig2dec | 14:11.00 |
Robin_Watts | OK. My commit to ghostpdl didn't include the Makefile.am changes: http://git.ghostscript.com/?p=ghostpdl.git;a=commitdiff;h=e0fd2c6fe5cdba60e0252993319408ce8962afd3 | 14:12.27 |
chrisl | So, I wonder where I got them from - it's a kind of magic...... | 14:13.08 |
| Oh, right, you changed Makefile.in, which is a derived file - I worked out the changes in Makefile.am based on that | 14:13.53 |
| So I *do* need to pull that change into ghostpdl | 14:14.20 |
Robin_Watts | The makefile.am bit, yes. | 14:15.03 |
chrisl | Yeh, I remember it now - but I *thought* I'd included a commit comment to that effect, which seems to have gone AWOL :-( | 14:15.39 |
| Okay, so the two jbig2decs should now be up to date. | 14:30.32 |
Robin_Watts | Urm... Can anyone see where gscms_create is called ? | 15:14.07 |
| I don't think we ever call it. Which means all the cms memory is allocated through malloc/free, not through gs. | 15:17.20 |
kens | seems to be the case, I can't see it called | 15:20.26 |
Robin_Watts | which is just as well, cos gs_lcms2_realloc fails to cope with the ptr == NULL case :) | 15:27.19 |
ray_laptop | another thing to chat with mvrhel about on Tue | 15:31.30 |
| (that and getting rid of lcms tree) | 15:31.52 |
Robin_Watts | lcms 2.4 is about to be released, hence I'm looking at it again. | 15:33.29 |
ray_laptop | Robin_Watts: that would be a show stopper for cust 532 upgrade to 9.06 (if they weren't using UseFastColor and doing their own h/w based color conversion) | 16:02.49 |
| Robin_Watts: not 2.4, but using malloc/free (they don't have much malloc area) | 16:03.42 |
Robin_Watts | ray_laptop: Right. | 16:04.32 |
| I've got a hack here to fix it. | 16:04.36 |
| I suspect mvrhel can easily solve it though. | 16:05.01 |
henrys | hopefully shelly can put some time in, by coincidence his bountiable pcl bug is the same as a customer problem. I could fix it fairly quickly but he's already worked on the problem quite a bit. | 16:21.54 |
chrisl | henrys: shelly was looking for something interesting to do..... | 16:22.25 |
henrys | sorry this is a pcl problem. | 16:22.40 |
| ;-) | 16:22.55 |
chrisl | ;-) | 16:23.29 |
henrys | Robin_Watts, tor8:no word back on my end. | 16:24.29 |
Robin_Watts | I haven't heard anything either. | 16:26.06 |
henrys | also zeniko has reported more jbig2 problems he can work on I guess I'll make those bountiable. | 16:31.04 |
Robin_Watts | Personally, I'd be tempted to say "Let's get the existing stuff out of the way before we start work on new stuff". | 16:31.36 |
henrys | yeah actually the pcl bug he's working now has become p1 in our usual grading system. | 16:33.24 |
| not really feeling a p1 bounty is in order though, I've helped out a lot with it so I'm going to leave it. | 16:34.51 |
| going to the pro cycling challenge stage here in boulder Saturday. The ride they're doing is absolutely incredible, I've done pieces of it at a time, can't imagine doing it all in a single day. | 16:37.34 |
Robin_Watts | henrys: Isn't that a bit like going to see the Olympic Torch being carried past? | 16:41.30 |
| You wait for hours by the side of the road behind a mounting crowd, and then suddenly a rush of people zoom past on bikes, and 30 seconds later it's all over and you get to go home ? | 16:42.05 |
| I caught some bike racing on TV while in Malta, and the coverage there was probably far more interesting, cos the camera moves with the bikes, and you can see them jostling for position, suddenly making a burst for it etc. | 16:43.28 |
henrys | no a group of us are riding to strategic points along the race so there will be several 30 second meetings | 16:44.00 |
Robin_Watts | Right. | 16:44.09 |
| I love F1, but I'm reluctant to go see it live because you can guarantee that you'll never follow the action from the 100 yards of the track you can see. | 16:44.34 |
henrys | it is a big thing here though. The crowds are like the Tour | 16:44.35 |
Robin_Watts | I'm sure. I think the bike racing I was watching was from Spain or Italy, and it's similarly huge there. | 16:45.28 |
| A mail from zeniko. | 16:46.11 |
chrisl | Robin_Watts: vuelta a espana is on just now | 16:46.29 |
henrys | Robin_Watts:I don't understand why our arrangment differs from the google CLA I've never actually read it. | 16:49.48 |
Robin_Watts | Miles' form is longer :) | 16:50.03 |
henrys | upon reading his side of it though I'm much more sympathetic to his position. | 16:51.42 |
Robin_Watts | I think we can explain that we as a company first and foremost make our living selling it commercially. Therefore nothing we do (no patches we take etc) can limit our scope to do that. Having said that, we choose to make it available as open source, because we feel that's useful. | 16:53.42 |
| If he wants to make patches available to us then we'll gladly take them (bounty or not), as long as we are free to use them as we need to. | 16:54.29 |
| If he's helping up with areas that are of use to the commercial development of it, then we're happy to thank him for that with bug bounties. | 16:55.05 |
henrys | yes but there must be something in the wording of our license that he find problematic. The google CLA covers everything you just said. | 16:55.55 |
Robin_Watts | I suspect it's just the fact that our CLA is much longer than googles. | 16:56.26 |
henrys | http://code.google.com/legal/individual-cla-v1.0.html | 16:56.29 |
Robin_Watts | googles is 3 paragraphs or something. | 16:56.34 |
| ok. 8 paragraphs :) | 16:56.50 |
henrys | well I'm trying to post mortem this thing. It seems that with proper communication none of it would have happened, I'd like to prevent repeating the exercise. | 16:59.25 |
Robin_Watts | yeah. I thought we had it all sorted out in March (after a day of being very worried). | 17:01.29 |
sebras | but it is still the case that patches must go through bugzilla..? | 17:03.36 |
Robin_Watts | sebras: Unless we get a blanket statement from him that he's happy for us to take any and all patched from SumatraPDF.patch, I feel we ought to be only taking patches that go via bugzilla with an explicit statement attached. | 17:04.43 |
sebras | ok. | 17:05.26 |
Robin_Watts | (Maybe you could create bugs, and put the patch on, and just have him put a note on saying that it's OK to take that patch, but I feel that we probably need a statement from him on record, I think?) | 17:05.32 |
sebras | Robin_Watts: seems sensible. | 17:06.20 |
kens | Heading off, goodnight everyone. | 17:06.39 |
Robin_Watts | Night kens. | 17:06.52 |
| Have a good weekend. | 17:06.58 |
ray_laptop | Robin_Watts: offhand, do you know if I can cluster test just gs and xps ? | 17:20.08 |
| since my change is with transparency, I only need those two | 17:20.37 |
| Robin_Watts: nm. I'll just start with gs for now | 17:21.27 |
chrisl | ray_laptop: you should be able to do "clusterpush.pl gs xps", I think | 17:22.03 |
Robin_Watts | ray_laptop: What chrisl said, sorry. | 17:23.05 |
ray_laptop | thanks to both of you. I just decided to do gs for this run (first time with pdf14 clist optimization trick) | 17:30.25 |
| now that cust 532 is going to 9.06, they are going to need it | 17:30.55 |
| clusterpush is sure an easy way to test linux builds ;-) | 17:41.06 |
| darn. git problem and my commit didn't make it correctly up to cluster. | 17:48.16 |
| looks better now. Time to head over to cust 532.... bbiaw | 17:53.45 |
PEPPERON1 | hello all, what the difference between invoking ghostscript vs gs on the command line? | 18:27.13 |
| what is | 18:27.17 |
sebras | PEPPERON1: ghostscript is a symbolic link to gs, so most likely -- no difference. | 18:28.15 |
| PEPPERON1: I'm assuming you're on Linux. I don't run windows so it may not be true there. | 18:28.48 |
PEPPERON1 | sebras: thanks! we had some weird environment issues with cron | 20:56.26 |
| hence the confusion | 20:56.29 |
sebras | PEPPERON1: not sure I get the connection between gs and cron, but I'm happy you sorted out the problem. | 20:58.14 |
PEPPERON1 | sebras: we had a package from lenny and also a custom build of ghostscript | 20:58.41 |
| cron was using the lenny version, but bash was using our build | 20:58.54 |
| it was the environment all along | 20:58.59 |
| Forward 1 day (to 2012/08/25)>>> | |