| <<<Back 1 day (to 2011/07/11) | 2011/07/12 |
mvrhel2 | ray_droid: neat | 00:55.46 |
| we just first reading the half blood prince and that should be arriving here via netflix in a couple days | 00:56.22 |
| and we just bought the last book today | 00:56.33 |
| henrys: are you there? | 05:18.29 |
Robin_Watts | tor8: Let me make you a patch of what I have so far. | 10:46.52 |
tor8 | mroning robin | 10:47.04 |
Robin_Watts | morning | 10:47.10 |
| http://ghostscript.com/~robin/0001.patch | 10:50.42 |
tor8 | wrong patch I think | 10:51.11 |
kens | It isn't mupdf :-) | 10:51.27 |
Robin_Watts | ahem : != / | 10:51.38 |
| Try again. | 10:51.40 |
| I noticed that in pdf_show_path we make a group and then throw it away even for non stroke, non fill paths. | 10:52.26 |
| (i.e. clip paths) | 10:52.29 |
| I get rid of that, which saves some time. | 10:52.45 |
| Also, when we do fz_new_pixmap, we calloc them. Then we always call fz_clear_pixmap. | 10:53.16 |
tor8 | ah, right, good catch (non-painting paths causing pdf_begin_group) | 10:53.43 |
| our calloc is (misleadingly) not the same as system calloc (it's counted-malloc, not clear-malloc) | 10:54.29 |
Robin_Watts | Ah, ok. | 10:54.47 |
tor8 | though I guess we could just clear all memory when we allocate it, though I prefer not to since that'll force me to be more careful about initializing memory properly | 10:55.26 |
Robin_Watts | No, Unnecessary clearing of memory is bad. | 10:55.45 |
| I made the temporary groups we push in order to do blending be isolated ones. | 10:56.13 |
| and the page level one too. | 10:56.19 |
| (saves expensive blending back) | 10:56.32 |
| This way, we should only be keeping the extra shape plane (and hence incurring the extra expense) when we're really in non-isolated transparency groups. | 10:57.21 |
| But the equations for blending back with the shape are wrong, and it's hurting my head trying to figure the right ones out. | 10:58.01 |
tor8 | I tried to use the gs uncomposite blend voodoo naively once, but that failed even more spectacularly | 10:59.09 |
| zeniko mentioned he did something to support non-isolated blends with the "gdi+" device in sumatrapdf | 10:59.36 |
| I couldn't make heads or tails of that code though :( | 10:59.46 |
| odd, chrome keeps giving me the old non-mupdf 0001.patch but I can see it fine if ssh into casper :/ | 11:01.09 |
| robin_watts: the patch looks good to me so far | 11:05.59 |
| hm, we clobber the edges in the edge list. I was thinking a fz_rewind_gel would've been handy | 11:09.10 |
Robin_Watts | Yes, that would make it cheaper to render to two targets. | 11:14.28 |
tor8 | well, non-isolated blend modes are pretty rare in the wild so I'm not too worried about performance. keeping the code maintainable is a more pressing issue with the pdf 1.4 madness. | 11:17.01 |
sebras | tor8: I assume that you fished the isolated/knockout and the non-versions pdfs out from the adobe-testsuite? they should be there at least. I guess it would make sense to add them to the insane testsuite... | 11:49.37 |
tor8 | sebras: http://ghostscript.com/~robin/out.png | 11:50.50 |
sebras | tor8: ah.., I remember that file... | 11:51.22 |
tor8 | :) | 11:51.29 |
sebras | tor8: cool, then I can't really help you more until later when I arrive at your place to disturb.. eh.. help you. :) | 11:52.03 |
tor8 | you should go to london instead, I'm only grappling with the ghostscript+mupdf interpreter initialization madness :( | 11:52.46 |
kens | London wouldn't help ;-) | 11:53.03 |
sebras | kens: where have you hidden Robin? | 11:53.31 |
kens | Robin is 'near' Oxford | 11:53.43 |
| For sufficiently large values of near | 11:54.09 |
Robin_Watts | :) | 11:54.13 |
sebras | :) | 11:54.18 |
Robin_Watts | Morning mvrhel2 | 13:46.14 |
mvrhel2 | good morning | 13:46.24 |
| So I won't be around at 9am but will be back around 10am . Is there anything that anyone who is going to be gone by then needs from me? | 13:47.24 |
Robin_Watts | may be ready to come crawling to mvrhel2 for an explanation of blending with shapes by then. | 13:48.21 |
| but I'll be around after 10am. | 13:48.31 |
mvrhel2 | ok. too early in the morning for that sort of thing anyway :) | 13:48.53 |
Robin_Watts | Too early in the year if you ask me :) | 13:49.06 |
mvrhel2 | :) | 13:49.13 |
henrys | mvrhel2:woops missed you last night. | 14:25.36 |
chrisl | henrys: I'm making the changes so that Luratech can be used for Ghostscript without hand editing makefiles. Just to check: if the Luratech directories exist (under the GS directory) we want people to have to *opt out* of using Luratech? | 14:55.11 |
Robin_Watts | chrisl: Seems reasonable. If you have luratech, why wouldn't you use it ? | 14:55.48 |
chrisl | Well, the only people it might cause problems for (in general) would be us - remembering to opt out of Luratech when we want to. | 14:56.39 |
tor8 | ugh. jpeg library version mismatch hell in gs/mupdf | 14:56.55 |
henrys | chrisl:right that's what we decided. The arrangement is rather awkward. | 14:57.08 |
chrisl | tor8: what jpeg library version were you expecting? | 14:57.48 |
Robin_Watts | chrisl: Personally, I'd probably keep luratech somewhere else on my harddisc and symlink/junction it in as required. | 14:57.54 |
tor8 | fitz uses one, gs uses another, in the same static builds :/ | 14:58.13 |
mvrhel2 | henrys: no problem | 14:58.20 |
| I was just wanting to let you know that I am going to miss the meeting | 14:58.33 |
chrisl | tor8: ah, can't help then! | 14:58.36 |
tor8 | meaning I've got two copies of libjpeg around... | 14:58.40 |
chrisl | Robin_Watts: that's what I would do (use symlinks) but Ray wanted the opt out there. | 14:59.02 |
henrys | maybe the directory should be there and we should tell folks in the newsletter to enable it and make the default off. then the free release can be the same as the commercial? | 14:59.06 |
tor8 | i've been doing the gs/mupdf bridge by copying the object files from another build so I don't have to mess with the build system for getting things up and running | 14:59.10 |
Robin_Watts | henrys: chrisl: I think having the option to disable it in the makefile is smart. | 14:59.40 |
| that way people can choose to include/exclude the directory or edit the makefile as they see fit. | 15:00.14 |
chrisl | henrys: (almost) no one will read such technical stuff, and I though the idea was to *ensure* commercial customers used Luratech. | 15:00.34 |
Robin_Watts | henrys: I think most customers don't read the newsletter or just take the prebuilt binaries. | 15:00.41 |
chrisl | s/though/thought | 15:00.42 |
henrys | okay then as you had it. | 15:01.05 |
chrisl | As Robin_Watts says, if you have Luratech, why on earth would you not use it (except for test/debug stuff) | 15:01.46 |
henrys | Robin_Watts:I think prebuilt binaries are a minority. | 15:01.49 |
| chrisl:I was trying to keep the commercial and public releases as close as possible to avoid maintenance errors and problems. | 15:02.39 |
chrisl | henrys: I'm really not fussy either way, I just want to avoid, well exactly this conversation, after I've committed the changes. | 15:03.49 |
henrys | I'm fine with what you said before my suggestion. | 15:04.19 |
Robin_Watts | right, so we have exactly the same makefile in commercial and free releases, and it just behaves differently (by default) according to whether the Luratech directory is there. Sounds smart. | 15:04.52 |
henrys | I know how to do that in configure.ac but not on windows. | 15:05.46 |
chrisl | nmake you can have conditional rules based on whether a file/directory exists or not | 15:06.15 |
henrys | oh nvm - | 15:06.17 |
| right | 15:06.21 |
chrisl | This kind of thing worries me a bit, because it's great for those who "blindly" build as we tell, but they might confuse the more technically involved folk. | 15:07.50 |
henrys | yeah that's why I was liking my suggestion, but I don't feel strongly about it. | 15:08.58 |
| not why but another reason. | 15:09.19 |
Robin_Watts | "more technically involved" = "more able to cope with confusion" | 15:09.39 |
chrisl | Robin_Watts: your faith is inspiring, but I fear misplaced..... | 15:10.03 |
Robin_Watts | You want to argue that "those who blindly build" would be better placed to cope ? | 15:10.29 |
henrys | IMHO if we can just respond to all these bugs and performance problems with enable luratech word will spread quickly. | 15:10.41 |
chrisl | Robin_Watts: no, more that the "more technically involved" are often not *much* better placed to cope than the "blind builders". | 15:11.51 |
| henrys: I'll hold off committing until after the meeting, if anyone feels strongly, they can speak then. | 15:12.34 |
Robin_Watts | mvrhel2: You still here? | 15:13.30 |
henrys | chrisl:teach you to ask folks' opinions - next time better just do it and say it is too much to change back ;-) | 15:14.16 |
Robin_Watts | I'm looking in gxblend.c for the routine used for non-knockout, non-isolated blends, and I can't see it. I can see the other 3 cases though. Am I missing something? | 15:14.29 |
chrisl | henrys: the only reason I didn't do that is because this is for the commercial release - I feel safer getting feedback when it's for paying customers! | 15:15.32 |
henrys | alexcher I am still baffled about 692309 why can't kens reproduce it? are you going to analyze this bug? | 15:20.58 |
kens | henrys I can try again to reproduce it if you like, but it didn't seem worht it after Alex said he could. | 15:21.27 |
alexcher | henrys: The main problem is that normal operation of gs depends on the hard-coded default. | 15:23.00 |
henrys | well mvrhel2 said he tested compile inits 0 shouldn't it always fail? | 15:23.39 |
| I can't imagine why but could this be just postscript and michael tested pdf? | 15:24.26 |
| marcosw:are you doing compile inits 0 testing? | 15:25.23 |
alexcher | henrys: The error happens during the start-up. gs has no chance to run any external files. | 15:25.32 |
| henrys: We to make the hardcoded default an to point to a non-existing directory and check how does it work. | 15:28.02 |
henrys | how would your command line in comment 15 find the iccprofiles wihtout setting the profile directory? | 15:28.24 |
alexcher | henrys: The start-up code uses search path to find the directory and passes it to the interpreter as a user parameter. | 15:30.56 |
henrys | does the job work with -sICCProfilesDir=/home/alexcher/gs/iccprofiles | 15:35.34 |
| ? | 15:35.40 |
| the docs say you need a closing delimiter. | 15:37.23 |
alexcher | henrys: Probably, not. I need to re-test this. | 15:37.37 |
henrys | fairly important I don't think we can do a release with compile inits 0 completely broken. | 15:38.11 |
marcosw | henrys: I haven't started the compile inits 0 testing yet. I've been looking at bitmaps for the 9.03 release and also the luratech decoders. | 15:47.46 |
henrys | marcosw:okay thanks. | 15:48.31 |
marcosw | henrys: I need to leave at quarter after today, so I'll miss the 2nd half of the meeting. | 15:53.14 |
henrys | okay | 15:53.42 |
alexcher | henrys: -sICCProfileDir doesn't help whether it has the trailing separator or not. | 15:53.59 |
ray_laptop | morning, all | 15:54.00 |
henrys | alexcher:so compile inits 0 does not work at all | 15:54.26 |
alexcher | henrys: Yes | 15:54.40 |
henrys | hi ray_laptop | 15:55.38 |
ray_laptop | I was amused by tor's comment that mupdf has a calloc that isn't the real calloc. I used to shoot (at least as far as their job went) for that kind of stuff | 15:56.23 |
Robin_Watts | ray_laptop: It does seem... suboptimal, yes. | 15:57.12 |
marcosw | henrys, et.al.: I just added weekly regression testing compile inits 0 (better late then never :-) | 15:57.53 |
Robin_Watts | Yes! My first blend looks perfect. | 15:57.55 |
| Subsequent ones look knackered though :( | 15:58.01 |
henrys | marcosw:thanks | 15:58.08 |
ray_laptop | I also saw a comment about the closing delimiter (or not) on the paths. We really _should_ be forgiving about that (IMHO) | 15:58.13 |
| even if it technically violates the letter of Adobe's spec for GenericResourceDir | 15:58.43 |
kens | Closing delimiter seems to cause problems if present on -I and if not present on -sGenericResourceDir. We should be more flexible. | 15:59.00 |
Robin_Watts | At the very least we should be consistent. | 15:59.27 |
| But being forgiving on top of that seems sensible. | 15:59.46 |
ray_laptop | we also need to make sure that the ICCProfilesDir (and GenericResourceDir) paths are exempt from the file access protection (at least the command line settings, that is) | 15:59.53 |
henrys | so we'll start the meeting with alexcher's showstopper (at least for me) don't think we should ship with compile inits broken | 16:00.51 |
ray_laptop | the linux community always uses SAFER for almost all of the scripts and by most of the other tools (printing chain and imagemagick) | 16:01.08 |
henrys | alexcher generally if you have a bug like that on your desk you need to bother people more until you get some results. | 16:01.59 |
ray_laptop | I agree. Particularly for the oob customer release. I know for some of their builds, they don't use the romfs | 16:02.04 |
Robin_Watts | What needs to be done to fix it? Is it just a question of coding, or are there design decisions that need to be sorted first ? | 16:02.42 |
ray_laptop | if there is a -dSAFER issue, that won't affect the OOB customer | 16:02.46 |
henrys | Robin_Watts:I don't think we know, alexcher? | 16:03.16 |
ray_laptop | first, please, which bug # ? | 16:03.28 |
henrys | any other release stoppers that we know of? | 16:03.33 |
Robin_Watts | 692309, I believe | 16:03.44 |
henrys | 692309 | 16:03.46 |
Robin_Watts | I am not aware of any other showstoppers. | 16:04.17 |
marcosw | do we want to start doing -dSAFER regression testing as well? | 16:04.24 |
henrys | okay so other than that we can ship. | 16:04.43 |
| right? | 16:04.46 |
marcosw | the bitmap comparision of 9.03 vs 9.02 went well. There were some progressions but nothing else. | 16:04.58 |
henrys | alexcher:any device or just pdfwrite? | 16:05.01 |
| marcosw:great | 16:05.10 |
alexcher | henrys: Any device | 16:05.23 |
Robin_Watts | alexcher: What needs to be done to fix 692309? Is it just a question of coding, or are there design decisions that need to be sorted first ? | 16:05.44 |
chrisl | So does anyone have strong feelings about automatically using luratech if the directories exist (for both configure and Windows builds), with an *opt out* option? | 16:05.45 |
ray_laptop | alexcher: what path is it trying to open (strace or debug result) | 16:05.50 |
henrys | so GenericResourceDir has gs/iccprofiles on it by default? Hard to believe. | 16:05.53 |
ray_laptop | henrys: the spec is that if someone doesn't set ICCProfilesDir then we default to a sibling of the 'Resource' directory | 16:06.50 |
henrys | oh I didn't see that. | 16:07.19 |
| sorry | 16:07.21 |
alexcher | ray_laptop: gs tries to opoen %rom% | 16:07.35 |
ray_laptop | I don't know if we do, but we should search for an iccprofiles directory that is the sibling of ANY Resource directory on the LIBPATH list (IMHO) | 16:07.40 |
Robin_Watts | chrisl: Can we have a USE_LURATECH option, that users can set to 0 or 1 specifically? If unspecified by the user, set it to 1 if the directories exist, 0 otherwise ? | 16:08.12 |
ray_laptop | alexcher: it tries BEFORE failing to find it on the disk using the profile_dir ? | 16:08.22 |
| it is supposed to use the %rom%iccprofiles ONLY if it can't find one on the profile_dir (ICCProfilesDir) | 16:08.55 |
henrys | marcosw:I sent you mail about more stuff to do with that bug from 711. | 16:08.56 |
alexcher | Robin_Watts: Probably, interaction of the device and imager state should be clarified and documented. | 16:09.16 |
henrys | I've had relatives here and was sort of working a little at a time. | 16:09.17 |
chrisl | Robin_Watts: we'll have --with(out)-luratech on configure, for Windows I've kept the existing options, just added the "use it if it's there" logic | 16:10.10 |
marcosw | henrys: okay | 16:10.11 |
kens | I haev some solar panel engineers on our roof, I will be in and out for a while. | 16:10.11 |
ray_laptop | chrisl: as long as the customer release defaults to using luratech, I don't care if it is a manual edit for customers that don't want luratech | 16:10.19 |
henrys | alexcher:I guess we can postpone the discussion until michael gets here. | 16:10.25 |
Robin_Watts | kens: You too? | 16:10.27 |
| 9 of my 21 panels are up :) | 16:10.45 |
henrys | I didn't think you had enough sun for that. | 16:10.47 |
chrisl | ray_laptop: that's how I've got it, yes. | 16:10.54 |
henrys | kens:anything for the meeting? | 16:11.14 |
ray_laptop | chrisl: as long as we document how. But that may mean that the document may need to be different for customers and non-customers (but I hope not) | 16:11.17 |
Robin_Watts | chrisl: OK, sounds good. | 16:11.18 |
chrisl | ray_laptop: I don't see why the doc should be different: the UFST is covered in docs we always ship. | 16:11.53 |
ray_laptop | michael will be back in about an hour | 16:12.13 |
| chrisl: but the default doesn't differ for UFST. | 16:12.47 |
| chrisl: UFST customers are definitely the exception and usually get hand holding (since they are embedded printer customers) | 16:13.20 |
henrys | I do like my earlier idea more and more - just send the directory and tell folks how to enable it. | 16:13.25 |
| default should be off. | 16:13.34 |
ray_laptop | henrys: i thought you wanted luratech to be the default for customers ? | 16:13.52 |
| that was pretty clear to me previously | 16:14.09 |
ray_laptop | or I thought I had it straight | 16:14.28 |
henrys | I did now I don't but it is up to chrisl I talked him about this before. | 16:14.56 |
chrisl | me too, but then I opened my big mouth, and....................... | 16:14.56 |
marcosw | is off to school, I'll read the irc transcript later today | 16:15.03 |
henrys | bye marcosw | 16:15.10 |
ray_laptop | and a few customers use our pre-built binaries which I think _should_ have luratech. | 16:15.17 |
henrys | okay enable it. | 16:15.51 |
chrisl | Okay, topic closed! | 16:16.09 |
henrys | tor8 anything on the mupdf parser? | 16:16.14 |
ray_laptop | henrys: it's easier for support to have luratech ON for most customers (since we don't have to ask). Unless we make it the default, many customers won't use it. | 16:16.27 |
| darn internet connection ! | 16:16.33 |
tor8 | henrys: I hate the build system. other than that, I've been making progress :) | 16:16.48 |
chrisl | ray_laptop: for the docs, I think we just need to be clear about the two situations. | 16:17.07 |
ray_laptop | I would REALLY like to have luratech be the default build if we are going to ship it. It improves performance a LOT ! | 16:17.23 |
chrisl | ray_laptop: it is going to be the default - that topic is *closed*! | 16:17.57 |
henrys | chrisl has closed the topic | 16:18.02 |
| we're good. | 16:18.06 |
ray_laptop | chrisl: great! I agree that the topic is closed (since you agree with me) ;-) | 16:18.28 |
henrys | and chrisl did you see that regression I gave you that one surprised me. | 16:18.36 |
chrisl | henrys: I've got a fix for it, yes. Although I don't a memory use difference (to speak of). | 16:19.12 |
ray_laptop | Robin_Watts: 21 panels -- what is that, about a 3.5Kw (nominal) system ? | 16:19.31 |
chrisl | henrys: I don't see a memory usage difference, I mean | 16:19.39 |
Robin_Watts | 3.885 (185W per panel) | 16:19.45 |
henrys | I was not able to reproduce the memory problem either. | 16:19.56 |
chrisl | I knew there was a performance difference, but in all honesty, I'm surprised it's measurable. | 16:20.26 |
ray_laptop | BTW, what bug is that ? | 16:20.41 |
henrys | Robin_Watts:do you have some sort of government subsidy I can't imagine that economical. | 16:20.48 |
chrisl | ray_laptop: the performance one? Bug 692338 | 16:21.19 |
Robin_Watts | henrys: Yes. I get paid for every unit I generate, whether I use it or not. | 16:21.22 |
ray_laptop | here in socal, the tax and state rebates would let me recoup a 4Kw system in about 5 years (at my current usage and bill) | 16:21.49 |
kens | Robin_Watts : solar hot water, not PV. | 16:22.29 |
ray_laptop | chrisl: and the memory usage bug (that you can't reproduce) ? | 16:22.37 |
Robin_Watts | Unit = 1KWh. Typical price per unit (for me to buy) = 15p or so. I get paid 45p or so for every unit I generate. | 16:23.00 |
henrys | ray_laptop:I've asked marcosw to look at that and assign it to me if he fined anything - see your email. | 16:23.21 |
Robin_Watts | So I should pay off a 4KW system within 10 years, and the subsidy is guaranteed for 25 (tax free) | 16:23.24 |
henrys | s/fined/finds | 16:23.30 |
chrisl | ray_laptop: I don't think the memory usage issue has a bug yet | 16:23.35 |
| oh, beaten by henrys | 16:23.41 |
Robin_Watts | kens: Ah. I'm considering that for next year (rumour is they are going to do a feed in tariff for that too) | 16:23.55 |
kens | Hmm, we had it here when we moved in, but it broke this summer (perect timing) | 16:24.18 |
ray_laptop | the performance difference isn't nearly what the customer reports (henry reports 1.5x the customer claimed 2x). Do we need to ask the customer for profiles ? | 16:24.30 |
kens | It seems to be working again now thouhg, since the engineers stopped crawling over the roof. | 16:24.40 |
henrys | wind seems to be more popular in my area that solar I haven't looked at the financial incentives. | 16:24.58 |
| s/that/than | 16:25.05 |
chrisl | Robin_Watts: what's the git command to get a full list commits that exist in your repos (rather than just the log)? | 16:26.07 |
henrys | ray_laptop:yeah I was going to measure after chrisl checked in there's another problem lurking. | 16:26.17 |
Robin_Watts | henrys: wind is evil (on land) | 16:26.32 |
| kills bats, looks ugly, doesn't really work. | 16:26.52 |
| chrisl: git log --all ? Or git reflog ? | 16:27.27 |
ray_laptop | Robin_Watts: works better at night than solar :-) | 16:27.34 |
henrys | I'm scratching my head about solar in the UK? | 16:27.41 |
chrisl | Robin_Watts: reflog was what I was after, cheers! | 16:27.47 |
kens | henrys why ? we do get sunshine. | 16:28.01 |
| And it doesn't need to be brilliant sun either | 16:28.13 |
Robin_Watts | henrys: Solar doesn't need 'sunshine', it just needs 'daylight'. | 16:28.19 |
kens | brb | 16:28.41 |
Robin_Watts | It's not quite as much of a no brainer as it is when you move further south, true, but the government schemes push it back into no brainer territory. | 16:29.08 |
henrys | the amount of energy per square meter does matter. solar is much more effective further south. | 16:29.40 |
chrisl | henrys: the FAPI fix is committed and pushed. | 16:29.56 |
Robin_Watts | kens: http://www.rhincentive.co.uk/RHI/quick/ | 16:30.15 |
| Dunno if you can register given you're already installed. | 16:30.28 |
ray_laptop | did any of you hear about the IR (frequency antenna + diode) power collector that gets >85% efficiency >5x for the same area as PV | 16:30.30 |
Robin_Watts | ray_laptop: No... | 16:30.58 |
henrys | we have a huge wind operation here - and with the winds coming off the mountains I have to think it gets some use - we frequently have 30-40 MPH winds. | 16:31.06 |
ray_laptop | it was in a magazine I get from my alma mater (Mizzou Engineering). There's a guy there working with folks at MIT and other research places on it | 16:32.12 |
Robin_Watts | henrys: Yes, clearly it matters to the bottom line of "how much power you're generating". My point is that looking at it on a "return for money invested", it's the best long term investment out there at the moment, given the government scheme. | 16:32.25 |
| henrys: wind generation has problems with winds being too fast as well as too slow. | 16:32.53 |
| You need consistent winds of a speed within the right range. | 16:33.11 |
ray_laptop | Robin_Watts: I would think that bats could easily avoid the relatively slowly spinning blades | 16:33.22 |
Robin_Watts | ray_laptop: Its the low pressure area that the fans produce. | 16:33.40 |
| Bats die when they fly through it due to barotrauma. Their lungs explode :( | 16:34.00 |
ray_laptop | Robin_Watts: most of the wind systems out here have variable pitch blades to cope with variable wind speed | 16:34.23 |
Robin_Watts | Bats are HUGELY important for crop fertilisation, and pest control. | 16:34.57 |
| They are already massively endangered due to shrinking habitats, and a horrible fungal infection that's wiping them out worldwide. | 16:35.39 |
| so land based wind power really seems like a bad idea to me. | 16:35.57 |
chrisl | henrys: I was planning to "hide" this pending release from the normal scheduled releases, so the 9.04 changelog will go straight from 9.02->9.04, and I wasn't going to commit the 9.03 doc changes to trunk - is that okay? | 16:36.07 |
Robin_Watts | chrisl: That sounds wrong to me. | 16:36.34 |
ray_laptop | what are the bats doing, holding their breath ? with their breath rate i would think that they are always breathing in or out (I assume that this has been studied, however) Just doesn't seem obvious, but that's what science is for. | 16:36.49 |
Robin_Watts | ray_laptop: http://news.bbc.co.uk/1/hi/sci/tech/7581990.stm | 16:37.28 |
chrisl | Robin_Watts: well, including mention of a release that only one customer will (very likely) ever see, doesn't seem right to me - lesser of two evils? | 16:37.32 |
Robin_Watts | chrisl: We're Artifex, right? We do everything in the open. | 16:37.52 |
ray_laptop | Robin_Watts: what's wrong with "hiding" the release. It'll make it more clear for the mainstream what the changes have been since 9.02 | 16:37.59 |
Robin_Watts | By all means label it as an "interim" release, but hiding it seems wrong. | 16:38.28 |
chrisl | Robin_Watts: well, except risky, unplanned releases that we don't really want people to use. | 16:38.32 |
ray_laptop | otherwise folks that don't get 9.03 won't see the bulk of the changes since 9.02 | 16:38.34 |
Robin_Watts | ray_laptop: In the newsletter, we can document all the changes since 9.02 | 16:39.00 |
chrisl | Robin_Watts: I was going to document the reason for the apparently skipped number, though. | 16:39.16 |
ray_laptop | by 'hiding' I assume you mean not mentioning it in the 9.04 release notes and skipping it in the Changes.htm, etc. | 16:39.17 |
kens | Robin_Watts : I doubt we can register for the feed-in tariff, sadly. | 16:40.00 |
ray_laptop | chrisl: I agree with skipping it in the 9.04 release docs. Most users/customers don't need to know or care about 9.03 | 16:40.26 |
Robin_Watts | kens: well, leave it broken until next year, then get it repaired and claim it as a new installation :) | 16:40.26 |
chrisl | ray_laptop: basically yes. As I say, I'd explain why we appear to have skipped a version, but for the GPL release, and scheduled commercial release, the changelog would list changes as most people will see them - from 9.02 to 9.04. | 16:40.26 |
ray_laptop | chrisl: I agree! | 16:40.40 |
kens | Robin_Watts : Too late! Fixed.... | 16:40.42 |
Robin_Watts | I still think we should check in 9.03 as normal. | 16:40.56 |
henrys | chrisl:so you are proposing shipping with compile inits broken I don't think that a good idea. | 16:41.16 |
Robin_Watts | but hey, if the majority thinks otherwise... | 16:41.23 |
chrisl | Robin_Watts: that means making it a full release, which we agreed it wouldn't be. | 16:41.32 |
ray_laptop | Robin_Watts: the problem with the 'normal' release process is that it stages the 'deltas since last release' | 16:41.40 |
Robin_Watts | chrisl: did we? | 16:41.55 |
kens | henrys if we don't ship with compile_inits broken, we won't be shipping the customer's release until teh regular date anyway ;-) | 16:41.59 |
ray_laptop | chrisl and I did ;-) | 16:42.07 |
chrisl | Robin_Watts: it was discussed during an IRC meeting, yes. | 16:42.24 |
Robin_Watts | OK, I clearly spaced on that. Fair enough. | 16:42.47 |
ray_laptop | we don't know (yet) what's broken about COMPILE_INITS=1 do we ? Does is work with -sICCProfileDir=____ on the command line, for instance ? | 16:43.07 |
| s/ICCProfileDir/ICCProfilesDir/ | 16:43.24 |
kens | COMPILE_INITS=1 is OK do you mena 0 ? | 16:43.29 |
henrys | ray_laptop:I asked alexcher and he said it does not work. | 16:43.32 |
ray_laptop | I meant 1, yes | 16:43.38 |
ray_laptop | goes to try a couple of variations ... | 16:43.55 |
| which requires rebuilding ... | 16:44.45 |
chrisl | henrys: I don't think we should ship with this bug, no, especially as ray_laptop seemed to think that the customer in question uses COMPILE_INITS=0 | 16:45.19 |
ray_laptop | I meant 0 (darn!!) | 16:45.29 |
| I know they did for at least some of the builds they documented to us | 16:45.55 |
| at least I have it right on the command line for the version I'm building | 16:46.36 |
Robin_Watts | kens: Was the system installed since 15th July 2009 ? | 16:48.36 |
kens | No, before that. | 16:48.44 |
| *well* before that | 16:48.51 |
Robin_Watts | ah well. | 16:48.52 |
henrys | chrisl:is everything okay with shelly I told him to check in with you if he got stuck. | 16:52.45 |
| ? | 16:52.46 |
kens | He's asked a couple of questions | 16:52.58 |
chrisl | henrys: so far, yes, he seems to be making some progress. I've answered a couple things from him, and worked him through a couple of git subtleties. | 16:53.59 |
ray_laptop | he seems to have made progress on bug scrubbing :-) | 16:54.52 |
chrisl | I was kind of hoping more would be fixed, though! | 16:57.08 |
henrys | me too | 16:58.00 |
| ray_laptop:have you been able to work around the compile inits 0 problem? | 16:58.41 |
| Robin_Watts:looks like somebody is working on it, you just missed the conference ... http://www.batsandwind.org/ | 17:01.12 |
ray_laptop | henrys: well, I see at least one thing that's easy to fix that is related to my recent nulldevice change, but I need to go back to the 9.03 branch if we are talking about the OOB blocker. | 17:05.59 |
| I want to fix the issue on the master branch first, however | 17:06.17 |
henrys | the blocker is the same on the master branch it just heeds the error return and fails gracefully. | 17:09.42 |
chrisl | ray_laptop: just fix it on master, and mail me the SHA1 of the commit, I'll cherry-pick it into the 9.03 branch for you. | 17:10.32 |
ray_laptop | chrisl: That change isn't on 9.03, AFAIK. That'll be a separate exercise. | 17:12.01 |
chrisl | ray_laptop: oh, okay. Just the kind of nonsense we wanted to avoid :-( | 17:12.58 |
| henrys: is the Openjpeg integration project dead/dying/in a coma? | 17:14.51 |
henrys | ask alexcher | 17:15.12 |
| I thought it was close | 17:15.32 |
chrisl | Oh, yes, sorry, I forgot out-sourcing it didn't work out. | 17:16.08 |
| alexcher: does openjpeg look like it's going to work to replace jasper? I only ask because there's a build bug open about autoconf selecting which JPX decoder to use, and I want to know whether to leave it open. | 17:17.35 |
alexcher | chrisl: openjpeg won't be ready for v. 9.03 but I'll try to complete it for 9.04. | 17:19.09 |
chrisl | alexcher: okay, great. I'm not rushing you, as said, just whether to leave the bug open to remind me to add option when it's available. | 17:20.13 |
ray_laptop | henrys: OK, have a fix for COMPILE_INITS=1 on head that seems to work. I'll commit after a clusterpush (unless someone is in a rush), then switch to 9.03 and look into that.... | 17:27.55 |
| Note that I did NOT fix the trailing delimiter (yet) | 17:28.42 |
Robin_Watts | thinks that the inability to tell 0 from 1 is probably bad for a computer engineer :) | 17:29.22 |
henrys | alexcher's bug is COMPILE_INITS=0 | 17:29.23 |
| Robin_Watts:I think we have an interview question for that ;-) | 17:30.19 |
ray_laptop | hehe | 17:34.32 |
| 0 was a recent innovation | 17:35.04 |
Robin_Watts | Someone once told me you could spot computer engineers by asking them to count things. If they start from 0... | 17:35.07 |
ray_laptop | when I started programming we used roman numerals ;-) | 17:35.31 |
chrisl | Robin_Watts: what's the Windows command prompt replacement that you use? | 17:35.47 |
Robin_Watts | Console 2. | 17:35.57 |
| http://sourceforge.net/projects/console/ | 17:36.14 |
| http://sourceforge.net/projects/console/files/console-devel/2.00/ | 17:36.33 |
ray_laptop | chrisl: for the commercial release script, you need msysgit on windows | 17:36.44 |
chrisl | Cheers - I'll give it a try. I just about chucked my laptop out the window just now, because of the stock cmd.exe :-( | 17:36.47 |
| ray_laptop: will cygwin do? | 17:36.59 |
ray_laptop | (it's a shell script, not some WSH or .bat file) | 17:37.02 |
| chrisl: only if it has 'git' | 17:37.15 |
Robin_Watts | chrisl: Console 2 doesn't change cmd.exe. | 17:37.19 |
chrisl | ray_laptop: it has git, yes. It's how I do git stuff on Windows. | 17:37.49 |
Robin_Watts | It runs cmd.exe in a different window from "Windows Command Prompt". | 17:38.01 |
| but it lets me have msys and cygwin in the same window (different tabs) | 17:38.15 |
| and has a (configurably) sane cut/paste. | 17:38.27 |
| and cleartype. | 17:38.32 |
chrisl | Yeh, it's the cut and paste that cause the (nearly) violent outburst. | 17:38.53 |
ray_laptop | with a 150dpi screen, I don't need cleartype | 17:38.54 |
| chrisl: did you enable 'QuickEdit' mode (it's better than the title bar interface) | 17:39.30 |
Robin_Watts | with 2 1920x1200 monitors, you'd think I wouldn't need browsers/command windows with tabs either, but... | 17:39.49 |
ray_laptop | chrisl: just in case you are on a machine without console2 | 17:40.06 |
chrisl | ray_laptop: I hadn't noticed QuickEdit mode before, I'll give it a try. | 17:40.19 |
| Hmm, better, but still a bit icky..... | 17:40.52 |
ray_laptop | chrisl: It is left mouse drag to select, <enter> to 'copy' then right mouse to paste | 17:41.23 |
Robin_Watts | I have Console 2 configured to shift drag to select, Ctrl-Shift-C to copy, and Ctrl-V to past. | 17:42.13 |
| +e | 17:42.16 |
chrisl | ray_laptop: it still doesn't honor lines as I would like - 'course, I don't know it Console 2 does either. | 17:42.17 |
ray_laptop | the one thing that's sort of strange is it lets you cut a rectangle from the middle of the screen and doesn't wrap to the beginning of lines below | 17:42.29 |
Robin_Watts | console 2 wraps lines. | 17:42.37 |
| So cutting/pasting command lines works. | 17:42.50 |
chrisl | Robin_Watts: that might just save my laptop from failing to learn to fly ;-) | 17:43.25 |
Robin_Watts | I also recommend ansicon and cmdkey | 17:44.10 |
| cmdkey means Ctrl-A and Ctrl-E do sane things, and command line history works sanely. | 17:45.08 |
ray_laptop | I miss vi edit mode on the history -- does console2 support that ? | 17:45.42 |
Robin_Watts | That sounds more like a cmdkey thing than a console 2 thing. | 17:46.09 |
chrisl | Robin_Watts: do you have a link for cmdkey? | 17:46.29 |
ray_laptop | well since it hooks keys and handles history, I thought it might :-/ | 17:46.55 |
Robin_Watts | http://adoxa.110mb.com/ | 17:47.12 |
| ray_laptop: console2 doesn't handle history. | 17:47.38 |
chrisl | great, thanks! | 17:47.40 |
ray_laptop | how do I switch to the 9.03 branch ? | 17:47.48 |
Robin_Watts | git checkout ghostscript-9.03 | 17:47.58 |
chrisl | it's ghostpdl-9.03 | 17:48.19 |
ray_laptop | Robin_Watts: I tried that: error: pathspec 'ghostscript-9.03' did not match any file(s) known to git | 17:48.27 |
Robin_Watts | oh. Is that consistent ? | 17:48.33 |
ray_laptop | all of our releases are tagged as ghostscript-X.XX | 17:48.55 |
chrisl | Well, the branch is of the entire ghostpdl tree. | 17:49.18 |
ray_laptop | we can still tag a ghostscript-9.03, right ? | 17:49.38 |
Robin_Watts | Yes, we can have multiple tags pointing to the same commit. | 17:50.08 |
chrisl | Yes, I'll tag with both names ghostscript-9.03 and ghostpdl-9.03 - that will be consistent with what we've done before. | 17:50.34 |
ray_laptop | and 9.03 is ONLY going to exist as a branch, right ? | 17:50.41 |
chrisl | ray_laptop: no, I'll tag it, just in case the customer needs support in the future. | 17:51.18 |
| ray_laptop: oh, but you mean the docs for 9.03, yes, that'll only exist on the branch - sorry, I think my lines crossed there. | 17:54.00 |
| ray_laptop: is that what you meant, or are we still fuzzy? | 17:55.45 |
ray_laptop | sorry phone call.... pls wait | 17:58.26 |
| chrisl: I meant that 9.03 won't 'reconnect' to the 'master' branch it'll be a dead end branch as far as 'git' is concerned, and the tag will point to it. | 17:59.46 |
chrisl | ray_laptop: yes, exactly. | 18:00.10 |
ray_laptop | chrisl: no fuzz here then (except, maybe some pocket lint) | 18:00.38 |
chrisl | ray_laptop: :-) good, I just wanted to be sure. | 18:01.17 |
mvrhel2 | I am back | 18:01.51 |
Robin_Watts | http://ghostscript.com/~robin/out.png <-- Woo HOo! | 18:02.03 |
mvrhel2 | henrys: so do I need to dig into a compile inits issue? | 18:02.09 |
Robin_Watts | mvrhel2: I think ray_laptop is testing a fix as we speak. | 18:02.43 |
mvrhel2 | ok great | 18:02.52 |
Robin_Watts | (unless I have my polarity reversed) | 18:02.57 |
ray_laptop | marcosw: cust 532 isn't happy with the patch. Can you do the bisect for the patch with FT_BRIDGE=0 ? I tried it, but am new to git bisect. I _think_ it is rev 967e771 | 18:03.10 |
| mvrhel2: I am testing a fix for master, and am looking into the issue on 9.03 | 18:03.49 |
Robin_Watts | tor8: See previous link. | 18:04.06 |
mvrhel2 | ray_laptop: do you need me to look into anything? | 18:04.37 |
ray_laptop | mvrhel2: no. It works on the 9.03 branch | 18:07.11 |
tor8 | Robin_Watts: awesome! | 18:07.29 |
ray_laptop | mvrhel2: but I'm seeing some "funny" stuff when I specify a relative path | 18:07.42 |
Robin_Watts | tor8: Now I just need to tidy it all up. | 18:08.11 |
| tor8: How come mupdf renders page 1147 of the pdf reference manual perfectly? That had me confused for ages. | 18:17.33 |
| I guess the manual must not actually implement that diagram using the mechanisms it's documenting. | 18:18.13 |
tor8 | like most things in the pdf spec, they're just images | 18:18.19 |
| which makes it easy to see what's supposed to happen, but is useless as a test file | 18:18.38 |
Robin_Watts | They aren't quite just images. | 18:18.52 |
| The non-isolated knockout case shows structure in the antialiasing. | 18:19.59 |
ray_laptop | alexcher: there is a problem with the checking for the validity of the GenericResourceDir. When I cd to Resource, run "../bin/gswin32c -IInit " I get the default set to c:/gs/Resource/Init NOT the expected ../Resource/ | 18:21.58 |
| alexcher: but ../bin/gswin32c -I../Resource/Init works | 18:22.34 |
tor8 | hmm, well, they cheat a lot anyway was my point :) | 18:22.39 |
Robin_Watts | tor8: just pondering how we do knockout groups. | 18:23.28 |
| I guess we could put a non-isolated group around every drawing operation. | 18:24.11 |
| except rather than copying the background from one up the stack, we copy it from 2 up. | 18:24.33 |
ray_laptop | henrys: alexcher: mvrhel2: I think 9.03 is OK to go w.r.t. COMPILE_INITS=0 (or 1) | 18:24.34 |
mvrhel2 | ok so what is the deal with that bug? | 18:25.21 |
henrys | I'm confused | 18:25.46 |
ray_laptop | mvrhel2: there _is_ a bug with master | 18:25.55 |
mvrhel2 | oh | 18:26.00 |
| when was that introduced? what is the tag for 9.03? | 18:26.23 |
henrys | so it was broken fixed and then broken again? Is that what we have here? | 18:27.01 |
ray_laptop | I think 9.03 was tagged BEFORE the tagfix changes | 18:27.11 |
mvrhel2 | I wonder how the tagfix changes could have affected the compile inits icc directory stuff | 18:27.41 |
ray_laptop | henrys: if someone wants to bisect, fine. I don't | 18:27.44 |
| mvrhel2: it wasn't the tagfix that broke HEAD, but the 12-07D.PS "fix" for the nulldevice color profile | 18:28.16 |
henrys | it's probably just reproducible on linux - that would explain a number of issues. | 18:28.22 |
mvrhel2 | ray_laptop: but you just put that in. this bug was in long before that | 18:28.44 |
ray_laptop | henrys: i will log into peeves and check 9.03 on linux | 18:28.50 |
| alex's testing began 7/7 | 18:29.15 |
ray_laptop | goes to check the logg dates for the patches.... | 18:29.42 |
mvrhel2 | uhoh email from customer 711. Looks like they are not going to update... | 18:30.11 |
henrys | mvrhel2:can you respond to the other part of the email? | 18:35.03 |
mvrhel2 | yes. I will do that | 18:35.12 |
ray_laptop | strange, on peeves, I am getting error: pathspec 'ghostpdl-9.03' did not match any file(s) known to git. | 18:35.38 |
henrys | when did you last rebase? | 18:36.49 |
| I'll test it on the mac while we're at it. | 18:37.04 |
ray_laptop | henrys: I just did git pull --rebase it said "Current branch master is up to date." | 18:37.42 |
henrys | hmm works fine for me - I did git checkout ghostpdl-9.03 | 18:39.20 |
ray_laptop | Robin_Watts: do you have any idea what I need to look at ? the git config -l looks OK (except it doesn't show the branch.ghostpdl-9.03 as my laptop does, but I don't know if I have to checkout first) | 18:39.58 |
henrys | chrisl:--disable-compile-inits doesn't work for me, unknown option. | 18:42.57 |
| continuing with manual disable | 18:43.09 |
Robin_Watts | ray_laptop: Do a git fetch | 18:43.31 |
henrys | nvm cockpit error | 18:43.51 |
ray_laptop | Robin_Watts: OK. nothing from git fetch | 18:46.08 |
| and git checkout ghostpdl-9.03 still fails as before | 18:46.50 |
Robin_Watts | I don't know :( | 18:47.37 |
| do: git logg -all | 18:48.13 |
ray_laptop | Robin_Watts: OK | 18:48.33 |
Robin_Watts | do you see an origin/ghostpdl-9.03 in there, about 2 pages or so down? | 18:48.43 |
ray_laptop | Robin_Watts: I see: | * 487c62e (refs/remotes/origin/ghostpdl-9.03) Updates PDL README.txt | 18:49.35 |
Robin_Watts | Right. | 18:49.43 |
| What vintage git ? | 18:50.06 |
| git --version | 18:50.21 |
| My git here is 1.7.4.msysgit.0 | 18:50.38 |
| I'm betting the one on ubuntu is stone aged. | 18:50.48 |
| and hence doesn't cope with automatically creating a local copy of a remote branch when you check it out. | 18:51.13 |
henrys | alexcher can you reproduce this on windows? | 18:53.46 |
ray_laptop | sorry ANOTHER phone call :-( | 18:55.24 |
| git version 1.6.3.3 | 18:55.42 |
| on linux (peeves) | 18:55.51 |
Robin_Watts | ray_laptop: worth updating, I suspect. | 18:56.00 |
chrisl | henrys: sorry, phone call - I take it you got it working? | 18:57.19 |
ray_laptop | Robin_Watts: upgrading ... | 18:57.35 |
| oops. apt-get doesn't show an upgrade for git :-( | 18:58.41 |
sebras | Robin_Watts: I just saw out.png, cool! :) | 19:00.40 |
henrys | chrisl:yes and I can't find the iccprofiles | 19:00.48 |
chrisl | henrys: that's not good :-( | 19:01.11 |
Robin_Watts | sebras: yes, and it turned out to be less painful than I'd feared. | 19:01.20 |
henrys | at least setting GenericResourceDir - I haven't tried other options. | 19:01.30 |
chrisl | ray_laptop: do you need to be more specific on the package name, like git-core? | 19:02.14 |
ray_laptop | Robin_Watts: I think I figured it out. I did: git branch ghostpdl-9.03 then set the remote=orign and merge=refs/heads/ghostpdl-9.03 and it _seemed_ to work (did a lot of stuff) when I did git co ghostpdl-9.03 | 19:07.00 |
chrisl | ray_laptop: the latest commit on the branch should be: 487c62 - Updates PDL README.txt | 19:08.13 |
henrys | even copying the the resource to the default expected directory /usr/local/share/ghostscript9-03/ doesn't work - I can't imagine why this works on windows if it does maybe folks aren't actually doing COMPILE_INITS=0 on windows. | 19:08.28 |
ray_laptop | chrisl: I'll check... | 19:08.30 |
| henrys: I _know_ it is doing COMPILE_INITS=0 on windows -- I looked with the debugger and also there is not gsromfs1 ! | 19:09.21 |
henrys | okay that would explain why michael didn't see it in the first place. | 19:10.22 |
ray_laptop | I'm going to re-clone my git on peeves. It is honked up somehow | 19:11.02 |
chrisl | henrys, ray_laptop: I'm calling it a night now. Unless someone mails to tell me different, I *won't* be doing the release in the morning - I'll wait until we have a consensus on a "go", okay? | 19:11.21 |
henrys | chrisl:right | 19:11.50 |
chrisl | Cool, g'night all! | 19:14.03 |
alexcher | henrys: I'm using 64-bit GNU/Linux. A Windows box is still available for the special cases. | 19:15.45 |
henrys | mvrhel2:tracing opens it is opening the profile once but it fails the second time - does the interpreter and device open the profiles separately? | 19:19.22 |
| the first open it uses the complete path the second open it tries to open the file without a path. | 19:20.21 |
ray_laptop | Robin_Watts: I did a fresh clone on peeves and git checkout ghostpdl-9.03 _still_ fails | 19:22.06 |
Robin_Watts | ray_laptop: Yes. | 19:22.14 |
| It's your version of git. | 19:22.19 |
| New versions of git know that if you try to checkout an origin/branch then they should create a local branch of that name from it and swap to it. | 19:23.00 |
| (I think) | 19:23.12 |
ray_laptop | Robin_Watts: apt-get install git-core tells me: git-core is already the newest version. | 19:23.29 |
Robin_Watts | Right. Ubuntu is years behind. | 19:23.39 |
| http://blog.avirtualhome.com/2011/01/31/git-1-7-4-released-and-available-for-ubuntu-10-10-and-10-04/ | 19:24.08 |
ray_laptop | so how is anyone else working on ubuntu | 19:24.12 |
mvrhel2 | henrys: multiple profiles are opened. one for the device and several for the icc manager | 19:24.21 |
Robin_Watts | ray_laptop: You just can't work using git checkout blah. You need to create the branch manually. | 19:25.02 |
henrys | so it looks like the icc manager opens have paths and the device open does not have the path directory. | 19:25.44 |
| which I believe is consistent with what alexcher has been saying but why wouldn't we see the same thing on windows. | 19:26.50 |
| ? | 19:26.51 |
| alexcher:I think for your job you need a windows and linux box ready to go. | 19:27.25 |
| most of our customers use windows. | 19:27.37 |
| I have git 1.7.1 on ubuntu 10.10 fwiw | 19:30.02 |
| should I check if git works? | 19:30.20 |
| for the 9.03 branch I know it works so far. | 19:30.43 |
ray_laptop | Robin_Watts: git co -t origin/ghostpdl-9.03 seemed to do the trick (I think) | 19:32.03 |
| how can I make sure ? | 19:32.18 |
Robin_Watts | git logg ? | 19:32.44 |
ray_laptop | well it shows me | * 487c62e (refs/remotes/origin/ghostpdl-9.03, refs/heads/ghostpdl-9.03) Updates PDL README.txt is that right ? | 19:33.18 |
Robin_Watts | That shows that you have both origin/ghostpdl-9.03 and ghostpdl-9.03 | 19:33.48 |
| but I would have expected HEAD to be there too. | 19:33.55 |
| git checkout ghostpdl-9.03 too maybe ? | 19:34.09 |
ray_laptop | * a06634a (refs/remotes/origin/master, refs/remotes/origin/HEAD, refs/heads/master) Update autoconf build ... | 19:34.24 |
Robin_Watts | not origin/HEAD, HEAD. | 19:34.39 |
ray_laptop | Already on 'ghostpdl-9.03' | 19:34.54 |
Robin_Watts | Maybe stoneage git doesn't show HEAD. | 19:35.06 |
ray_laptop | is there a way to show my current SHA1 ? | 19:35.58 |
Robin_Watts | git name-rev HEAD | 19:36.35 |
ray_laptop | Robin_Watts: my .git/refs/heads/ghostpdl-9.03 has 487c62e88ce2792a9b58fd23a9fcb12a0396f7dc which looks right | 19:36.58 |
henrys | git show probably a little more informative. | 19:37.14 |
ray_laptop | HEAD ghostpdl-9.03 | 19:37.19 |
| git show has the same rev as the first. Hurray!! I don't have to upgrade ;-) | 19:37.59 |
henrys | mvrhel2:where the call to open the icc for the device? | 19:38.03 |
Robin_Watts | right. git rev-parse HEAD | 19:38.07 |
ray_laptop | henrys: all of the icc opens use sfopen | 19:38.28 |
Robin_Watts | git rev-parse takes any kind of reference (like HEAD^^~4 or something) and returns the SHA. | 19:38.39 |
| git name-rev takes the same kind of thing and tries to give a human readable version. | 19:39.03 |
henrys | ray_laptop:it would really be best to upgrade. there will be other problems. | 19:39.38 |
Robin_Watts | food. | 19:39.56 |
ray_laptop | henrys: I will when I can find time. I'd rather see if I can reproduce any 9.03 problems | 19:40.25 |
| I just wanted to make sure I was really on 9.03 | 19:40.57 |
alexcher | Current gs doesn't build on Windows: Undefined external symbol _wmemset | 19:44.36 |
| I'm looking into this problem. | 19:44.54 |
ray_laptop | alexcher: I just built on Windows ! | 19:45.20 |
| alexcher: what version VS ? | 19:45.37 |
ray_laptop | uses VS 9 (2008) | 19:45.53 |
alexcher | 2003 | 19:47.03 |
ray_laptop | alexcher: iisn't that MSVC 7 ? | 19:49.07 |
| I have 6 and 7, but not installed. I only have 8 and 9 installed | 19:50.05 |
chrisl_away | alexcher: it looks like you'll need to leave out the unicode changes | 19:51.46 |
alexcher | Yes, it's 7.1. I have v. 9 on my laptop. | 19:52.10 |
ray_laptop | henrys: i get an error with the default device on peeves (x11alpha) but NOT with ppmraw (with 9.03) | 19:53.39 |
mvrhel2 | henrys: sorry, I stepped off to grab a bite to eat | 19:54.15 |
| I am back now | 19:54.17 |
chrisl_away | alexcher: if you add /DWINDOWS_NO_UNICODE to the CFLAGS in msvc.mak that *should* get it building, I think. | 19:54.55 |
ray_laptop | interestingly 9.03 also fails with: debugbin/gs -IResource/Init -dNODISPLAY -c quit | 19:56.08 |
| that uses the 'nullpage' device. | 19:57.28 |
| probably the 'copydevice' isn't setting up a reasonable icc_dir since it comes in as NULL | 19:58.05 |
| henrys: do you want me to continue and look into it, or do you want to do it. | 19:58.31 |
| I'll check logs. Have to run an errand now... | 19:58.55 |
henrys | if you are busy I'll do it. | 19:59.02 |
ray_laptop | henrys: I just need to do something for cust 532 and run an errand, so i won't be back to it for a couple of hours. If you want to dig into it in the meantime, fine. I'll check IRC logs and email before diving back into it. | 20:00.16 |
| at least I can reproduce it now (not with -sDEVICE=ppmraw but some devices fail) | 20:00.41 |
henrys | okay | 20:00.45 |
ray_laptop | interestingly, -dNODISPLAY _works_ on Windows... | 20:01.18 |
henrys | right ppmraw works for me to - bbox, X11 and friends do not. | 20:12.20 |
mvrhel2 | whew. ok. so the object dependent color management on the source side seems to be working now | 20:18.27 |
| both in clist and non-clist cases | 20:18.40 |
| henrys: almost done with my reply to customer 711 | 21:24.02 |
| They basically need to figure out a transfer function | 21:24.20 |
| at least that is the way that I would solve this issue | 21:24.41 |
henrys | uhoh ray_laptop's patch with 0 probability of breaking something caused a regression in our system. | 21:50.49 |
| thanks mvrhel2 for answering that I didn't know if there was something special you might know about the odd technology. | 21:52.20 |
mvrhel2 | in the end it is a common problem for most print technologies | 21:53.00 |
| and one that should be handled before spending a bunch of time with optimal screens | 21:53.35 |
| sending the email now. I cc'd you and support | 21:53.53 |
| going to take a break for bit. | 21:56.05 |
henrys | profile_struct = dev->icc_array hmm think I'm not going to like the terminology in this one. | 22:21.06 |
mvrhel2 | henrys: I am back | 23:41.12 |
| did you find something that gave you heartburn? | 23:42.28 |
henrys | well no big deal icc_array isn't really an array but a struct so I found the terminology confusing. | 23:46.44 |
mvrhel2 | yes. I should rename that | 23:46.54 |
| when I started it was going to be an array | 23:47.02 |
| and then I had to include the rendering intents | 23:47.09 |
| which I am about to set up a test profile to demonstrate the use of those | 23:47.27 |
| I will rename that in the process | 23:47.37 |
henrys | but that's trivial, for x11 and bbox the device profile is not initialized that "sync" routine isn't being called. I have to leave in a bit but I can probably track it down later this evening. | 23:47.50 |
mvrhel2 | bbox should not have a device profile | 23:48.14 |
| but it should forward to the x11 device | 23:48.21 |
| to get it | 23:48.24 |
henrys | well you can run the code with the bbox device and folks do. | 23:48.37 |
mvrhel2 | oh as a target? | 23:48.49 |
henrys | it shouldn't error out. | 23:48.49 |
| yes | 23:48.53 |
mvrhel2 | ah | 23:48.57 |
| hmm | 23:48.58 |
| in that case it should be be set for forward | 23:49.19 |
| when the bbox is installed we should check if it has a target and adjust the proc with respect to this | 23:50.15 |
| the proc for getting the icc_array that is | 23:50.23 |
henrys | you should be able to see that one on windows - compile inits 0 | 23:50.30 |
mvrhel2 | and run with bbox as the output device? | 23:50.42 |
henrys | yes | 23:50.49 |
mvrhel2 | ok. I will give that a try | 23:50.54 |
| so also with compile inits 0 the x11alpha device fails? | 23:51.21 |
| on linux? | 23:51.24 |
| or the x11 device? or both? | 23:51.41 |
henrys | yes | 23:51.42 |
| both | 23:51.45 |
mvrhel2 | ok. | 23:51.51 |
| I will stop the profile stuff for a bit and look at the bbox right now | 23:52.30 |
henrys | okay and I'll continue exploring later tonight if there is more to do. | 23:53.29 |
| Forward 1 day (to 2011/07/13)>>> | |