| <<<Back 1 day (to 2012/09/25) | 2012/09/26 |
henrys | Robin_Watts:for the logs any image should crash with -ZBb - problem with the memory parameter debug change. | 02:44.37 |
mvrhel | black point compensation control seems to be working. need to check tomorrow that it works based upon object type. done for the night now though | 05:07.32 |
henrys | kens:your arrival is a sign I should be done for the evening. Have a good day. | 07:11.56 |
kens | good night henrys | 07:12.05 |
Robin_Watts | kens: So your customer seem to have solved their problem; I guess that's an implicit admission of not being able to build. | 09:58.11 |
| How do I do a luratech build ? | 10:00.41 |
| Never mind. | 10:05.07 |
paulgardiner | Robin_Watts: did you have a chance to look at those commits? | 10:13.48 |
Robin_Watts | sorry, no, let me do so now. | 10:24.05 |
paulgardiner | Robin_Watts: ta | 10:28.10 |
Robin_Watts | paulgardiner: app.launchURL("url", 0, 1); | 10:37.55 |
| What would that do in Acrobat ? | 10:38.00 |
paulgardiner | I haven't looked yet. I do intend to at some stage. Really, that should never appear in a file, though of course it might in some. | 10:38.50 |
Robin_Watts | I'd be more tempted with: default: if (argc <= 0) return NULL; in app_launchURL. | 10:38.55 |
| but otherwise they all look fine. Let me pass it as is. | 10:39.46 |
paulgardiner | Yes that is probably better. I'll look at that in a follow up later | 10:40.08 |
Robin_Watts | pushed. | 10:42.00 |
tkamppeter_ | Can someone have a llok at https://bugs.launchpad.net/ubuntu/+source/ghostscript/+bug/1042111 Here a user complains about a "Font Renderer Plugin ( FreeType ) return code = -1" error message in 9.05 which is solved in 9.06. Can someone check if this is a small patch which one could release as bug fix update? | 10:45.14 |
Robin_Watts | tkamppeter_: git bisect is your friend :) | 10:46.02 |
paulgardiner | Robin_Watts: thanks | 10:46.12 |
Robin_Watts | tkamppeter_: chrisl is the guy who might know, but he's out today. | 10:46.56 |
| If you open a matching bug on bugs.ghostscript.com, then marcos may look for you, but we have enough to do tracking our own bugs than to go scouring other peoples bug trackers for more work. | 10:48.42 |
| Or of course, you can run git bisect yourself and have an answer very quickly :) | 10:49.06 |
tkamppeter_ | Robin_Watts, thank you very much. | 11:02.23 |
saper | any chance to get http://bugs.ghostscript.com/attachment.cgi?id=8749 committed? (http://bugs.ghostscript.com/show_bug.cgi?id=692527). It even makes svg writer usable (for simple cases) | 11:18.22 |
Robin_Watts | tor8: I'm going to look into the mail to support from Paul Hudson. | 12:12.49 |
tor8 | Robin_Watts: right. we might take this reminder as an opportunity to fix the hyperlink UI. I didn't like the tri-state button, and I wanted to implement the same for iOS to release at the same time. | 12:15.12 |
Robin_Watts | I wasn't planning to implement anything, just to figure out what was wrong with the iWork thing (it's graphics being clipped to text, and the text gets too large, so we get an fz_warn("unimplemented"). | 12:16.52 |
| And to tell him that following links is supported, just not enabled in that version. | 12:17.11 |
tor8 | Robin_Watts: oh. hm. thought I fixed that, but maybe I just warned about that case... | 12:19.55 |
Robin_Watts | You fixed the fill case, but not the clip case. | 12:20.51 |
kens | Robin_Watts : yesy I think there's an implicit admission form the customer that they patched the sources incorrectly for the Mac (though why they wouldn't use the same source as for Windows escapes me). | 12:30.18 |
| The other possibility of course is that they failed to rebuild the Mac code, and were using unpatched binaries, which would explain the resutl too. | 12:30.41 |
ray_laptop | morning, all | 13:46.37 |
Robin_Watts | Morning ray | 13:46.47 |
ray_laptop | sorry I missed the meeting yesterday. Anything I need to go look at in the logs ? | 13:46.55 |
Robin_Watts | Yeah, I have a new unicode patch up. | 13:47.05 |
| and I'm canvassing for opinions. It's in my repo on casper. | 13:47.20 |
ray_laptop | I saw that, but where is the patch, and what I supposed to look at it for ? | 13:47.45 |
Robin_Watts | http://git.ghostscript.com/?p=user/robin/ghostpdl.git/.git;a=commitdiff;h=0d3ce254a0a03d3530623c5be5b4c1d92466f87c | 13:48.16 |
| Read the docs etc to see if you're happy with the approach. | 13:48.34 |
ray_laptop | Robin_Watts: OK. looking now | 13:49.10 |
| Robin_Watts: OK. So Windows command line args will always be 'local' encoding UNLESS the app calls gs_set_arg_encoding ? | 13:52.25 |
Robin_Watts | yes. | 13:52.59 |
| hence maintaining backward compatibility. | 13:53.13 |
ray_laptop | Robin_Watts: and we don't have a way to have a windows utf8 command line ? | 13:53.33 |
Robin_Watts | Sorry? | 13:53.45 |
ray_laptop | Robin_Watts: a gswin32c that takes utf8 args | 13:54.00 |
| not sure if that even makes sense to the windows user world | 13:54.23 |
Robin_Watts | The windows apps, unless they are specifically compiled with WINDOWS_NO_UNICODE will be built as unicode things that take unicode command lines, and convert that down to utf8 before calling the core. | 13:54.27 |
| No one just calling gswin32c.exe will probably ever spot the difference. | 13:55.14 |
ray_laptop | Robin_Watts: OK, I see. So the gswin32(c) is now always unicode (wchar) ? | 13:55.17 |
Robin_Watts | Yes. | 13:55.23 |
| (Unless explicitly built with WINDOWS_NO_UNICODE) | 13:55.38 |
| But people that are using gsapi within their own apps have the choice of which one to use. | 13:56.08 |
| local or utf8. | 13:56.15 |
ray_laptop | Robin_Watts: I read the docs and it seems like a good approach to make unicode support there by default. I haven't looked at the diffs yet. Do I need to ? | 13:56.22 |
Robin_Watts | Only if you are feeling particularly diligent. | 13:56.39 |
ray_laptop | Robin_Watts: and it cleverly avoids breaking people's (customer's) apps | 13:56.56 |
| Robin_Watts: if I start feeling diligent, I'll lie down until it passes ;-) | 13:57.36 |
Robin_Watts | That was the thing that Sags was objecting to | 13:57.42 |
| :) | 13:57.43 |
ray_laptop | Robin_Watts: yes, me too. Breaking customer apps is generally something to be avoided | 13:58.32 |
| Robin_Watts: so I'm good with it | 13:58.54 |
Robin_Watts | cool. | 13:59.12 |
ray_laptop | I have to get the kids up, but I'll be here off and on. My youngest has the stomach flu and with 'camping' at Karen's mom's house while ours is being worked on, we are disorganized a bit | 14:00.17 |
Robin_Watts | ray_laptop: What are you having done to the house? | 14:01.46 |
| tor8: Patch up for review. http://git.ghostscript.com/?p=user/robin/mupdf.git;a=commitdiff;h=afabdc63e8976ee2f7475845f0d8387104fd8480 | 14:08.19 |
| That implements text clipping and stroked text clipping. | 14:08.30 |
ray_laptop | Robin_Watts: sorry. Kitchen, flooring (remove tile in kitchen/den carpet in living room dining room) replace with laminate, ceiling texture removal (popcorn) install can lights, new paint, new cabinets, ... | 14:11.24 |
| Robin_Watts: then after that's finished they do the ceiling and painting upstairs and reno the bathrooms | 14:12.14 |
tor8 | Robin_Watts: right, looks like it should work! | 14:12.21 |
| but there's a 1.0 without 'f' suffix in there | 14:12.54 |
| Robin_Watts: use either 1 (and let it promote) or 1.0f please, to avoid the doubles | 14:13.27 |
Robin_Watts | sure. | 14:13.34 |
| ray_laptop: Cor. Lots of work. I understand why you're camping elsewhere :) | 14:13.48 |
tor8 | my preference is 1 but I'm not too annoyed by 1.0f | 14:13.54 |
Robin_Watts | 1 is fine by me. | 14:14.04 |
henrys | ray_laptop:when you get a chance can you look at Guillaume's email and see what he is doing, it makes absolutely no sense to me. well it makes sense but it breaks many invariants in devices and shouldn't work at all. | 14:27.53 |
Robin_Watts | henrys: I think he needs to be told "Do not do this". | 14:29.52 |
henrys | Robin_Watts:maybe a script to run through all the -Z letters and run a few jobs would be a good test for your work. | 14:36.01 |
Robin_Watts | henrys: I'm sure we'll run into them over the months before the next release. | 14:37.05 |
ray_laptop | henrys: OK. I'll have a look. I suspect that the device was doing things that there didn't used to be a way to do. | 14:38.02 |
henrys | Robin_Watts:true but it is kind of a nuissance to trip over it. Usually when you use -Z you are in the middle of debugging something and don't wnat to be interrupted to fool with that, but I don't feel strongly. | 14:39.26 |
Robin_Watts | Just had a bug reported from my Skypey mupdf customer of a file that takes 600Meg+ to render a particular page. I think it's to do with shadings. | 14:44.43 |
| This is the second one in a week. | 14:45.17 |
kens | Oh I thought it was the same one you were talking about | 14:45.28 |
Robin_Watts | I wonder if we should maybe keep shadings in their compressed form and uncompress them bit by bit at render time. | 14:46.02 |
tor8 | Robin_Watts: what kind of shading? | 14:46.18 |
Robin_Watts | Dunno yet. 137Meg file still transferring. | 14:46.30 |
| Triangle mesh, I believe. | 14:46.34 |
tor8 | the shading representation is fairly compact, in general | 14:46.51 |
Robin_Watts | Really? 34 floats per vertex. 3 vertexes per triangle... | 14:48.15 |
tor8 | 3 floats per vertex for shadings with a function, 5 floats per vertex for RGB shadings | 14:49.19 |
| not sure where you got the number 34 from | 14:49.30 |
Robin_Watts | pdf_shade.c: | 14:50.13 |
| struct vertex | 14:50.15 |
| { | 14:50.17 |
| float x, y; | 14:50.19 |
| float c[FZ_MAX_COLORS]; | 14:50.21 |
| }; | 14:50.22 |
tor8 | that's not what gets stored in the fz_shade though | 14:50.57 |
| only used in the parser and to pass to the pdf_add_xxx functions | 14:51.19 |
| see: pdf_add_vertex | 14:51.31 |
Robin_Watts | ok. So maybe this file has lots of small shadings, and it's the fz_shade overhead itself that's large. | 14:52.26 |
| I will dig into it all a bit later. | 14:52.34 |
| Right now, I'm looking at the cluster to think about mupdf submodules. | 14:52.48 |
| tor8: Does it hurt for me to git submodule init more than I need to ? | 14:53.02 |
tor8 | well, if it's a 130 meg large shading, represented with 8-bit coords, then it could conceivably blow up a fair bit when turned into floats | 14:53.07 |
| Robin_Watts: no, it's benign. you can also run "git submodule update --init" | 14:53.24 |
Robin_Watts | There are 270+ pages in the file. | 14:53.37 |
tor8 | and skip the --init altogether | 14:53.41 |
Robin_Watts | Cool. | 14:53.47 |
tor8 | s/--init/git submodule init/ | 14:53.50 |
henrys | so we lack gs malloc in jbig2dec also, not just luratech right? | 14:55.53 |
Robin_Watts | I can't see calls to malloc/realloc/free in sjbig2.c, where I could in sjbig2_luratech.c | 14:58.42 |
| but then I can't see any allocation calls in sjbig2.c | 14:58.53 |
| Right jbig2.c uses malloc/realloc/free too. | 14:59.29 |
| Looks like a simple enough fix though. | 14:59.49 |
| (famous last words) | 14:59.57 |
henrys | kens and Marcos for the logs: the test file associated with 689980 made me think of interesting testing ideas. for example being able to set the device and output in a postscript file would allow us to go through all the devices with a single file that could be run as part of a regular clusterpush. | 15:00.20 |
kens | henrys it works for pdfwrite. | 15:00.39 |
| More or less, I didn't spend much imte on it | 15:00.48 |
Robin_Watts | unleashes git submodules on the mupdf tests in the cluster. | 15:01.12 |
kens | Robin_Watts : I've just been looking at Virgin flilghts for December, and they are already £725 | 15:02.20 |
Robin_Watts | So, if we want to attend the last day of the show, we'd need to be there in time for the show on the 7th. | 15:04.38 |
| which would mean, I think, flying out on the 6th. | 15:04.49 |
kens | Yes, but the prices are very similar | 15:04.53 |
| You cna get £40 lower flygin back on Monday or out on Tuesday (I think it was Tuesday) | 15:05.12 |
| Still £685 though | 15:05.18 |
Robin_Watts | price difference is dwarfed by the extra cost of the rooms. | 15:05.45 |
| Thursday -> Sunday is 705 | 15:06.05 |
kens | Actually its a lot cheaper to fly on the Thursday, comes out at £603 | 15:06.24 |
| Robin_Watts : I'm seeing 603 | 15:06.38 |
Robin_Watts | how strange. I'm not. | 15:06.50 |
kens | depart 11 am on the 6th Dec, return 16:25 on the 9th December | 15:07.14 |
| I am logged in, I wonder if that makes a difference | 15:07.35 |
Robin_Watts | I am logged in too. You are going LHR -> SFO, right? :) | 15:08.40 |
kens | Yes :-) | 15:08.47 |
| I actually selected London All, let me check again | 15:09.03 |
| Yes definitely Heathrow | 15:09.30 |
Robin_Watts | I had London ALl too. | 15:09.31 |
kens | Still comes up as £603.19 for me | 15:10.12 |
Robin_Watts | 603 is for PE return? | 15:11.01 |
kens | PE ? | 15:11.10 |
Robin_Watts | sorry. I get 603 if I tell it miles + money | 15:11.24 |
kens | AH, maybe I have that checked, let me see | 15:11.34 |
Robin_Watts | with premium economy return. | 15:11.48 |
kens | Aha, yes I did have that checked, strange..... | 15:12.01 |
| That's better, now I see 705 | 15:12.29 |
Robin_Watts | PE both ways is 918. Not doing that. | 15:12.55 |
| but PE back makes sense. | 15:13.03 |
kens | That's what I saw too, but I might go for it | 15:13.10 |
| You're seeing 918 ? Is that miles+money ? I'm seeing 1083 | 15:14.09 |
| Thurs->Sun | 15:14.28 |
Robin_Watts | 918 = miles + money thurs ->sun, PE both ways. | 15:14.44 |
kens | Yes, that makjes sense, I was looking at just 'fare' | 15:14.58 |
| So spending miles saves you 165 quid, hardly seems worthwhile | 15:15.27 |
Robin_Watts | Well, for me it's "spending miles saves 100 quid and gets me PE on the way back". | 15:16.17 |
kens | I used up a load of miles going to New York.... | 15:17.04 |
Robin_Watts | The only reason for not doing it would be because I need the points to stay a "silver" member, but then I don't think I'm ever going to have enough points to keep me there as they changed the requirements. | 15:17.37 |
kens | the requirement change makes it highly unlikely, yes | 15:17.59 |
| I have to go put a futon in a car and take it away. back shortly. | 15:18.45 |
Robin_Watts | actually... once you're in silver, it supposedly only takes 10 points a year to keep you silver, and I've had 8 this year already. | 15:24.18 |
marcosw | Robin_Watts: you kind of broke run.pl: | 15:36.41 |
| Global symbol "$newTime" requires explicit package name at run.pl line 798. | 15:36.41 |
| Global symbol "$newTime" requires explicit package name at run.pl line 799. | 15:36.42 |
| BEGIN not safe after errors--compilation aborted at run.pl line 1300. | 15:36.43 |
Robin_Watts | ah. | 15:37.01 |
marcosw | you _really_ need to check to make sure run.pl at least compiles if you modify it. Running it without any parameters on casper will accomplish that. | 15:37.39 |
Robin_Watts | ok, I've fixed it. Sorry. | 15:38.15 |
| "I can't possibly break anything if i'm just commenting stuff out... right?" :( | 15:38.38 |
marcosw | you python hackers are all alike... | 15:38.58 |
Robin_Watts | I don't speak python. | 15:39.08 |
marcosw | really? what other modern (?) language doesn't require variables to be declared? | 15:39.39 |
Robin_Watts | BASIC? :) | 15:39.52 |
marcosw | don't forget FORTRAN (and Forth and LISP), but I'm dating myself. | 15:40.29 |
Robin_Watts | Ah, the narcissists reason for wanting cloning. | 15:41.14 |
marcosw | Now we get compiler errors in mupdf: | 15:41.22 |
| CC build/release/image_jpx.o | 15:41.23 |
| fitz/image_jpx.c:4:22: error: openjpeg.h: No such file or directory | 15:41.24 |
Robin_Watts | yeah, that's more like it. | 15:41.32 |
marcosw | and I'm thinking this isn't even the "major commit to mupdf that will break the cluster" event that you were talking about yesterday. or is it? | 15:42.27 |
Robin_Watts | yes. | 15:45.40 |
| it's exactly that. | 15:45.45 |
| Instead of getting thirdparty from thirdparty.zip we now get it from git submodules. | 15:46.10 |
| but obviously it's not checking out the thirdparty stuff correctly, hence it's not finding openjpeg.h | 15:46.29 |
| so I need to check what I did wrong. | 15:46.34 |
kens | Robin_Watts : where did you find the info on staying in solver ? | 15:55.59 |
Robin_Watts | kens: On their website, but I just phoned them and was told that it has changed and you really do need 15 points a year to stay there, not 10. | 15:56.45 |
kens | Yeah thought so | 15:56.55 |
| So I'll be falling out then | 15:57.06 |
| Ah well. | 15:57.10 |
Robin_Watts | The website does seem to contradict that in at least 2 places, but hey... | 15:57.13 |
| So are you going to book then? | 15:57.23 |
kens | Robin_Watts : I just marked the file for bug #693360 as private | 15:58.38 |
Robin_Watts | Thanks. | 15:58.44 |
kens | 9.2MB file, so its 'fairly' large | 15:58.58 |
Robin_Watts | kens: I'm going to book @ 603 then. | 16:00.45 |
| Fly on 6th, back on 9th. | 16:00.57 |
kens | Right, I'm out of time but will do so tomorrow | 16:01.12 |
| I hear a pizza calling me.... | 16:01.22 |
| Goodnight all | 16:01.58 |
Robin_Watts | marcosw: I think git commits should be testing OK now. | 16:58.07 |
| Next thing is to fix user tests. Currently we don't 'push' thirdparty, right? | 16:58.44 |
marcosw | right. | 16:59.04 |
| actually, wait, we might | 16:59.17 |
| I think the push code assumes that the mupdf code is complete, so that would include the third party. | 16:59.42 |
Robin_Watts | OK, so it might just work. | 17:00.19 |
marcosw | there is still a problem with the git commit code. | 17:00.40 |
| inches is reporting: | 17:00.43 |
| Directory 'thirdparty/jbig2dec' exists, but is neither empty nor a git repository | 17:00.44 |
Robin_Watts | Hmm. I pushed a job earlier that was supposed to delete thirdparty/jbig2dec entirely. | 17:01.32 |
marcosw | maybe not all nodes ran that job? | 17:01.55 |
| i7a, inches, and macpro appear to be in the same state. | 17:02.13 |
Robin_Watts | let me edit run.pl now to do it again then. | 17:03.35 |
| OK, that should clear thirdparty the next time run.pl is run. | 17:04.50 |
| I'll need to fix the git clusterbridge, as that will get confused by the thirdparty stuff. | 17:06.29 |
ray_laptop | mvrhel: are you available for a conference call with cust 532 (Len and Michael Chang and I) ? | 17:24.07 |
mvrhel | ray_laptop: I can do a call. I don't know if I am ready | 17:24.35 |
marcosw | Robin_Watts: the cluster appears to be working as well as it ever does | 17:24.37 |
Robin_Watts | marcosw: Yeah. | 17:24.47 |
| After this runs through, I'll try some user jobs. | 17:25.07 |
ray_laptop | mvrhel: I just sent you cust 532's hacked up gxclimag.c | 17:30.27 |
mvrhel | ok | 17:30.33 |
ray_laptop | oops you need thegximage.c as well. | 17:31.06 |
| mvrhel: i 11:00 OK with you for a call > | 17:31.51 |
| ? | 17:31.53 |
mvrhel | ray_laptop: yes that is fine | 17:32.04 |
| oh they should maybe replace these calls with the real API calls that we have | 17:32.51 |
| like lcmsicc_transform_color | 17:33.02 |
henrys | ray_laptop:I just signed you up for a conference call also, scott should get with you shortly. | 17:33.14 |
mvrhel | oh I forgot all about this stuff | 17:33.18 |
marcosw | Robin_Watts: I want to make a change to the cluster code later day. can you git commit the changes you recently made so that I when I screw things up I can revert? | 17:33.33 |
henrys | ray_laptop:it's about a new ETS customer. | 17:34.44 |
ray_laptop | mvrhel: OK I sent the gximage1.c | 17:34.51 |
| henrys: for what time ? | 17:34.55 |
henrys | ray_laptop:scott should call and let you know. | 17:35.23 |
ray_laptop | henrys: OK | 17:35.27 |
mvrhel | ray_laptop: thanks for telling miles about the payroll. I was starting to wonder what was going on | 17:35.47 |
ray_laptop | none of the rest of you noticed ADP's mess up ? | 17:36.22 |
mvrhel | oh I did | 17:36.27 |
| I had to write a monster check last week for a re-fi that I was doing and was wondering when the payroll deposit was going to happen | 17:37.03 |
ray_laptop | I have been paying for the renovation and so my account has also been running lower than usual | 17:37.45 |
| mvrhel: I thought on re-fi's you either got money out or it was a break even. :-/ | 17:38.13 |
henrys | ray_laptop:I don't even know when the regular deposit date is. | 17:38.18 |
mvrhel | well I had to prime the pump for the escrow account | 17:38.31 |
| which i get back from the other lender | 17:38.39 |
| this is property tax payment time here | 17:38.50 |
Robin_Watts | marcosw: Will do. | 17:38.51 |
ray_laptop | henrys: Karen tells me 11th and 25th, but if those are on a weekend or holiday, it backs up to the previous business day | 17:39.00 |
| mvrhel: I see. Property taxes here are Dec 10 and Apr 10 | 17:39.27 |
mvrhel | ok. spring and fall here | 17:39.36 |
Robin_Watts | Interesting. So you pay tax on property twice a year? | 17:40.10 |
mvrhel | in washington we do | 17:40.23 |
Robin_Watts | At the moment in the UK, we pay 'stamp duty' when we buy somewhere, but that's a 1 off thing. | 17:40.35 |
mvrhel | property taxes vary from state to state | 17:40.36 |
| most states have three sources of revenue, property, sales, and income (plus business taxes). here in washington though there is no income tax. In oregon there is no sales tax. | 17:41.32 |
ray_laptop | mvrhel: tell me about it. And out of state owners don't get home owner's exemption, so it is easy to ding us without upsetting voters | 17:41.40 |
henrys | it's in my mortgage monthly. I don't know how it would go if I paid off the loan. | 17:41.51 |
mvrhel | yes, mine is wrapped up in the monthly mortgage too. | 17:42.45 |
henrys | mvrhel:no income tax wow! didn't know that. | 17:42.45 |
mvrhel | yes. that is nice | 17:42.51 |
ray_laptop | henrys: they generally transfer the impound account on re-fi, but sometimes the current lender makes you wait until they send you a check, so the new lender has to fill their impund acct | 17:42.53 |
mvrhel | yes. I should get a check from my old lender in about a month... | 17:43.29 |
henrys | I'm thinking about refi also... just lazy | 17:44.22 |
Robin_Watts | mvrhel: So the cunning plan would be to have a house just inside washington, and to shop in oregan? | 17:44.39 |
mvrhel | Yes the best place to live is in washington near oregon and do your shopping across the river | 17:45.10 |
Robin_Watts | marcosw: Committed. | 17:45.13 |
marcosw | thx | 17:45.17 |
Robin_Watts | Though there were changes of yours in there too. | 17:45.24 |
mvrhel | I always pick up a few things I need when I am in oregon | 17:45.33 |
| unfortunately here though, amazon includes taxes since their headquarters is in washington | 17:46.02 |
| they can't deny that one | 17:46.21 |
henrys | I wonder if prices aren't just a bit higher in Oregon. Vendors get as much as they can suck out of you... | 17:47.11 |
mvrhel | yes. probably | 17:47.36 |
| ray_laptop: did you send 2 files? | 17:48.28 |
henrys | mvrhel I assumed ray_laptop would be a better person for the ETS meeting, I didn't think you had looked at that code. | 17:49.31 |
mvrhel | henrys: absolutely | 17:49.41 |
henrys | alexcher can you take care of the jbig2dec memory issue along with luratech. It looks very straightforward to me (famous last words) - there is a struct already setup the memory members to be filled in. | 17:53.31 |
| alexcher_ ^^^ | 17:54.13 |
marcosw | Robin_Watts: I see you are brave enough to test my experimental cluster code :-) | 17:57.28 |
henrys | if you let these bugs sit for about 4 or 5 years they are easier to dispense with ;-) | 17:58.31 |
ray_laptop | mvrhel: It should be one .zip with two files in it (in the second email) | 17:58.31 |
mvrhel | ray_laptop: I never got a second email | 17:58.57 |
ray_laptop | mvrhel: sent 4 minutes after the first -- Subject has ____'s modified code (both files) | 18:00.05 |
marcosw | "We will fix no bug before its time." | 18:00.41 |
ray_laptop | mvrhel: calling ... | 18:02.08 |
mvrhel | ok | 18:02.13 |
henrys | marcosw:I've been spending the last couple days trying to prune my list. | 18:02.54 |
marcosw | I did that _before_ the staff meeting. | 18:03.19 |
ray_laptop | mvrhel: The code in gximage.c that I didn't send that does the conversion is: | 18:03.55 |
| lcmsicc_transform_color(info->icc_link, planes[0].data + k * planes[0].raster, | 18:03.57 |
| planes[0].data + k * planes[0].raster, info->plane_widths[0]); | 18:03.58 |
| mvrhel: that's in gx_image_plane_data_rows if the icc is needed | 18:04.29 |
henrys | mvrhel:I did copy you in on 693055 to check a gray to rgb mapping but I didn't do it in an obvious way. No hurry about that. | 18:06.59 |
mvrhel | henrys: ok. bbiab | 18:24.45 |
henrys | I'm down to 28 bugs, and everybody else's tally is up ;-) | 18:26.57 |
Robin_Watts | how many customer bugs? :) | 18:27.18 |
henrys | Robin_Watts:not much improvement on that front. | 18:31.16 |
malc_ | tor8: hi, around? | 18:32.02 |
tor8 | evening malc_ | 18:32.44 |
malc_ | tor8: god kvÐll | 18:33.07 |
| tor8 http://www.itu.int/rec/dologin_pub.asp?lang=e&id=T-REC-H.264-201201-I!!PDF-E&type=items | 18:33.14 |
ray_laptop | mvrhel: thanks for the call | 18:33.18 |
tor8 | malc_: writing a video decoder? ;) | 18:33.45 |
malc_ | tor8: links lead to unrelated places (in mupdf and xpdf too) not sure about acroread (don't have one), malformed? | 18:33.47 |
| tor8: messing with mpegts/es :) | 18:33.57 |
Robin_Watts | malc_: screwed in Acrobat reader too. | 18:34.50 |
malc_ | Robin_Watts: case dismissed, thanks | 18:35.42 |
Robin_Watts | np. | 18:37.21 |
ianneub | hi all. is it possible to get the text contents of an annotation when I have the fz_annot struct? | 18:50.22 |
| with mupdf | 18:51.07 |
tor8 | ianneub: not trivially, no. if you have the pdf_obj you can pdf_dict_gets(obj, "Content") to get the pdf string out, if it exists | 19:32.42 |
| adding that to the pdf_annot struct should be a fairly trivial task too | 19:32.58 |
| s/Content/Contents/ | 19:33.19 |
ianneub | ahh ok, I'm sure i can figure that out. thanks tor8 | 19:33.25 |
tor8 | I'd start by looking where annot->rect and similar are extracted, in pdf_load_annots | 19:34.17 |
ianneub | Cool thanks for the pointer, I'll checkout that function and work from there. | 19:36.21 |
malc_ | Robin_Watts: still here? | 19:38.03 |
Robin_Watts | I am. | 19:38.15 |
malc_ | Robin_Watts: does ARMv7 has mov reg, reg or it's substituted with something like ori rdest, rsrc, 0? | 19:38.57 |
| s;has;have | 19:39.06 |
Robin_Watts | sure does. | 19:39.06 |
malc_ | Robin_Watts: with full support for conditional bits? | 19:39.22 |
Robin_Watts | yes. | 19:39.28 |
malc_ | Robin_Watts: just saw a claim that it's those things are "not as fast as they used to be" on newer cores (conditional moves), your experience aswell? | 19:40.19 |
Robin_Watts | malc_: I can't say I've looked at v7 timings recently. | 19:41.11 |
| The condition flags used to be a really clever system, but as ARM has gained more pipeline stages, it's become more and more problematic. | 19:41.49 |
malc_ | Robin_Watts: aha, thanks, i seem to recall hearing that Thumb2 doesn't have conditional bits (or that they are restricted in one way or the other), correct? | 19:41.57 |
Robin_Watts | because you often have to wait for 1 instruction to move through enough pipeline stages for the condition bits for the next instruction to be known. | 19:42.43 |
| Thumb and Thumb2 instructions are 16 bits, so the conditionals are much more restrictive. | 19:43.13 |
| ianneub: If you get stuck, please come back and we'll try to help some more. | 19:44.16 |
ianneub | thanks Robin_Watts! You guys have been great to me the last few months I've been working on this project | 19:45.12 |
jen__ | howdy all. | 20:14.11 |
radistao | hi. Can sombody advice me on postscript EndPage usage? | 20:14.35 |
| Hi, Robin Watts. Can you help me? | 20:18.43 |
| guys, i have a problem: i need to watermark (put sample big text over the pages) every page in PDF document | 20:22.02 |
| for this reaso i use in postscript /EndPage function | 20:22.26 |
| << /EndPage { pop pop gsave /Helvetica_Bold 120 selectfont .85 setgray 130 70 moveto 50 rotate (Sample) show grestore true } bind >> setpagedevice | 20:22.59 |
| but this script creates for every page in input PDF additional empty page. So, output document will be twice bigger and every second page will be empty (but with watermark) | 20:24.01 |
| how avoid this pages duplicating? | 20:24.39 |
| or maybe you know more suitable solution? | 20:25.10 |
henrys | radistao kens will probably continue helping you on stackoverflow he's probably off for the evening. | 20:26.07 |
radistao | the same script with BeginPage works excellent, but watermark is under the page text. I need it over the page | 20:26.23 |
| @henrys: ok, thx | 20:26.50 |
henrys | bbiab | 20:27.53 |
| this xps stuff is awfully stringy | 21:14.54 |
Robin_Watts | radistao: For the logs - I can't help, I'm afraid. You should try again in about 8-10 hours (8-10am UK time) when Ken will be around again. | 23:06.53 |
| Forward 1 day (to 2012/09/27)>>> | |