| <<<Back 1 day (to 2013/02/04) | 2013/02/05 |
kens | Ghostscript 6.50 ?? | 08:03.04 |
kens | boggles | 08:03.13 |
chrisl | Terrific! Maybe we can convince them to upgrade to Windows NT 4.0 at the same time...... | 08:10.47 |
saper | no, that would mean migrating off PowerPC | 08:48.33 |
chrisl | Or DEC Alpha..... | 08:48.47 |
saper | did try NT 3.51 once on IBM RS/6000 | 08:49.01 |
chrisl | Those were the days..... ;-) | 08:49.11 |
saper | one had a choice | 08:49.33 |
chrisl | You still have a choice - Intel, or nothing.... or use a "device" rather than a computer | 08:50.25 |
saper | there is ARM with everyone's favourite msm platform | 08:51.43 |
chrisl | Not many ARM workstations floating around these days | 08:52.09 |
kens | Hmm 5 cluster nodes down. | 09:13.30 |
| Marcos fell off the internets ? | 09:13.55 |
| Oh no 3 of them came back | 09:14.05 |
tor8 | Robin_Watts: there's a typo in your html... 12tp rather than 12pt | 09:17.41 |
| I think we have to use relative dimensions to allow zooming by changing the font size | 09:18.08 |
| that is, ems rather than pts | 09:18.20 |
saper | chrisl: an AVR "workstation" is lying on my desk :) running amforth | 09:36.16 |
chrisl | :-) | 09:36.57 |
Robin_Watts | tor8, paulgardiner : Morning | 10:17.56 |
paulgardiner | Hi. We have styles. | 10:18.07 |
Robin_Watts | paulgardiner: tor8 just pointed out a typo in my code. | 10:18.19 |
| 12tp rather than 12pt | 10:18.26 |
paulgardiner | Just got rid of a byte[] -> String -> byte[] transformation | 10:18.28 |
| Might that explain why it's huge | 10:18.44 |
Robin_Watts | paulgardiner: You mean you've taken my patch on? | 10:18.55 |
paulgardiner | Don't like the grey border, but besides that | 10:18.59 |
| Robin_Watts: yes | 10:19.06 |
| Just pushed the patch to get rid of the unnecessary conversions | 10:19.50 |
Robin_Watts | removing the grey border is easy. | 10:20.46 |
| mupdf.c | 10:20.48 |
| line 1308 | 10:21.02 |
| body{background-color:gray; | 10:21.33 |
paulgardiner | just make margin 0 and remove the gray? | 10:21.50 |
Robin_Watts | Whatever fits best with your scaling stuff. | 10:22.15 |
paulgardiner | Text is huge on my phone | 10:22.29 |
tor8 | nooo! not 0 margins... | 10:22.51 |
| 0.5 em or 1 em margins please! | 10:23.01 |
paulgardiner | There's still a margin between the edge of the screen and the text | 10:23.23 |
| The grey was extra | 10:23.31 |
tor8 | paulgardiner: ah, well in that case I retract my panic :) | 10:24.14 |
paulgardiner | Robin_Watts: were you testing on a tablet? | 10:24.14 |
Robin_Watts | no. on my phone. | 10:24.41 |
paulgardiner | Robin_Watts: strange the grey border was huge and so was the text, more so than I'd have imagined you intended. | 10:25.16 |
Robin_Watts | paulgardiner: I was aware it was not right. | 10:25.31 |
| ANd I'd not realised the border was my fault. | 10:25.45 |
paulgardiner | Ah good. So it's just adjustment needed not that it varies from device to device necessarily | 10:26.06 |
Robin_Watts | I gave up working last night cos I could feel a migraine coming on, so what you saw was just how far I got. | 10:26.15 |
paulgardiner | Looks great though. Styles make all the difference | 10:26.27 |
Robin_Watts | tor8: Where do you see that typo? I can't find it. | 10:26.28 |
paulgardiner | Unicode is still working, in that I still see "for all" signs and all sorts | 10:26.46 |
| Robin_Watts: think that typo was on the grey margin | 10:27.09 |
Robin_Watts | Ah. paulgardiner: line 1308 again ( same line) | 10:27.09 |
| yeah. | 10:27.15 |
| so zoom is still broken, right? | 10:27.30 |
paulgardiner | Yeah, that could be hell to make work | 10:28.24 |
Robin_Watts | tor8 had a suggestion about that just before you arrived. | 10:28.56 |
paulgardiner | Can you access style sheets in javascript, rather than the styles of individual tags | 10:29.07 |
| ? | 10:29.08 |
Robin_Watts | tor8: "I think we have to use relative dimensions to allow zooming by changing the font size" | 10:29.19 |
| tor: "that is, ems rather than pts" | 10:29.32 |
paulgardiner | It's working | 10:29.33 |
| I just retried the zoom style | 10:29.41 |
| Must have messed up when I tried it last | 10:29.52 |
Robin_Watts | really? | 10:29.54 |
paulgardiner | Yep | 10:30.06 |
Robin_Watts | I found that the gaps between lines moved, but the text size didn't. | 10:30.14 |
| but if it's working, fab! | 10:30.20 |
| Can I see an apk? | 10:30.24 |
paulgardiner | Yeah, that's what happened with fontSize, but the zoom style works. | 10:30.45 |
| When I tried before I attempted to set it to a number not knowing how to make it a percentage in javascript, but it turns out you just have to set it to a string, which as the digits of the number followed by a percent sign | 10:31.47 |
| Zoom is even more clunky now we have styles, but that's webview's fault | 10:32.32 |
| The view is still resizing correctly too. | 10:32.56 |
Robin_Watts | I found it VERY slow to get pages in when I tried it last night. | 10:33.22 |
| (which may be the extra conversions etc) | 10:33.38 |
| Flip forward a few pages, then wait a long time. | 10:34.11 |
| No 'loading' icon shown like we get from the PDF views. | 10:34.31 |
paulgardiner | Robin_Watts: I may be able to add that to the reflow view possibly. I think people will let us off for now though | 10:35.44 |
Robin_Watts | sure. | 10:35.56 |
| Maybe avoiding 'churning' would improve the speed? Might that be possible now ? | 10:36.20 |
paulgardiner | Coo. It looks really good | 10:36.25 |
Robin_Watts | Show me the money (apk)! | 10:36.48 |
paulgardiner | Robin_Watts: ah yes, of course. Now that the sizing is working | 10:36.49 |
| I'll just cure the churn | 10:37.05 |
| We just want to make it come up initially with smaller text. Not sure where best to do that. | 10:38.07 |
Robin_Watts | paulgardiner: Change the default zoom ? | 10:38.42 |
paulgardiner | But then I have to set zoom on first load which is awkward - hopefully not impossible, but can't currently | 10:39.32 |
| Even formula with subscripts look correct. How is that working? | 10:41.06 |
| Maybe superscripts wouldn't work | 10:41.29 |
| http://intranet.glidos.net/~paul/MuPDFReflow20130205.apk | 10:43.32 |
Robin_Watts | paulgardiner: Can we not just put 'zoom "25%"' in the body style ? | 10:57.38 |
| (downloading the apk now) | 10:58.01 |
paulgardiner | Yeah maybe I can just do that. I was just going to take a look at the styles we are producing, but I imagine they are actually correct for the normal page display, so we should stick with them | 10:59.33 |
| We have fonts with sizes 36pt and 48pt. Is that what we should expect? | 11:02.07 |
Robin_Watts | paulgardiner: apk looks good to me. | 11:02.43 |
| except I get occasional SEGVs. | 11:02.49 |
paulgardiner | Erk!! | 11:03.02 |
Robin_Watts | Start pdf_reference17.pdf | 11:03.16 |
| go into reflow mode. | 11:03.26 |
| Tap the right hand edge of the screen to "smart advance". Keep tapping. | 11:03.42 |
| Very neat that smart advance works in the webview too :) | 11:03.56 |
| All of a sudden it bales back to the menu. | 11:04.34 |
paulgardiner | Yeah, was meaning to say that. been finding smart advanvce very usefuk in both modes | 11:04.37 |
| SEGV or java exception | 11:04.55 |
Robin_Watts | It's a SIGSEGV | 11:05.03 |
| (so sayeth adb logcat) | 11:05.10 |
paulgardiner | In the library then, presumably | 11:05.17 |
| So do we think that the large point sizes on fonts like 36 and 48 are to be expected? | 11:09.04 |
tor8 | which fonts are that big? are they 36pt in the source document? | 11:12.55 |
paulgardiner | tor8: oh yeah, I guess that's worth checking :-) | 11:13.28 |
tor8 | the sizes you get stored in the fz_text_style depend on the matrix used to execute the page | 11:13.54 |
Robin_Watts | paulgardiner: I think we expect fonts that large for titles. | 11:14.09 |
| Like "PDF Reference" on page 1 of pdf_reference17.pdf | 11:14.21 |
paulgardiner | 36 was the smallest in the doc I was looking at | 11:14.36 |
Robin_Watts | I'm gonna go for a run, then I'll go hunting for the SEGV when I get back if you want. | 11:14.49 |
tor8 | well, large font sizes for titles should stay large | 11:14.51 |
paulgardiner | So should we use id matrix for text? | 11:14.59 |
tor8 | we're doing reflow not cleanup | 11:15.01 |
Robin_Watts | paulgardiner: That sounds odd. Unless it's a bigprint book :) | 11:15.05 |
tor8 | paulgardiner: yes. | 11:15.06 |
| paulgardiner: yes, to identity matrix in text extraction. | 11:15.20 |
paulgardiner | I think we were doubling them | 11:18.37 |
| Slightle more in fact | 11:18.48 |
| That's better | 11:22.14 |
| Robin_Watts: so far not been able to reproduce the SEGV | 11:45.21 |
kens | lunches | 12:11.53 |
Robin_Watts | paulgardiner: Guess it's down to me to find it then :) | 13:04.43 |
| This version seems MUCH faster than the last one. | 13:05.03 |
paulgardiner | Yeah, big SEP field around it | 13:05.10 |
| Might be removing the conversions and not churning | 13:05.29 |
| Sheridan was correct for hotel in Miami? | 13:06.08 |
Robin_Watts | yes. | 13:06.59 |
| so can I pull the latest version from your reflow branch? | 13:12.11 |
paulgardiner | yep | 13:20.47 |
Robin_Watts | Damn. The crash is in webcore. | 13:49.47 |
| maybe it's to do with rapid recycling? | 13:50.18 |
| nope, even when I go slowly. | 13:51.12 |
Seus | what am i missing with ghostscript output? I'm trying to run this commandâ¦receive no errorsâ¦yet there's no output: gs -o file%d.jpg main.pdf | 14:10.24 |
kens | no device specified | 14:11.02 |
| try adding -sDEVICE=jpeg | 14:11.15 |
| with that line you should get the dispplay device, possibly you built without X | 14:11.36 |
Seus | that did the trick! | 14:12.17 |
| on osx it would show up in X11â¦on my linux box, i just installed via apt | 14:12.33 |
kens | Well, we can't help what package maintainers do | 14:12.52 |
paulgardiner | Robin_Watts: I pushed a patch that may fix that SEGV | 14:53.11 |
Robin_Watts | paulgardiner: Trying now. | 14:58.56 |
| paulgardiner: Are you getting the tech@ mails now ? | 14:59.35 |
paulgardiner | yeah thanks | 15:00.02 |
Robin_Watts | mvrhel_laptop: I thought you were camping today? | 15:00.08 |
henrys | paulgardiner: if not obvious we'll skip the meeting today. | 15:00.39 |
paulgardiner | henrys: oh good. I'd forgotten. :-) | 15:01.02 |
Robin_Watts | paulgardiner: Nope, still crashes :( | 15:01.08 |
paulgardiner | Robin_Watts: ah no!! | 15:01.15 |
Robin_Watts | and it still crashes when I comment out the requestHeight from the onPageFinished thing. | 15:03.23 |
| It's the requestHeight in setScale that's killing it I think. | 15:05.07 |
| If I comment that out, it's solid. | 15:05.14 |
paulgardiner | The requestHeight not the other loadUrl? | 15:12.55 |
| Maybe conjoining them into one call | 15:13.04 |
Robin_Watts | the requestheight, yes. | 15:13.06 |
| I just tried that. still fails. | 15:13.12 |
| It's the window.HTMLOUT.reportContentHeight() that fails. | 15:15.04 |
| if I change that to be var fred = (); it's fine. | 15:15.16 |
| (i.e. the expression is not the problem) | 15:15.22 |
paulgardiner | But did it not work with the first LoadUrl in place and the requestHeight removed? | 15:16.05 |
Robin_Watts | it worked fine with the fist loadURL in place and the requestHeight removed. | 15:16.31 |
| so I tried to roll the requestHeight into the first loadURL and that failed. | 15:16.48 |
| so I nobbled the combined expression to replace: window.HTMLOUT.reportContentHeight with var fred= and it works fine. | 15:17.29 |
paulgardiner | Maybe it's the call to the java-side javascript interface | 15:17.47 |
Robin_Watts | presumably that can't be an immediate call? | 15:18.05 |
| i.e. it must take time ? | 15:18.16 |
paulgardiner | I'd think so | 15:18.24 |
Robin_Watts | so maybe that gets queued, and then we remove the page, then it goes off and boom. | 15:18.45 |
paulgardiner | Did you say you tried not reusing the views? | 15:19.42 |
Robin_Watts | I wouldn't have a clue how to do that :) | 15:19.55 |
paulgardiner | In MuPDFReflowAdapater | 15:20.15 |
| change if (convertView == null) to if true | 15:20.38 |
| If that works, there's an alternative we can try. | 15:22.56 |
Robin_Watts | That works. | 15:23.29 |
| pages are white for longer, but it's still no where near as slow as it used to be. | 15:23.54 |
paulgardiner | I'd probably better describe the alternative on the phone. | 15:24.42 |
Robin_Watts | ok. | 15:24.46 |
| helen is on the phone. I'll call in a mo. | 15:24.56 |
paulgardiner | ta | 15:25.25 |
marcosw | henrys: I'm going to be ~15 minutes late for the 9:00 meeting today. | 15:53.49 |
henrys | okay | 15:54.14 |
Robin_Watts | paulgardiner: I've rewritten the gtt history a touch and pushed to robin/reflow | 16:00.04 |
| If we're not having a meeting, I will just pop to the post office. | 16:01.13 |
paulgardiner | Robin_Watts: still there? | 16:01.29 |
Robin_Watts | I am. | 16:01.34 |
paulgardiner | I also had one more commit on my branch | 16:01.52 |
Robin_Watts | did you? | 16:01.59 |
paulgardiner | Android: remove secondary padding in reflow mode | 16:02.30 |
Robin_Watts | I rolled that one back. | 16:02.40 |
| I mean, that was part of the git rewriting I did. | 16:02.49 |
paulgardiner | Ah right | 16:02.59 |
mvrhel_laptop | Robin_Watts: we are leaving in a few minutes | 16:03.45 |
Robin_Watts | I'll do a release build and put it up for people to try. | 16:04.16 |
mvrhel_laptop | not camping, but a "field trip" here is any outing for a school class room | 16:04.23 |
Robin_Watts | Ah. The cardboard box factory. | 16:04.40 |
| (If you're a Simpsons fan) | 16:04.47 |
mvrhel_laptop | in this case we are off to Olympia to see the state govt. and maybe meet the governor of washington state | 16:04.50 |
| a little better than the cardboard box factory :) | 16:05.05 |
paulgardiner | Robin_Watts: all looks fine | 16:05.18 |
mvrhel_laptop | heading off... | 16:05.38 |
Robin_Watts | paulgardiner: I'll get that pushed when I get back. | 16:05.47 |
| henrys: Miles made a quote from the roadmap thing, so that's all done and dusted now, I believe. | 16:07.04 |
tor8 | oops. forgot the time. meeting time yet? | 16:19.36 |
kens | cancelled :-P | 16:19.47 |
| wyour one | 16:19.53 |
henrys | tor8:you are working non stop and don't have time for a meeting | 16:20.21 |
tor8 | henrys: fab! | 16:20.30 |
Robin_Watts | tor8: you have been eaten by a grue | 16:21.23 |
| http://ghostscript.com/~robin/MuPDF.apk | 16:37.38 |
| That's the latest version with reflow enabled in, as a release build. If people could try it out, I'd be grateful. | 16:39.36 |
paulgardiner | Robin_Watts: Works with xps files too of course | 16:42.31 |
tor8 | Robin_Watts: do you have pdfref13.pdf around? | 16:42.37 |
Robin_Watts | tor8: possibly somewhere. | 16:42.50 |
| why ? | 16:42.52 |
tor8 | page 23 has an extra tiny "2" just left of the "Overview" heading | 16:43.05 |
Robin_Watts | I see stuff like that on pdf_reference17.pdf I think. | 16:45.37 |
tor8 | I just installed dropbox on my tablet. hopefully that's more convenient than using the built-in browser. I really hate typing on a touch screen. | 16:46.17 |
Robin_Watts | Page 47 of pdf_reference17.pdf for example. | 16:46.17 |
tor8 | Robin_Watts: yeah. exactly the same. | 16:47.02 |
Robin_Watts | and I see the same tiny '3' with the text extraction device. | 16:47.11 |
tor8 | right. so either it's there or we have a bug in the core extraction device. | 16:47.41 |
Robin_Watts | well, there seems to be an fz_fill_text call for it. | 16:54.46 |
tor8 | Robin_Watts: is it invisible text mode perhaps? | 16:58.46 |
| brb | 16:58.47 |
henrys | We should have a very quick meeting - marcosw late, michael out, Robin and Tor working on stuff... | 16:58.53 |
kens | :-) | 16:59.00 |
| Quick meetings are good meetings | 16:59.07 |
ray_laptop | agrees | 16:59.19 |
Robin_Watts | henrys: Our mad panic is over, I believe. | 16:59.21 |
ray_laptop | so where are we w.r.t. the release ? | 16:59.38 |
henrys | Robin_Watts: be quiet I'm shortening a meeting | 16:59.51 |
Robin_Watts | tor8: wmode = 0 ? | 16:59.58 |
| sorry. | 16:59.59 |
alexcher | lcms2 doesn't compile on ppc32. | 17:00.11 |
chrisl | ray_laptop: we're waiting on marcosw finishing a "weekly" test run | 17:00.32 |
henrys | alexcher:I think we should create a bug but not hold up the release. | 17:00.43 |
| chrisl:yea actually marcosw is the person we need to talk this meeting but he's late | 17:01.27 |
| other than the release I don't have anything. | 17:01.40 |
chrisl | henrys: He's not due to finish until, I think, Thursday, or maybe tomorrow | 17:01.56 |
henrys | I think the 7 day business really has to change - way too long | 17:02.24 |
ray_laptop | alexcher: henrys: that may be an issue for cust 532, but we can easily support them directly for build issues. They are pretty good about fixing those kinds of things, too | 17:02.27 |
henrys | ray_laptop:oh right | 17:03.08 |
ray_laptop | henrys: I agree. even if we have a "nightly" run in nice 20 mode on each of the clusters, we'll be done in a day | 17:03.14 |
Robin_Watts | tor8: It's printed in white (! g) | 17:03.18 |
| (1 g) | 17:03.22 |
chrisl | alexcher: we had this with 9.06, I think - didn't we fix it then? | 17:03.38 |
ray_laptop | henrys: but to be clear, I don't think ppc is worth holding up to release (as you said) | 17:03.41 |
henrys | alexcher copy in Robin_Watts and he'll report upstream also (ppc compile problem) | 17:04.15 |
ray_laptop | cust 532 has 9.06 running on PPC 32-bit. They didn't mention anything | 17:04.24 |
henrys | alexcher: is it some obvious setting in a configuration file or what? | 17:05.16 |
chrisl | alexcher: http://git.ghostscript.com/?p=ghostpdl.git;a=commitdiff;h=004aeab5298b0a42ff08a1c9e626f41cf799039c has gone awol | 17:06.07 |
alexcher | henrys: I have a fix. I'm opening a bug report and attaching the fix. | 17:06.17 |
ray_laptop | alexcher: OK. Great! | 17:07.04 |
henrys | alexcher is this the fix chrisl is pointing us to? | 17:07.39 |
| awol? | 17:08.01 |
kens | Absent Without Leave | 17:08.08 |
henrys | yes I was in the military just trying to figure out how it relates to a commit | 17:08.32 |
alexcher | henrys: yes, this time I've coded it a bit differently but it doesn't matter. | 17:08.34 |
chrisl | henrys: I guess it wasn't noticed when we updated lcms2? | 17:08.55 |
henrys | oh got it. | 17:09.23 |
| so that is a problem release wise - you are going to break 532 | 17:09.43 |
| assuming that commit made it into the August release | 17:10.24 |
chrisl | Oh, yeh, 'cause they are obsessive about not using patched code!! | 17:10.27 |
Robin_Watts | Surely we can cherry pick that in? | 17:10.53 |
henrys | and restart the 7 day count? | 17:11.07 |
chrisl | The fix was in 9.06 - we can pull the fix in without bothering marcosw's testing, since he only tests on platforms with a 64 bit data type | 17:11.14 |
Robin_Watts | It's not going to affect marcosws testing as he's not using big endian systems. | 17:11.19 |
chrisl | Robin_Watts: it's not an endian issue | 17:11.58 |
henrys | reviewing the patch chrisl points to it not restricted to big endian | 17:12.11 |
Robin_Watts | ok, then what you said. | 17:12.16 |
chrisl | ;-) | 17:12.20 |
ray_laptop | chrisl: he does do some 32-bit tests, that's where the spate of 32-bit build bugs came from. But we didn't stop the release for them | 17:12.24 |
chrisl | ray_laptop: yes, but x86 linux and x86 windows both have 64 bit integers, which ppc32 linux doesn't always | 17:13.13 |
ray_laptop | but since there were no build issues on x86 32 bit, I wouldn't hold things up for PPC 32 | 17:13.17 |
henrys | It doesn't really matter it's only feb 5 | 17:13.32 |
alexcher | chrisl: the compile problem happens only on 32-bit big endian platforms. | 17:13.44 |
henrys | I think if you restricted the patch to only affect those systems I'd be happier with a cherry pick and no test. | 17:14.30 |
chrisl | alexcher: oh, okay, so there is an endian aspect. But it needs more than just a 32 bit platform - it needs a platform without a usable 64 bit integer type | 17:14.57 |
henrys | OTOH releasing FEB 12 is fine by me. | 17:15.01 |
| I can put off the stupid newsletter | 17:15.13 |
alexcher | chrisl: Probably, missed int64_t by the configuration. | 17:15.46 |
chrisl | BTW, once again, *my* PPC32 box builds just fine both on Linux and MacOS | 17:16.02 |
| alexcher: btw, did you report that problem to Marti? | 17:16.48 |
henrys | what os and version is alexcher using? | 17:16.53 |
marcosw | On the assumption there won't be a whole bunch of bitmap differences I can do the weekly testing in 3 1/2 days. It's about twelve hours to run each test not including the fuzzy compare step, so can do two weekly tests per day. | 17:17.04 |
chrisl | henrys: IIRC, it's an old-ish debian PPC32 that has the problem | 17:17.29 |
alexcher | Not yet, gcc 4.3.2, Linux ppc32 2.6.26-1-powerpc #1 Sat Jan 10 14:00:38 CET 2009 ppc GNU/Linux | 17:17.36 |
chrisl | Surely reporting stuff like this upstream should be a priority | 17:18.40 |
henrys | yes, this needs to be reported now and should have been reported when we patched our system. | 17:19.22 |
| August 1 | 17:19.31 |
| I thought Robin_Watts talked to Marti but if you want me to do it let me know and I'll do it, but let's do that today | 17:20.27 |
chrisl | I wonder if we lost the valgrind quieting changes in lcms2 as well...... | 17:20.55 |
henrys | so much for the short meeting | 17:21.01 |
Robin_Watts | henrys: I've talked to Marti before about various things, but I was unaware of this fix. | 17:21.12 |
| I can forward it on if you want, or alexcher can. | 17:21.24 |
chrisl | alexcher: do you think your current fix is better than the original? | 17:22.20 |
alexcher | chrisl: No, the first one looks better. | 17:23.57 |
chrisl | Okay, so we cherry pick that onto master and onto the release branch? | 17:24.20 |
henrys | I do think it makes sense that michael should own lcms and talk to marti in the future. If we all agree with that can marcos make a color management bug component defaulted to michael? | 17:24.51 |
| marcosw ^^^ | 17:25.44 |
chrisl | henrys: fine by me, but then if any of us make changes in there, we need to make sure that Michael is aware of them - and not just assume he'll see the commit..... | 17:25.53 |
henrys | chrisl:yes I don't know if this will completely solve the problem but it seems a step in the right direction. | 17:26.53 |
tor8 | Robin_Watts: oh great. wonder why they do that! | 17:27.15 |
chrisl | henrys: yeh, I agree | 17:27.24 |
Robin_Watts | tor8: Not a clue, but they're aint much we can do about it :( | 17:27.35 |
henrys | I don't think we are very good at dealing with third party stuff because we just don't do it often. | 17:27.35 |
tor8 | Robin_Watts: but once we get colors in the styles (should be easy enough to add) we can mimic that too | 17:27.36 |
Robin_Watts | tor8: true. | 17:27.46 |
henrys | anyway kens did you have anything meeting-wise almost time for you to get going. | 17:28.09 |
tor8 | Robin_Watts: I'm banging on the RTL stuff. I'm going to have to suck in a unicode character database. | 17:28.14 |
kens | henrys, nothgin from me | 17:28.23 |
| Except to note that customer #1 is sending me questions relating to the work they want to do for pdfmarks | 17:28.44 |
Robin_Watts | tor8: Ok. Is that going to be large? | 17:28.44 |
tor8 | Robin_Watts: fortunately it's a pretty small one! adds only about 100k | 17:28.46 |
Robin_Watts | ok. | 17:28.50 |
henrys | anybody who has android devices please help out with testing. Robin_Watts has posted a new apk | 17:29.06 |
tor8 | and it will let us do unicode normalization and some other stuff we ought to be doing in the text extraction | 17:29.08 |
Robin_Watts | henrys: I've found a problem with it (not huge, but...) | 17:29.24 |
| When you zoom the reflowed screen on my tablet the page changes size instantly, but the text takes a bit longer. | 17:30.05 |
| and sometimes I have to tap the screen to make it right itself. | 17:30.23 |
| paulgardiner has an idea what it might be. | 17:30.33 |
henrys | for the next few releases can you post apk test releases to tech email? | 17:30.40 |
Robin_Watts | Sure. | 17:30.48 |
henrys | have a good week everyone meeting over 10:31 | 17:31.17 |
| I cursed it by saying it would go fast | 17:31.33 |
chrisl | Oh, just quickly - I'm taking tomorrow off | 17:31.40 |
henrys | as long as your back in 3.5 days | 17:32.02 |
| ;-) | 17:32.05 |
chrisl | we'll see...... | 17:32.18 |
kens | goodnight all | 17:32.59 |
chrisl | alexcher: I've pushed the cherry-picked fix to master and gs907 if you want to check it | 17:33.32 |
alexcher | I'll check how it compiles and run a few tests. | 17:34.21 |
henrys | chrisl:will you check if we lost other patches? | 17:34.42 |
chrisl | henrys: I'm not terribly familiar with the lcms code...... | 17:35.17 |
henrys | chrisl:so maybe a bug for michael to review that? | 17:37.17 |
| I'll create one. | 17:37.36 |
chrisl | alexcher: actually, I'm not at all sure about that patch - it dereferences the pointer *and* uses array indices..... | 17:38.00 |
| henrys: the last lcms import was Robin_Watts, but I guess he's tied up with mupdf just now | 17:38.32 |
Robin_Watts | chrisl: just use the git webview thing to look at the history of the lcms files from before the last import | 17:39.37 |
alexcher | chrisl: I need to check how the old patch works. | 17:39.53 |
henrys | the patch also looks wrong because it looks like it effects all 32 bit machines am I missing something? | 17:41.33 |
| that is not what the comment says it should do. | 17:41.53 |
| s/comment/log message/ | 17:42.05 |
chrisl | henrys: it's in the #else section..... | 17:42.32 |
| ... of #ifndef CMS_USE_BIG_ENDIAN | 17:42.54 |
henrys | oh I am just looking at the email | 17:43.54 |
| sorry | 17:44.20 |
chrisl | Yeh, I don't like the use of ifndef in this context, but...... | 17:44.36 |
| OKay, I *think* there's only the one lcms2 patch that went missing in the update - another isn't there, but the new code does the same thing slightly differently..... | 17:47.56 |
malc | Robin_Watts, tor8: another naming "issue" textchar still has a bbox field, it is a bbox, but it's type is a rect, ought to cause some confusion http://www.youtube.com/watch?v=_f_p0CgPeyA | 17:48.21 |
Robin_Watts | chrisl: Got a link? | 17:48.28 |
chrisl | Actually, maybe not..... http://git.ghostscript.com/?p=ghostpdl.git;a=commitdiff;h=c2a982ff | 17:49.22 |
| Robin_Watts: it makes no functional difference, it's only to quieten valgrind | 17:50.26 |
| alexcher: so, on PPC 32 bit Debian 6.0.3 (Squeeze) with gcc 4.4.5 I didn't see a problem - perhaps you should update ;-) | 17:58.57 |
| Robin_Watts: I think I'm going to open a bug for Micheal to review that LCMS2 patch and decide whether to reapply or do it some other way, then he can deal with Marti on it - sound fair? | 18:03.10 |
tor8 | Robin_Watts: how about renaming fz_bbox to fz_irect to reduce confusion when we merge in your pass-by-reference patch? seeing as that will require api changes for clients, it'd be a good time for it. | 18:10.54 |
| malc: yes, I am aware of it. | 18:11.06 |
| but in general, everything except pixmaps use the fz_rect type, and fz_bbox should be called fz_irect instead. | 18:11.34 |
Robin_Watts | tor8: yeah, why not. | 18:16.12 |
| chrisl: OK. | 18:18.52 |
chrisl | Ta] | 18:19.01 |
| Done, hopefully it'll be an easy one for Michael! | 18:20.24 |
JakeSays | so i need to alter some pdfs to create test documents. basically i need to drop specific objects. any idea how i'd go about doing that? | 18:21.53 |
Robin_Watts | You want to remove certain objects from the files? | 18:28.11 |
| Or you want to keep all but certain objects ? | 18:28.21 |
| Acrobat pro can let you delete objects. | 18:28.35 |
ray_laptop | JakeSays: you can use mutool clean -d to make an editable PDF and then modify it. I change all the characters in the area that I want to omit to blanks. That way the xref stays OK. | 19:09.00 |
| JakeSays: if you delete objects from a contents stream then you can make a new pdf with mutool -d -ggg orig.pdf new.pdf The -ggg deletes unreferenced objects (or is supposed to), but if they are still named in Resource dicts, it may not | 19:11.24 |
Robin_Watts | When ray_laptop says "an editable pdf" he means "one you can edit in an editor". It relies on you knowing something about the PDF format. | 19:12.06 |
ray_laptop | JakeSays: much more "brute force" than using Acrobat, but you know exactly what is happening | 19:12.26 |
Robin_Watts | but the act of mutool cleaning changes the compression used in the files, which may be what you want to test. | 19:12.42 |
ray_laptop | Robin_Watts: true. But doesn't everybody ? ;-) | 19:12.51 |
Robin_Watts | But it's the way that I edit files, personally. | 19:12.53 |
| :) | 19:13.02 |
JakeSays | Robin_Watts: i tried acrobat pro, but it wouldnt let me delete anything | 19:13.09 |
Robin_Watts | JakeSays: You need to use the... touch up tool, I think it is. | 19:13.28 |
JakeSays | i couldnt even get that to work | 19:13.39 |
| Robin_Watts: but your idea of mutool clean and then messing up the characters would work well | 19:14.00 |
ray_laptop | JakeSays: send a bug report to Adobe ;-) (but don't hold your breath) | 19:14.08 |
| I thought that was _my_ suggestion | 19:14.31 |
JakeSays | ray_laptop: lol yes, that was directed at you - was too quick with the tab key | 19:14.56 |
ray_laptop | just kidding. | 19:15.08 |
| JakeSays: good luck. I use 'gvim' on Windows (since I know vi/vim so well). It doesn't mess up the binary parts of the file (thinks that -d leaves binary) | 19:16.08 |
JakeSays | but yeah - good answers. its just what i need. just need to tweak pdfs enough to cause my tools to fail. | 19:16.13 |
ray_laptop | JakeSays: you WANT your tools to fail ??? :-/ | 19:16.39 |
JakeSays | ray_laptop: well, this is for unit testing the tool | 19:17.01 |
| so i need corner cases, etc | 19:17.06 |
ray_laptop | JakeSays: OK. Well, good luck! | 19:17.25 |
JakeSays | thanks! | 19:17.30 |
| Forward 1 day (to 2013/02/06)>>> | |