| <<<Back 1 day (to 2013/05/26) | 2013/05/27 |
Robin_Watts | mvrhel_laptop: (For the logs). Will do tomorrow morning. | 00:16.21 |
unverbraucht | hi all | 08:56.01 |
kens | Hello | 08:56.25 |
ghostbot | salut, kens | 08:56.25 |
kens | Be quiet ghostbot | 08:56.33 |
unverbraucht | Robin_Watts: did you have an opportunity to check if the v8 lib works? | 08:57.07 |
kens | Robin_Watts : may well not be here, this is a public holiday in both the UK and US | 08:57.28 |
chrisl | paulgardiner: ping | 08:59.43 |
paulgardiner | hi chrisl | 08:59.56 |
chrisl | What version of VS 2012 did you use for the WinRT GS stuff? | 09:00.28 |
paulgardiner | Hang on. I'll just spark up the VM | 09:02.06 |
chrisl | I assume it was Express either "for Windows 8" or "for Windows Desktop" | 09:03.08 |
| Or one of the 90 day trial versions | 09:03.42 |
paulgardiner | Bought off Amazon, so not Express | 09:03.57 |
| Professional | 09:04.42 |
| I'll be able to see the version number in a minute. It's taking a while | 09:05.05 |
| 11.0.60315.01 | 09:05.46 |
chrisl | Crikey! 600 quid?!?! | 09:05.48 |
paulgardiner | Oh! No. Less than a 100. Let me look at my order history | 09:06.22 |
| Doh! Sorry | 09:07.03 |
| That was WIndows 8. VS 2012 is the one that Michael obtained. | 09:07.34 |
| It's on casper in my home dir | 09:07.47 |
chrisl | Oh, great | 09:08.01 |
| And it's "only" 1.5Gb - I thought it would be 3+ Gb | 09:10.18 |
paulgardiner | Yeah. Smaller than I was expecting too | 09:11.56 |
chrisl | They usually try to fill up the media | 09:12.26 |
unverbraucht | kens: ok thx, good to know | 09:15.21 |
Robin_Watts | unverbraucht: Hi | 09:19.40 |
| Yes, it seems to work. | 09:19.45 |
| so I need to check with henrys, but I think you can probably go ahead and claim the bounty. | 09:20.09 |
tor8 | Robin_Watts: (and mvrhel in the logs when he's come back tomorrow) I've found out some smart git tricks about dealing with whitespace issues | 09:24.06 |
| there are options in newer gits to automatically fix whitespace errors on commits | 09:24.22 |
| for mupdf: git config core.whitespace trailing-space,space-before-tab,tab-in-indent,tabwidth=4 | 09:26.23 |
| for mupdf: git config apply.whitespace fix | 09:26.29 |
| the second config there should convert spaces back to tabs (and fix other whitespace errors) using the first config settings | 09:26.57 |
Robin_Watts | tor8: But that will operate regardless of file type, right? | 09:27.11 |
tor8 | Robin_Watts: only on file types with the "text" attribute | 09:27.41 |
Robin_Watts | right, still not ideal. | 09:27.51 |
| mvrhel found a cunning VS trick. | 09:28.07 |
tor8 | no, not for ghostscript since indent with spaces doesn't mesh with makefiles | 09:28.10 |
| which files in mupdf need exceptions? | 09:28.22 |
Robin_Watts | We now both have shortcuts set up so that we can launch different options in VS for gs and mupdf. | 09:28.37 |
| That way we'll get TABs for mupdf and spaces for gs. | 09:28.46 |
tor8 | I'll keep looking, there must be a way to have exceptions per filetype (so we can exclude makefiles in gs) | 09:29.56 |
| then we could just let people do whatever and have git canonicalize the text on commit | 09:30.11 |
| and add a script "gs-git-setup" for convenience | 09:30.32 |
Robin_Watts | tor8: Third party code does not conform to tabs/spaces as we'd like in mupdf. | 09:30.56 |
tor8 | Robin_Watts: we don't commit to third party code from the mupdf git anyway | 09:31.24 |
Robin_Watts | the XML files in the mupdf android and winrt solutions doesn't follow tabs/spaces either. | 09:31.27 |
| tor8: true. | 09:31.36 |
tor8 | Robin_Watts: the latter could either be fixed, or excepted if I figure out how | 09:31.50 |
Robin_Watts | and the VS solutions are tabs based. Not sure how having them check out to be spaces based would work. | 09:32.20 |
tor8 | Robin_Watts: meh. apply.whitespace=fix doesn't work on git commit, only on rebase :( | 09:49.23 |
| so git commit with errors, then "git rebase -f HEAD~1" will clean the last commit from whitespace errors | 09:49.49 |
| Robin_Watts: in .gitattributes you can set different whitespace=...options... for files using wildcard patterns | 09:50.56 |
unverbraucht | Robin_Watts: excellent, I'll do that then. Glad it works :) | 09:59.10 |
| Erm, how would I go about claiming it? Just write a comment in the bug tracker? | 10:01.50 |
Robin_Watts | unverbraucht: That's part of what I need to check with henrys - essentially I think that we put a note in the bug that it's claimable, and you contact miles and he'll pay you. | 10:05.01 |
| but henrys is probably not about today as it's a holiday in the states. | 10:05.15 |
tor8 | Robin_Watts: I have a tentative pre-commit hook that will use the whitespace setting and fix errors | 10:17.56 |
paulgardiner | Having problems running mujstest on the cluster: make: *** No rule to make target `build/release/opj_bio.o', needed by `build/release/libopenjpeg.a'. Stop. | 10:25.53 |
Robin_Watts | paulgardiner: Have you git submodule ipdate --init ? | 10:27.48 |
| or update even. | 10:27.57 |
paulgardiner | yeah. Did that | 10:28.29 |
tor8 | on the cluster too? | 10:29.16 |
Robin_Watts | paulgardiner: Using clusterpush.pl or git cluster ? | 10:29.44 |
paulgardiner | ah! Didn't know about that | 10:30.08 |
| robin_watts: git cluster | 10:30.19 |
Robin_Watts | paulgardiner: Hmm. That should git submodule update for you. Let me run some tests. | 10:30.41 |
paulgardiner | ta | 10:30.48 |
Robin_Watts | I'm seeing the same failures here. | 10:34.10 |
| paulgardiner: Try now. | 10:42.58 |
| The git bridge script wasn't updating the submodules. Now it should do. Seems to work for me. | 10:43.20 |
paulgardiner | Thanks. I've set another run off. | 10:44.06 |
Robin_Watts | paulgardiner, tor8: So, my plan for today is to try and get a new build done for google play. | 10:46.17 |
| so we can get that up there before the show. | 10:46.25 |
| Do either of you have things you need to fix ? | 10:46.34 |
| sebras: was there an issue to with arabic ? | 10:46.51 |
paulgardiner | Nothing to fix, but I do have a commit I want to get in as soon as possible because it's likely to repeatedly conflict with others. On the other hand, it might be a bad time just before a release. Assuming it passes on the cluster, take a look and see if you think it's risky. | 10:50.04 |
sebras | Robin_Watts: yes, there were problems with searching for arabic. | 10:53.56 |
| couldn't really try since seeking was completely broken. | 10:54.18 |
| and I havent tried doing it myself since. | 10:54.26 |
Robin_Watts | sebras: Let me make a build and upload. | 10:54.33 |
henrys | unverbraucht, Robin_Watts: email miles.jones at artifex dot com with the bug # you fixed and tell him that it is your first bounty and ask him to send you the financial information Artifex needs to pay him. | 11:01.56 |
Robin_Watts | early morning, henrys? | 11:03.25 |
paulgardiner | henrys: fasting disturbing your sleep? | 11:04.21 |
henrys | my 11'th go at this race: http://bolderboulder.com/ | 11:06.05 |
| I ought to get a discount | 11:06.45 |
kens | What time does it start ? | 11:07.13 |
Robin_Watts | tor8: if you search for "is a registered", it doesn't find things. | 11:07.27 |
| should we treat multiple whitespace in the search string as a single one ? | 11:07.47 |
henrys | my wave is in 2 hours - it's big > 50,000 | 11:08.24 |
Robin_Watts | henrys: crumbs. | 11:09.18 |
paulgardiner | More of a tsunami | 11:09.19 |
kens | yeah that's a lot of people | 11:09.34 |
henrys | I think it's the 5th largest road race | 11:10.25 |
tor8 | Robin_Watts: possibly, we collapse multiple whitespaces in the text to be searched, so I think it makes sense to do the same on input | 11:11.13 |
Robin_Watts | Sponsored by Crocs. | 11:11.14 |
tor8 | Robin_Watts: even better would be some rudimentary wildcards in the search | 11:11.27 |
| like * and ? at least | 11:11.40 |
Robin_Watts | henrys: Helen says good luck. She's about to take the dog out for a cat hunt. | 11:11.53 |
| aka a run around the village. | 11:12.05 |
tor8 | full regexes would need a library or some non-trivial amount of code to get to work | 11:12.10 |
henrys | Robin_Watts:sounds like a cartoon | 11:12.35 |
Robin_Watts | henrys: The one piece of information I can't see on their site is, how long is the race? | 11:15.28 |
| I'm guessing a half marathon? | 11:15.56 |
| oh. 10K. | 11:16.55 |
henrys | 10 - it says BB10K - bolderBoulder | 11:16.56 |
Robin_Watts | sorry. | 11:17.03 |
henrys | like the entire world is supposed to know BB .. | 11:17.34 |
| I know the race director I'm going to bug him about the web page's low signal/noise ratio. | 11:18.59 |
Robin_Watts | henrys: I suspect he'll argue that with 50,000 participants already they can't afford any more signal :) | 11:19.40 |
| coo. My 10K times would qualify me. | 11:20.19 |
kens | Quick get on a plane :-) | 11:20.45 |
Robin_Watts | kens: Not til wednesday :) | 11:21.01 |
kens | had forgotten that.... | 11:21.17 |
| If you'd gone early you could join henry | 11:21.30 |
Robin_Watts | If I'd gone earlier I could have joined in the rerun of the boston marathon at the weekend. | 11:21.47 |
| or at least the last mile. | 11:21.51 |
henrys | Robin_Watts: it is one of the funnest races I do you'd be interested to know they have bacon on the course every year. | 11:21.59 |
| maybe next year. | 11:22.05 |
Robin_Watts | staff meeting in boulder next year ? :) | 11:22.21 |
paulgardiner | Bacon! Even I'd run a 10K for that. | 11:22.44 |
kens | Wrong time of year, no snow | 11:22.50 |
Robin_Watts | tor8: regexps sound too hard for now. | 11:22.59 |
| tor8: but I might look at collapsing white space. | 11:23.13 |
| Anyone with an android device, sebras, tor8, paulgardiner: http://ghostscript.com/~robin/MuPDF.apk | 11:23.41 |
henrys | heading out happy holidays all | 11:23.48 |
Robin_Watts | sebras: Does arabic.pdf work with that? | 11:23.53 |
| henrys: good luck! | 11:23.57 |
paulgardiner | henrys: yeah. Good luck | 11:24.05 |
Robin_Watts | henrys, kens: someone popped up with a pcl/pdfwrite problem at the weekend. I told them to open a bug. | 11:25.25 |
kens | Robin_Watts : I saw the log, they didn't open a bug | 11:25.37 |
paulgardiner | robin_watts: cluster looks stuck. Still running your test with one machine down, and all others in Standby or Idle | 11:35.02 |
Robin_Watts | tor8: So I've just committed a couple of changes (with alexchers blessing) to gs/jbig2dec. How long til it gets reflected in the standalone repo ? | 11:35.27 |
| paulgardiner: hmm, yes. | 11:35.53 |
| paulgardiner: I restarted the clustermanager. Hopefully it'll all shake out. | 11:39.54 |
| and I'll abort my job. | 11:40.13 |
tor8 | Robin_Watts: at midnight, unless you poke me | 11:48.44 |
Robin_Watts | tor8: poke! | 11:48.52 |
tor8 | the job takes a minute to run, since it has to extrac the history each time (or I haven't figured out how to get git subtree to be smarter and cache results) | 11:50.10 |
| Robin_Watts: anyway, it's done | 11:50.14 |
Robin_Watts | tor8: Thanks. | 11:51.13 |
tor8 | Robin_Watts: the scripts are in /home/tor/bin/mirror-jbig2dec.sh and mirror-tthirdparty.sh but they need to run as me (since they use temp directories in my home) | 11:51.17 |
| and the /home/git/jbig2dec.sh is set up to pull changes from those temporary directories | 11:52.08 |
| s/sh/git/ | 11:52.16 |
| paulgardiner: the java and xml files for android are 50/50 whether they use spaces or tabs for indentation | 11:53.10 |
Robin_Watts | paulgardiner: I think your job finished. | 11:54.48 |
paulgardiner | robin_watts: yeah. I realised it was broken just after you said you'd killed yours, but I thought may as well let it run because it wouldn't in any case get very far | 11:55.54 |
| tor8: Hmmm, I intended tabs for the Java. I thought I'd checked before commits. I guess not then | 11:56.38 |
tor8 | paulgardiner: want me to fix? | 11:56.53 |
| I've got some git hackery to deal with whitespace stuff in the pipeline | 11:57.05 |
| so I can do the cleanups as part of that | 11:57.12 |
| I have some git hooks to automatically fix whitespace errors on commit that I'm going to check in | 11:57.40 |
paulgardiner | tor8: yes great. | 11:57.45 |
tor8 | and an alias to do it manually | 11:57.46 |
| if it works out well, I can do the same for ghostpdl | 11:58.01 |
| Robin_Watts: where is the canonical version of memento.[ch] hosted? | 11:58.40 |
Robin_Watts | on gs or on mupdf. | 11:58.52 |
paulgardiner | Oh yeah. I started aligning things in the java before you told me of the MuPDF convention. That's where the spaces come from | 11:59.15 |
Robin_Watts | I try to keep the 2 in sync, but in general, it's whichever was committed last. | 11:59.16 |
tor8 | paulgardiner: I fixed those once, these spaces are in new files (AsyncTask.java and the other new ones) | 12:00.10 |
| Robin_Watts: right. so it makes sense to keep memento.[ch] indented with spaces then | 12:00.25 |
Robin_Watts | AsyncTask.java is an import from the android source, I believe. | 12:00.35 |
| tor8: oh, yes, please. | 12:00.53 |
| So is... another one whose name escapes me. | 12:01.08 |
tor8 | the vcxproj files are all spaces (with 2 space indents) | 12:01.17 |
Robin_Watts | Deque.java ? | 12:01.20 |
paulgardiner | tor8: One trouble with fixing white space is that it creates a point in the git tree around which diffs are difficult to interpret. | 12:01.51 |
tor8 | paulgardiner: diff -w | 12:02.40 |
paulgardiner | In my own projects in the past, I've used one of the git tools (the name of which I've forgotten) to rerun the whole development tree and make it look like correct spacing was always used. | 12:03.02 |
tor8 | but yes, batch changes of whitespace suck, which is why I'm now spending time to make sure we commit correct whitespaces | 12:03.06 |
| (prompted by the winrt branch) | 12:03.15 |
paulgardiner | tor8: ah right. That'll do it | 12:03.23 |
tor8 | paulgardiner: git filter-branch? | 12:03.39 |
| can't really do that with mupdf, it'd break too many people's repos | 12:03.49 |
paulgardiner | That's the one. | 12:03.50 |
| tor8: yeah. I'd guessed that was the case | 12:04.03 |
tor8 | I did try that with the ghostpdl tree when we first converted it from svn | 12:04.23 |
| but there were way too many files that needed exceptions for it to work | 12:04.35 |
| all makefiles, source files with strings using embedded tabs, postscript files, etc | 12:04.50 |
Robin_Watts | Also, the mupdf source is pretty small compared to gs, and it's very consistent already. | 12:07.53 |
| The source in gs is WAY more disparate. | 12:08.06 |
| oh, but you did a big global rewrite of gs, didn't you. | 12:08.39 |
paulgardiner | robin_watts: cluster looks stuck again | 12:15.26 |
Robin_Watts | alex_i7a says "starting jobs" | 12:16.07 |
| hopefully it should time out in a mo. | 12:17.16 |
| If it doesn't get moving by 13:30, I'll prod it. | 12:17.34 |
| I disabled alex_i7a and it recovered. | 12:24.34 |
| well, I thought it had recovered. | 12:26.49 |
paulgardiner | I'm off out. Back in a few hours | 12:39.11 |
Robin_Watts | oh, the android app doesn't use text_search.c :( | 12:46.27 |
tor8 | Robin_Watts: paulgardiner: whitespace hook (and batch whitespace fixes) on tor/master | 12:54.33 |
| run "bash scripts/gitsetup.sh" to install the hook | 12:54.47 |
Robin_Watts | looking now. | 12:55.54 |
| I see them not. | 12:56.28 |
| tor8: ^ | 12:56.55 |
tor8 | Robin_Watts: sorry. forgot to do the actual push... | 12:57.31 |
| I think I'm in need of feeding | 12:57.38 |
Robin_Watts | likewise. | 12:57.46 |
| ok, they all look good. will push. | 12:59.47 |
| tor8: pushed. And 2 commits for you to look at on robin/master | 13:05.11 |
| after you've eaten :) | 13:05.20 |
| We have problems searching for hypenated text. | 13:05.45 |
| but searching for "Content streams" now works, even across line boundaries. | 13:07.21 |
| http://ghostscript.com/~robin/MuPDF-18.apk <- armeabi build, should work on anything. | 13:07.51 |
| http://ghostscript.com/~robin/MuPDF-19.apk <- armeabi-v7a build | 13:24.14 |
| just reuploading MuPDF-18.apk now | 13:24.35 |
| and building an x86 version too. | 13:25.59 |
tor8 | Robin_Watts: the jbig2dec update commit also includes changes to openjpeg in the android build | 13:54.45 |
| I don't think you need both -DHAVE_STDINT_H and -DOPJ_HAVE_STDINT_H in ThirdParty.mk | 13:55.43 |
Robin_Watts | tor8: sadly, we do. | 14:14.22 |
tor8 | who needs the first? | 14:14.31 |
Robin_Watts | jbig2dec | 14:14.37 |
| on android. | 14:14.43 |
| the jbig2dec code doesn't define int64_t otherwise | 14:14.56 |
| tor8: Updated patch online - I changed the commit message. | 14:20.20 |
| tor8, sebras, paulgardiner: I am uploading 18/19/20 to google play at the moment. | 14:24.49 |
| so any testing you can do would be much appreciated. | 14:24.59 |
| Does anyone on here have an x86 based android device? | 14:25.16 |
sebras | Robin_Watts: 18/19/20? | 14:25.58 |
Robin_Watts | sebras: yeah, the links I posted above. | 14:26.19 |
| 18 = armeabi | 14:26.23 |
| 19 = armeabi-v7a | 14:26.31 |
| 20 = x86 | 14:26.34 |
sebras | oh, I missed them. | 14:26.39 |
tor8 | Robin_Watts: right. jbig2dec looks good now. | 14:26.51 |
sebras | no, no x86-based android here. | 14:26.52 |
Robin_Watts | Google play will serve the highest versioned apk it has that will work on a given device. | 14:27.17 |
sebras | Robin_Watts: do you know if it is possible download an older version? | 14:27.48 |
Robin_Watts | sebras: No, I have to remove older versions from being published in order to put the new ones up. | 14:29.35 |
| ooh, they have a new beta testing feature. | 14:30.37 |
sebras | ah, that explains why I can't download older versions. | 14:31.19 |
Robin_Watts | https://play.google.com/apps/testing/com.artifex.mupdfdemo <- Everyone with an Android device should visit that link and opt in. | 14:32.23 |
| That way you'll be on the beta test list. | 14:32.31 |
| but that link ain't working for me yet. | 14:33.54 |
| mvrhel_laptop: Morning | 15:14.40 |
| I pushed all your mupdf changes this morning. | 15:14.47 |
| Do either Miles or Scott have a WinRT device ? | 15:15.03 |
| ok. I have an x86 android emulator set up here, and that all works. | 15:18.51 |
| paulgardiner, tor8: Have either of you tested the android builds ? | 15:57.25 |
kens | OK goodnight everyone | 16:30.41 |
henrys | kens:goodnight | 16:31.01 |
| Robin_Watts: can I email shelly that alexcher is going to be the lead on the openjpeg stuff? or you can do it, let me know. | 16:31.57 |
Robin_Watts | henrys: What openjpeg stuff? | 16:32.13 |
| all the outstanding stuff has been checked by alexcher and gone in. | 16:32.26 |
| no, wait, that was the jbig2dec stuff. | 16:32.38 |
| So, sorry, what openjpeg stuff? | 16:33.26 |
| The OpenJPEG stuff that shelly/simon did for mupdf has gone into the mupdf repos. | 16:33.50 |
| I/We need to point alexcher at that and let him pull it over to gs when he's happy. | 16:34.11 |
henrys | yes I thought there was more and Shelly is asking about it. | 16:34.16 |
Robin_Watts | I/We/Shelly need to pass it back to openjpeg to see if they want the fixes. | 16:34.32 |
| AFAIK there are no more openjpeg problems (at least no more we know about). | 16:34.49 |
| There is still one outstanding jbig2dec problem. | 16:34.57 |
henrys | ah I thought more address sanitizer bugs were in openjpeg | 16:35.21 |
| so alexcher should contact the openjpeg folks - I'll remind him Tuesday. | 16:36.06 |
alexcher | Yes, I'm lurking here. | 16:36.15 |
Robin_Watts | henrys: I believe we have 1 remaining crash, in jbig2dec. | 16:37.18 |
| Some code accesses off the end of GR_stats. | 16:37.29 |
| but it's not entirely clear how to solve it, as GR_stats can be 2 possible sizes and the size is not available at the point at which the access is done. | 16:37.56 |
henrys | assign it to alexcher and he can fork it off to shelly or simon or work on it. | 16:38.48 |
Robin_Watts | alexcher: http://git.ghostscript.com/?p=thirdparty/openjpeg.git;a=shortlog;h=refs/heads/zeniko_fixes | 16:38.59 |
alexcher | OK | 16:39.13 |
Robin_Watts | alexcher: There are 12 fixes there that we've adopted for mupdf, that you may want to pull into gs too. | 16:39.21 |
| henrys: right. I will raise a bug for that in a mo. Am trying to get the mupdf android stuff polished before the show. | 16:39.53 |
henrys | Robin_Watts: I just got back should be able to download at test the google play stuff shortly | 16:41.11 |
Robin_Watts | henrys: Thanks. How was the race? | 16:41.25 |
henrys | Robin_Watts: best I've done at that particular race - it's a very tough course and I usually don't fair well. | 16:42.38 |
| a good 2 1/2 minutes slower than other 10k's I do in the area. | 16:44.10 |
| we could make it a a company meeting - lots of folks walk it. I really is just a huge party all moving in one direction. | 16:45.08 |
| s/I/It | 16:45.23 |
heoyea | is there a way to change the default Mupdf hotkeys? | 17:04.49 |
Robin_Watts | heoyea: On what platform ? | 17:06.09 |
heoyea | linux | 17:06.18 |
Robin_Watts | Look in apps/pdfapp.c and edit the large switch statement. Otherwise, no, sorry. | 17:06.41 |
heoyea | ic Thanks | 17:07.08 |
Robin_Watts | paulgardiner: I see the problem, I think. | 17:15.15 |
paulgardiner | ah... brilliant | 17:15.50 |
Robin_Watts | finds more fixes for mupdf waiting uncommitted on peeves. Rats. | 17:26.57 |
| paulgardiner: Yeah, I have a fix. Will put review up asap. | 17:29.07 |
paulgardiner | robin_watts: where did you say the updated v8 libs were? | 17:42.12 |
Robin_Watts | ghostscript.com/~robin/v8-3.9-android.... | 17:42.36 |
paulgardiner | ta | 17:42.49 |
Robin_Watts | I think the missing text for the delete might be because there is no android:layout_toLeftOf thing for the appropriate layout xml stuff. | 17:43.32 |
paulgardiner | Quite possibly | 17:44.03 |
Robin_Watts | android:layout_toLeftOf="@+id/inkButton" | 17:44.22 |
| android:contentDescription="@string/strike_out" | 17:44.24 |
| That seems suspicious too. | 17:44.30 |
| no, ignore me. | 17:44.44 |
| paulgardiner: Sorted! | 18:16.14 |
paulgardiner | Superb | 18:16.37 |
vtorri | hey | 18:20.36 |
| the implementation of fz_search_text_page() is the simplest way to find text in a pdf page ? | 18:20.57 |
Robin_Watts | That's the plan. | 18:21.15 |
vtorri | it's not so simple, actually :) | 18:25.01 |
| i suppose that the api is quite low level | 18:25.22 |
Robin_Watts | tor8, paulgardiner: 4 reviews for your delight and delectation on robin mupdf | 18:27.35 |
vtorri | :) | 18:27.51 |
Robin_Watts | Hi marcosw. | 18:28.24 |
paulgardiner | robin_watts: you spoil us! | 18:28.35 |
Robin_Watts | The cluster was misbehaving a bit this morning. I had to prod it a couple of times. | 18:28.38 |
vtorri | when searching a text, should I cache the fz_text_page or speed ? | 18:33.18 |
| for* | 18:33.24 |
Robin_Watts | vtorri: When would you ever reuse the same text_page ? | 18:33.48 |
vtorri | hmm | 18:33.58 |
| ok... | 18:34.01 |
paulgardiner | robin_watts: if (w <= -1) {} else if (w < 0) | 18:34.08 |
vtorri | i should think a bit before asking | 18:34.11 |
Robin_Watts | paulgardiner: There are 2 versions of draw_scale.c | 18:35.36 |
| draw_scale and draw_simple_scale | 18:35.41 |
| and I'd added that code to one, but not the other. | 18:35.50 |
| so this commit is just pulling the 2 into line. | 18:36.09 |
paulgardiner | oh okay | 18:36.24 |
Robin_Watts | Now, you can argue that I could have written that in a different (more legible) way and you may not be wrong :)( | 18:36.31 |
paulgardiner | Surely it never takes the second branch... or am I goign mad | 18:37.19 |
vtorri | which mediabox should I pass to fz_new_text_page() when calling fz_search_text_page() ? | 18:37.32 |
paulgardiner | My spelling suggest the latter | 18:37.35 |
| Oh floats!!!!! | 18:37.56 |
| Doh! | 18:38.04 |
Robin_Watts | yes, floats. | 18:38.04 |
| sorry. | 18:38.13 |
paulgardiner | First two are fine | 18:38.45 |
Robin_Watts | vtorri: You don't pass a mediabox in. | 18:39.08 |
vtorri | ? | 18:39.51 |
| i pass NULL ? | 18:40.02 |
Robin_Watts | vtorri:int fz_search_text_page(fz_context *ctx, fz_text_page *text, const char *needle, fz_rect *hit_bbox, int hit_max); | 18:40.52 |
| ctx = the context. | 18:41.02 |
| text = the page to search in | 18:41.07 |
vtorri | yes | 18:41.15 |
| text | 18:41.17 |
Robin_Watts | const char *needle = the thing to search for. | 18:41.18 |
vtorri | i have to create it | 18:41.23 |
| so i call fz_new_text_page | 18:41.35 |
| and the last call of this function is a rect | 18:41.46 |
Robin_Watts | vtorri: oh, I see, sorry. | 18:42.04 |
vtorri | last param, not last call | 18:42.05 |
Robin_Watts | fz_infinite_bbox | 18:42.09 |
vtorri | ok | 18:42.12 |
| thank you | 18:42.15 |
| robin_watts: isn"t it fz_infinite_rect instead ? | 18:45.42 |
Robin_Watts | vtorri: It is. sorry. | 18:45.57 |
vtorri | ok | 18:46.02 |
| btw, in the implementation of fz_search_text_page, you're using | 18:46.54 |
| if (strlen(needle) == 0) | 18:46.55 |
| maybe a better test would be if (!needle || !*needle) | 18:47.22 |
Robin_Watts | vtorri: possibly. | 18:50.19 |
| actually it should be a test that checks for needle not being entirely whitespace, probably. | 18:50.52 |
| http://ghostscript.com/~robin/MuPDF-30.apk (armeabi) is online now. | 19:39.03 |
| http://ghostscript.com/~robin/MuPDF-31.apk (armeabi-v7a) is online now. | 19:39.11 |
| http://ghostscript.com/~robin/MuPDF-32.apk (x86) is online now. | 19:39.22 |
| http://ghostscript.com/~robin/MuPDF-33.apk (mips) is online now. | 19:40.21 |
| Unless anyone finds any problems by tomorrow lunchtime (UK time), I'll hit the button to publish them. | 19:40.54 |
henrys | Robin_Watts: copy text doesn't seem to work in reflow mode - but that might be expected. | 19:52.43 |
Robin_Watts | henrys: yes, but it should be disabled :( | 19:53.30 |
| I'll talk to paul about that tomorrow. | 19:53.41 |
henrys | Robin_Watts:is the hardware floating point significant? I'm not seeing much difference in v7a vs older releases I've looked at. | 19:56.39 |
Robin_Watts | older releases were all v7a. | 19:56.55 |
henrys | oh okay | 19:57.11 |
Robin_Watts | and I think it's only really v8 that will be affected. | 19:57.30 |
| and we don't really care too much about the speed of that. | 19:57.42 |
henrys | ship it ;-) | 20:00.50 |
| Forward 1 day (to 2013/05/28)>>> | |