| | <<<Back 1 day (to 2014/06/26) | 2014/06/27 |
| pedro_mac | morning all | 06:49.22 |
| chrisl_away | good morning | 06:50.01 |
| chrisl | How's sunny Glasgow? | 06:50.28 |
| pedro_mac | uncharacteristically bright :) | 06:55.18 |
| kens | You're up early pedro_mac | 06:55.32 |
| chrisl | Nice! It's dull, overcast and possibly raining here..... perfect for a few days away! | 06:55.46 |
| pedro_mac | chrisl: (Iâm @home just now, but heading through in 5 mins) | 06:56.04 |
| kens | chrisl, amazingly fast fix last night! | 06:56.05 |
| pedro_mac | cool | 06:56.07 |
| chrisl | kens: I spotted the problem literally as you logged off for the evening! | 06:56.38 |
| pedro_mac | chrisl: hopefully IoW will miss a lot of the inland downpours⦠| 06:56.45 |
| chrisl | pedro_mac: we'll see - if not, we can just stay in the pub! | 06:57.09 |
| pedro_mac | result :) | 06:57.16 |
| chrisl | As long as the beer's good | 06:57.43 |
| | kens: of course, the fix is another one-liner | 06:58.25 |
| kens | Yeah, but the trick is to know *which* line..... | 06:58.39 |
| chrisl | Well, once I saw a restore freeing the memory while the ref count was 30, it was a strong hint where the problem was | 06:59.17 |
| kens | I suppose I should have guessed that myself, but I'm not so familiar with the memory stuff | 06:59.46 |
| | I feel sure this was the same problem I saw earlier in the day and then couldn't reproduce | 07:00.08 |
| chrisl | Possibly the confusion was the fact it's reference counted! I can't say I endorse the idea of reference counted objects in garbage collected memory | 07:00.40 |
| kens | Yeah, our mulitple memory managers just sow confusion here | 07:00.58 |
| chrisl | But moving it to non-gc memory was a much bigger fix, so I opted for the simpler one | 07:01.42 |
| kens | Seems like the right move anyway | 07:01.55 |
| chrisl | Right, everything else is shutdown, so I'll be off now - back on Tuesday....... | 07:02.54 |
| kens | have a good time | 07:03.01 |
| chrisl | Thanks - bye! | 07:03.07 |
| mattchz | morning | 09:10.51 |
| | fwiw, SOT (app store version) seems to work fine on the Android L preview (in my very limited testing) | 09:12.45 |
| pedro_mac | cool - we should be doing a refresh pretty soon so itâll be good to test on L as well | 09:20.04 |
| mattchz | MuPDF seems fine too. | 09:24.22 |
| | interestingly, SOT completely massacres rendering the annotation that I added in MuPDF :) | 09:24.42 |
| pedro_mac | worth raising a bug on that I guess with the test file | 09:25.33 |
| mattchz | Iâm not sure which app is to blame :) | 09:25.57 |
| pedro_mac | grins | 09:26.04 |
| mattchz | and SOT just crashed. | 09:26.13 |
| pedro_mac | if it renders ok in adobe then I âd point at SOT | 09:26.24 |
| mattchz | Iâve seen some bug reports about stuff not rendering properly in Adobe too, so Iâm not sure about this. It renders fine in QuickOffice | 09:27.01 |
| kens | mattchz : if you put the PDF somewhere I can look at it with GS | 09:43.33 |
| mattchz | sure, no problem. | 09:44.49 |
| | bug report for the crash: http://bugs.ghostscript.com/show_bug.cgi?id=695338 | 09:45.46 |
| kens | ok give me aminute, got another report to look at also | 09:46.27 |
| mattchz | kens: thatâs not got the PDF in it. | 09:46.51 |
| | Iâll send you that separately. | 09:46.54 |
| kens | Oh, that's just the logcat, not the PDF file | 09:46.58 |
| | ahOK thanks | 09:47.02 |
| mattchz | yeah, thatâs a crash, rather than a rendering problem | 09:47.09 |
| kens | Indeed, but I can look at the PDF file and see if its legal :-) | 09:47.50 |
| | Though that doesn't excuse a crash of course | 09:48.00 |
| mattchz | yeah, np. The two are unrelated I think | 09:48.01 |
| | the PDF I used had sensitive data in so Iâm going to try and create another one | 09:48.15 |
| jogux_mac | mattchz: I sent you the details for the SO2 android beta (possibly I did that before, dunno) | 09:53.37 |
| mattchz | oh, you did, but I forgot. Ta. | 09:54.10 |
| | Iâll try that in a bit. | 09:54.18 |
| | kens> http://ghostscript.com/~matt/Index%20of%20-~matt.pdf.pdf | 09:56.34 |
| | I can open a bug as well if you like... | 09:56.41 |
| kens | No I don't need a bug, I'll just look at the file thanks | 09:56.53 |
| mattchz | FWIW, Preview and pdf.js render it fine. | 09:57.22 |
| kens | is trying to explain that there are no 'tabs' in PDF :-( | 09:58.04 |
| | The person from Unisys is going to be disappointed..... | 09:58.37 |
| pedro_mac | kens: almost as much fun as trying to explain to folks why it can be hard to search a pdf or do reflow ;) | 09:58.45 |
| | âbut I can see the textâ | 09:59.23 |
| kens | Yeah, its the same sort of problem, they make a document with spaces in it to do positioning (a stupid idea already) and want the spaces retained in teh PDF, even though the first thing they do is print it to PostScript..... | 09:59.35 |
| | Of course the 'spacing' spaces get turned into absolute positions | 10:00.00 |
| | So no more spaces | 10:00.05 |
| | "But it works if I save as PDF from Word"...... | 10:00.27 |
| | Which begs the response "well do that then...." | 10:00.40 |
| | Hmm that's a really naff way to make a PDF file..... Talk about inefficiency | 10:02.41 |
| jogux_mac | kens: urgh :-S | 10:02.47 |
| kens | : pedro_mac your PDF looks OK to me, can't see anything wrong with it. My guess would be that SOT is drawing the annotation itself rather than using hte Appearance stream, and the InkList is moderately complex, so its probably barfing on that | 10:06.14 |
| | Ghostscript renders it without even a warning which is a pretty good indicator as well | 10:07.30 |
| pedro_mac | kens: quite plausible | 10:08.44 |
| jogux_mac | it's odd given that the core of SOT has a full postscript renderer. | 10:09.23 |
| | path renderer, even. | 10:09.32 |
| | mattchz : was there hints at a release date for Android L? | 10:10.12 |
| mattchz | oh well, Iâll open a bug then | 10:10.19 |
| | thanks kens | 10:10.21 |
| | jogu> good question. âThis fallâ apparently. | 10:11.07 |
| | I get the impression that this crash is related to ART rather than just the new OS, in which case it can be reproduced in KitKat by enabling the appropriate dev options. | 10:11.53 |
| kens | jogux : but PDF isn't PostScript.... | 10:11.56 |
| jogux_mac | kens : oh, mm, true :) | 10:12.48 |
| kens | We do know that Acrobat always draws annotations rather than using the Appearance stream, so its plausible that other viewers do too, for compatibility | 10:13.01 |
| mattchz | Hmm. Thereâs no PDF component in GhostDocs :) | 10:13.04 |
| kens | Tell marcosw :-) | 10:13.21 |
| mattchz | :) | 10:13.41 |
| | So, is that three PDF renderers you guys own now? Or have I missed some? ;) | 10:13.56 |
| kens | I guess its three, though the Ghostscript one is written (mostly) in PostScript | 10:14.29 |
| | Anyway, Ghostscript itself has 4 (I think its 4, could be more) type 1 font interpreters built in..... | 10:15.12 |
| mattchz | heh | 10:15.36 |
| | bug opened: http://bugs.ghostscript.com/show_bug.cgi?id=695339 | 10:16.43 |
| Robin_Watts | kens, jogux: If only there was a bug in the SOT path rendering :) | 11:03.29 |
| | That'd be one I'd be qualified to fix. | 11:03.49 |
| jogux_mac | :) | 11:05.43 |
| Robin_Watts | jogux_mac, pedro_mac: So, the indentation problems on page 2 of Artifex2014 are down to a cockup with tabs. | 11:08.42 |
| | In office documents at least, tabs are supposed to be relative to the left margin. | 11:08.55 |
| | In SOT that isn't happening. | 11:09.06 |
| | I have a pretty trivial fix for it, but I suspect it's going to screw up other docs. | 11:09.28 |
| jogux_mac | cool. so in theory that's a relatively simple bug in layout... | 11:09.32 |
| Robin_Watts | Anyone have any feeling for whether tabs should be made relative to the left margin for everything? Or just for office docs? | 11:10.03 |
| jogux_mac | I don't think tabs are used outside of office docs? | 11:10.23 |
| | unless we support them in .txt files | 11:10.44 |
| Robin_Watts | http://intranet.picsel.com/~ats/cgi-bin/results-tgvath.pl?resultid=1068&diff=0 | 11:10.49 |
| | Those look like txt files. | 11:10.54 |
| jogux_mac | http://intranet.picsel.com/~ats/cgi-bin/fetch-log.pl?trid=228&start=2014-06-26%2022:32:11&fullpage=1&label=scriptlogepagecorecharacter-setsbn-inbengali-ppt.ath suggests it's a ppt? | 11:12.04 |
| pedro_mac | Robin: do you mean the page margin ? | 11:12.11 |
| jogux_mac | I need to fix the script viewer to work with git :-S | 11:12.14 |
| | and the next one is a .doc | 11:12.43 |
| Robin_Watts | jogux: OK :) | 11:12.50 |
| jogux_mac | you can view the 'script' by clicking on the 'log' link in the previous page, as the script commands are printed as debug as they're run | 11:13.20 |
| pedro_mac | Robin_Watts: left margin sounds sensible; even for text we should be setting a margin | 11:13.30 |
| jogux_mac | robin_watts : did you mean left margin, or left edge of container? | 11:14.01 |
| Robin_Watts | left margin, I believe. | 11:14.42 |
| | I'm going to run through some of those files checking the renderings. | 11:15.33 |
| jogux_mac | good plan | 11:16.40 |
| | (and we should keep a note of the worst one it fixes for marcosw :) ) | 11:17.05 |
| Robin_Watts | Interestingly, the microsoft powerpoint viewer screws up page 2 of bengali.~1~.ppt | 11:19.12 |
| kens | lunch.... | 11:30.04 |
| mattchz | robin_watts: http://git.ghostscript.com/?p=user/matt/mupdf.git;a=commitdiff;h=063a723b9fa50200192b01ba976ecd551461a8dc | 11:33.01 |
| | Howâs that? I was wondering if we could just get away by checking that file was a PDF and had a password, rather than adding the extra isUnencryptedPDF() call, but I wasnât sure if the encryption and passwords were orthogonal. | 11:33.47 |
| | (so I just went with what you originally suggested) | 11:33.55 |
| | (about to add the same fix to iOS too) | 11:41.29 |
| paulgardiner | My predicition was right. I've just after hours of messing about manged to build at the branch point and the result is a grotesque creature completely unsuited to any testing. So that was a big waste of time. :-( | 11:44.49 |
| Robin_Watts | mattchz: Looks good to me. paulgardiner may have an opinion too, as this is his code. | 11:45.13 |
| paulgardiner | The ulr 5 lines back? | 11:45.42 |
| | url even | 11:45.48 |
| mattchz | yep | 11:45.52 |
| jogux_mac | paul: oh, urgh :-( | 11:45.55 |
| | paul: you may still be able to bisect, assuming you can find some historical point at which it works :-S dunno if it's worth it though. | 11:46.28 |
| mattchz | btw, I noticed we are using get_globals() in various functions that we should probably change to be using get_globals_any_thread() | 11:46.38 |
| Robin_Watts | mattchz: Have at it. | 11:46.54 |
| | paulgardiner: You could start at the tip of the branch and 'probe' backwards down the branch to see if you can reach a point where it still builds, but fails ? | 11:47.52 |
| paulgardiner | jogux_mac: the trouble is I don't even know what the behaviour is at the branch point. It could be "ruler stays with menu", "ruler doesn't stay with ruler" or "menu doesn't move" | 11:47.58 |
| jogux_mac | ah. yes. :-S | 11:48.10 |
| | when I tested the SO+ build, that seemed to have 'menu doesn't move' behaviour (it's older than the SO2 build) | 11:48.29 |
| | appstore builds, that is. | 11:48.35 |
| paulgardiner | I don't know ether it became fixed on the branch, became broken on master, or was implemented on both but wrongly on master | 11:48.59 |
| | If it was implemented on both but wrongly on master then bisect isn't going to help much anyway | 11:50.16 |
| | I think my best chance is use the Windows build where it works, debug through the mechanism where by it works, and find what the break in the chain is on iOS. | 11:51.26 |
| mattchz | so, paul are you ok for me to push that change|? | 11:51.54 |
| paulgardiner | Just looking | 11:52.11 |
| mattchz | oh, sorry. | 11:52.15 |
| Robin_Watts | It's definitely left margin. Hanging indents don't change the tabs. | 11:52.21 |
| jogux_mac | paul> yeah, that sounds sensible. sitll don't understand why it would work on windows really :) | 11:52.50 |
| | paul> btw, it may be worth checking if -define=REMOVE_SIMPELIB fixes it. if it does we could just switch that on, as we may want to do so in the long term anyway... | 11:53.18 |
| paulgardiner | jogux_mac: do we think that work was reasonably complete then? | 11:54.13 |
| Robin_Watts | I suspect that REMOVE_SIMPLELIB is going to break loads of other stuff though. | 11:54.29 |
| jogux_mac | paulgardiner: one customer was getting it | 11:54.40 |
| | but we don't really know. | 11:54.45 |
| | robin_watts: I'd bet on "some" rather than "loads", but it's hard to know :-S | 11:55.03 |
| | the customer that was getting it didn't have editting, so if it breaks something it'll likely be in editting. and they were on winrt, so maybe ios/android brokeness. | 11:55.52 |
| | but probably not, it's not really platform specific. | 11:56.01 |
| paulgardiner | mattchz: so the plus sign in a circle icon still appears but does nothing? | 12:01.27 |
| mattchz | No, itâ/s hidden | 12:01.40 |
| mattchz | lunches | 12:01.44 |
| | (as it already is for XPS) | 12:02.01 |
| paulgardiner | Ah yeah got it. | 12:02.30 |
| | Looks fine | 12:02.32 |
| | Oh, for the branch build things improve a bit when I force portrait only, but now if I try to create a spreadsheet it says "Partial document contents cannot be displayed"!!!! | 12:05.50 |
| jogux_mac | paulgardiner : boggle | 13:15.51 |
| paulgardiner | Tracing how it works is looking more fruitful and I'm seeing some iOS-specific stuff, although not sure whether it's relevant yet | 13:16.56 |
| mattchz | thanks paul | 13:24.47 |
| | http://developer.android.com/guide/practices/verifying-apps-art.html | 13:36.49 |
| | looks useful | 13:36.50 |
| | I think the stack size issues may be one of the things that we trip up on. | 13:37.35 |
| jogux_mac | mattchz: hm, we're using soft threading. I wonder if alien threads might work better... | 13:46.07 |
| | android alien doesn't support that though - but possibly the linux pthread version of alien threads might just work. dunno. | 13:46.55 |
| mattchz | there is a mention on that link about using various pthread funcitons that may cause bother. | 13:47.31 |
| | odd that the app did briefly work fine. | 13:47.42 |
| | I guess we are maybe near the stack liimt | 13:47.48 |
| jogux_mac | mm, possibly. | 13:49.54 |
| pedro_mac | Robin_Watts: I notice OfficeSuite suggests an additional font pack download | 13:51.49 |
| Robin_Watts | pedro_mac: yeah, seems to be free though. | 13:52.52 |
| jogux_mac | paulgardiner / robin_watts : allegedly the worst thing REMOVE_SIMPLELIB might break is the ath scripts :-S | 13:53.21 |
| Robin_Watts | jogux_mac: boggle, if true, | 13:54.12 |
| | If that's true, then we should put that on the list of things to do. | 13:54.37 |
| paulgardiner | oh yeah doesn't ath communicate with simplelib | 13:54.54 |
| jogux_mac | robin_watts : just slight differences in where things scroll to etc :-S | 13:54.58 |
| paulgardiner | bagsy I delete simplelib | 13:55.51 |
| jogux_mac | paulgardiner: sold! | 13:56.09 |
| | no backsies! | 13:56.13 |
| paulgardiner | :-) | 13:56.23 |
| | cos I'm guessing we should be aiming at removing REMOVE_SIMPLELIB as an option, hardwiring all it implies and deleting heaps of cruddy code... in the long run at least | 13:57.19 |
| jogux_mac | nods, correct. | 14:00.21 |
| paulgardiner | For me that would feel like the end of a crusade. | 14:00.57 |
| mattchz | neil was saying we should make sure we re-enable the Wasp patent stuff. | 14:07.18 |
| jogux_mac | yeah, looks to me like the patented wasp bits are disabled, we'd need to set WASPPATENTSCAN to enable it | 14:10.25 |
| | fuzzy redraw is off too I think, looking at the --samsung-patents option in tgvbuild.py | 14:10.55 |
| paulgardiner | What affect do the patented wasp bit have? | 14:11.33 |
| Robin_Watts | speed. | 14:11.44 |
| jogux_mac | the patented version is 9ish times faster allegedly | 14:11.52 |
| paulgardiner | ah | 14:11.54 |
| Robin_Watts | that much? really? | 14:12.08 |
| jogux_mac | that was Neil's recollection. | 14:12.15 |
| Robin_Watts | in the C, maybe I guess. | 14:12.30 |
| jogux_mac | I'll quickly check back to the review see if there's any notes | 14:13.02 |
| Robin_Watts | jogux_mac: There almost certainly aren't. | 14:14.03 |
| mvrhel_laptop | marcosw: are you there? | 14:14.05 |
| Robin_Watts | cos I did the work, and I don't remember ever having had clear timings for the benefits etc. | 14:14.39 |
| mattchz | iOS equivalent fix: http://git.ghostscript.com/?p=user/matt/mupdf.git;a=commit;h=5587d318d8495d15fb9157ea8c0a925e6e590e00 | 14:15.22 |
| jogux_mac | robin_watts : yeah, I've gone through a few related reviews and the most commital statement I found was "this is less efficient". | 14:16.21 |
| Robin_Watts | It's all rubbish of course. | 14:17.06 |
| paulgardiner | mattchz: LGTM | 14:17.18 |
| mattchz | ta | 14:17.30 |
| jogux_mac | robin_Watts: http://intranet.picsel.com/~cvs/showreview.pl?id=81668#c12 has some actual figures, though doesn't state if it was C or ARM :-S | 14:22.18 |
| Robin_Watts | they talk about devices, so almost certainly ARM. | 14:23.47 |
| | oh, sorry, wasp-perf-test. | 14:23.57 |
| | That's probably on a host, so C. | 14:24.12 |
| jogux_mac | You'd hope that would be on a target device really, but it is unclear from context. | 14:24.21 |
| Robin_Watts | I'd be happy to go back to the original wasp stuff. It's much cleverer/nicer. | 14:24.55 |
| jogux_mac | that said, we do have at least one customer using the C | 14:25.03 |
| Robin_Watts | anyone on non-ARMs, yes. | 14:25.21 |
| jogux_mac | robin_watts: Artifex definitely licensed the wasp patent then? | 14:25.25 |
| | it seems odd to me that it appeared SOT where still paying or meant to be paying patent licensing fees if they'd removed use of the patents. or I've missed something. | 14:26.15 |
| | wtf. a few lines earlier up it says the build command for waspperftest was 'scripts\winbuild.py wasp-perf-test -debug -waspaltscan -define=SEARCH_ENABLE'... errr. | 14:27.09 |
| Robin_Watts | jogux_mac: Artifex has a license for all the samsung patents, yes. | 14:27.13 |
| jogux_mac | I vaguely hope they didn't test performance in a -debug build. | 14:27.40 |
| Robin_Watts | hehe | 14:27.46 |
| paulgardiner | wanders into uicontrol-docview-item.c and realises he has only himself to blame | 14:40.34 |
| mattchz | iirc, they always wanted to workaround the patents, just in case their patent license was not renewed and/or so they could run in other products | 14:44.17 |
| | i.e. non FV | 14:44.25 |
| Robin_Watts | yeah, Picsels license to the patents was always "for office viewing products" AIUI. | 14:45.28 |
| mattchz | hah. I just google to try and find the solution to an android problem (for MuPDF), and the main hit was a stackoverflow q by over very own Robin :) | 14:59.44 |
| Robin_Watts | mattchz: Probably complete rubbish then :) | 15:03.12 |
| mattchz | http://stackoverflow.com/questions/9215631/multiple-file-extension-mimetype-intent-filters-with-one-activity | 15:03.29 |
| | This one. I noticed that XPSâs arenât associated properly in the Android app. | 15:03.38 |
| Robin_Watts | ah yes. Never really got a straight answer on that, I felt. | 15:03.49 |
| mattchz | although Iâm not sure weâre doing anything wrong. | 15:03.50 |
| | from what I can see, itâs possibly the file manager thatâs doing it wrong, as even apps like âIntent Analyzerâ arenât able to receive the intent. | 15:04.33 |
| | it looks like no mimetype is being sent at all (not even */*0 | 15:04.47 |
| mvrhel_laptop | sigh. visual studio 2010 crashes quite often for me with SOT. I had hoped getting the service pack would fix this but apparently not :( | 15:05.08 |
| Robin_Watts | mattchz: ISTR that paulgardiner found a bug in EX file explorer and submitted a fix for something like that. | 15:05.10 |
| Ron___ | greetings ... may I ask for some of your time? I have sample forums up in both Discourse and phpBB | 15:05.17 |
| mattchz | robin> you mean ES? | 15:05.29 |
| kens | Hi Ron | 15:05.32 |
| jogux_mac | hi Ron, cool :-) | 15:05.40 |
| kens | Canyou point us to the forums ? | 15:05.43 |
| Robin_Watts | mvrhel_laptop: It's been relatively stable for me. (curse you for making me say that, cos now it'll die everywhere...) | 15:05.49 |
| mvrhel_laptop | :) | 15:05.59 |
| paulgardiner | Hi Ron | 15:06.02 |
| Robin_Watts | mattchz: ES, EX, whassa difference? :) | 15:06.05 |
| Ron___ | I'd like to ask that you try:http://www.artifexforum.com for phpBB and .. | 15:06.09 |
| mattchz | :) | 15:06.10 |
| | Was just interested to see his fix, thatâs all. But I didnât realise ES was opensource | 15:06.26 |
| Ron___ | artifex.discoursehosting.net for the Discourse app | 15:06.40 |
| jogux_mac | ron__ : you're going to nuke this after, right, so we can post any old junk? | 15:06.50 |
| Ron___ | http://artifex.discoursehosting.net | 15:06.58 |
| paulgardiner | Yes, ES weren't creating intents correctly. It's been fixed for some time though | 15:06.59 |
| mattchz | paul> was the issue that it wasnât setting a mimetype of */* ? | 15:07.17 |
| | which it seems to, now, if you try and open it as âother' | 15:07.36 |
| kens | Ron___ : It all seems to work OK for me. Two more places for me to watch in addition to Stack Overflow :-( | 15:07.56 |
| Ron___ | Yes jogux- both forums are out of the box, with default settings. I put up a few topics. | 15:07.56 |
| paulgardiner | It was something along those lines. Can't remember the detail now | 15:08.12 |
| jogux_mac | feels I should get a small price for passing the phpbb captcha first attempt | 15:08.22 |
| mattchz | paul> I guess thatâs what happening with the other file apps Iâm seeing a problem with. Not our fault then, at least. | 15:08.37 |
| Ron___ | At the moment, the forums both require new users to be approved by the admin. To start, I thought you should try registering as new users, and when I approve you, I'll make you admins so you can post at will, and dig into the set up areas too. | 15:09.30 |
| Robin_Watts | the phpbb one doesn't go through an email verification step. | 15:10.06 |
| jogux_mac | phpbb doesn't allow docx to be attached | 15:10.43 |
| kens | For anyone old enough to remember the 'Tennis Girl' poster: | 15:11.12 |
| | http://www.bbc.co.uk/news/uk-england-28036668 | 15:11.12 |
| pedro_mac | Ron: cool - phpBB seems to work fine - who is moderating just now? | 15:13.17 |
| Ron___ | Robin and jogux_: your posts are pending mode waiting for me to approve or disapprove. Pedro_ only I am the moderator for now. | 15:13.50 |
| | I'm going to go ahead and approve both of your posts ... | 15:14.20 |
| pedro_mac | thanks :) | 15:14.43 |
| pedro_mac | quite likes discourse | 15:17.54 |
| jogux_mac | hm. for some reason discourse doesn't trigger safari's password generator. | 15:19.11 |
| | and thunderbird thought the activation email was junk :-S | 15:20.07 |
| paulgardiner | same here | 15:21.12 |
| Robin_Watts | got through OK for me. | 15:21.43 |
| jogux_mac | can't upload a docx attachment to discourse either. | 15:21.45 |
| Robin_Watts | jogux_mac: Is that a plus or a minus? :) | 15:22.02 |
| jogux_mac | robin_watts : I feel, on a site where we're going to discuss bugs with various docx etc files, having an official way to share them would be helpful | 15:22.48 |
| Robin_Watts | yeah. | 15:22.54 |
| | so my tab changes make some definite progressions. There are some files that were wrong before and are still wrong now (probably font mismatching), but there is at least one that looks like a regression. | 15:23.44 |
| Ron___ | FYI - I am not fluent in either discourse or phpBB, but I would guess document attachments are blocked in the default set up. I'll look for the control on that feature. | 15:24.16 |
| jogux_mac | the 20 character minimum length on discourse may get annoying :) | 15:24.20 |
| Robin_Watts | jogux_mac: For topic titles? seems great to me. | 15:24.51 |
| jogux_mac | robin_watts : and for any comment | 15:24.59 |
| Robin_Watts | How many "bug report" or "problem" threads do we need? And if your reply isn't longer than 20 chars, it's probably "me too!" etc. | 15:25.33 |
| mvrhel_laptop | Robin_Watts: quick question for you. how do I do a clean and rebuild in the solution. I guess I can go in and remove the files but I was wondering if we can hook up the clean and rebuild so that they work properly | 15:26.01 |
| jogux_mac | robin: "That's unsupported." :) | 15:26.18 |
| mvrhel_laptop | hi rayjj | 15:26.30 |
| Robin_Watts | mvrhel_laptop: Most of the targets have rebuild set up. | 15:26.52 |
| mvrhel_laptop | hmm I tried it on the solution | 15:27.09 |
| Robin_Watts | The testshell target doesn't. | 15:27.21 |
| mvrhel_laptop | oh there it goes | 15:27.25 |
| | I must have been off | 15:27.29 |
| pedro_mac | jogux: its a forum: @Thatâs unsupported !!!!??!??!!! :):);) | 15:27.30 |
| mvrhel_laptop | sorry Robin_Watts | 15:27.31 |
| jogux_mac | hehe | 15:27.42 |
| Robin_Watts | mvrhel_laptop: No, it was partly me being lazy. | 15:27.51 |
| mvrhel_laptop | henrys: are you there? | 15:28.10 |
| Robin_Watts | The way to rebuild is to delete the appropriate section of genroot. | 15:28.18 |
| | Let me make a commit that does that. | 15:28.27 |
| jogux_mac | do all the reply buttons on the page in discourse do the same thing, or are there non-obvious differences? | 15:28.33 |
| Robin_Watts | mvrhel_laptop: I might err on the side of caution and delete more than is absolutely necessary though, so cleaning one target may affect another. | 15:29.20 |
| mvrhel_laptop | I understand | 15:30.48 |
| jogux_mac | ron___ : so, afaics, either work. I have a slight preference for phpbb, but only due to familiarity. | 15:31.45 |
| mattchz | kens> that PDF we were trying this morning with the annotations problems seems to have the same problem (no annotations displayed) when viewed in iOSâs native viewer. | 15:44.30 |
| Robin_Watts | I like discourse, I think. | 15:44.31 |
| kens | mattchz : interesting! | 15:44.44 |
| mattchz | kens> I think thato is probably what this bug is about: http://bugs.ghostscript.com/show_bug.cgi?id=694789 | 15:44.49 |
| kens | Hmmcould be, but without a smaple PDF file its rather hard to tell :-) | 15:45.12 |
| mattchz | yeah. | 15:45.24 |
| kens | It does seem kind of likely | 15:45.37 |
| mattchz | I just tried with another PDF annotated in Android, sent to iOS and the annotations are viewable in MuPDF (iOS), but not the native iOS viewer. | 15:46.01 |
| | oddly itâs fine in OS X Preview, which you might assume is the same code. | 15:46.18 |
| kens | As I said, I can't immediately see anything wrong with the PDF. GS uses the appearance stream and that works OK, Acrobat is known to generally discard Appearance streams and so will most likely be using the InkList, and the Acrobat display is fine too | 15:46.25 |
| mattchz | are both streams added by us? | 15:47.17 |
| kens | I would assume so yes | 15:47.48 |
| Ron___ | sorry .... i was digging through the admin control panels. phpBB is by default set to allow for attachments of any type, discourse only jpg, jpeg, png, gif. I can add file types to discourse, but which ones? pdf, doc, docx, xls, xlsx, more? | 15:48.20 |
| kens | Appearance streams are optional, and if they aren't present viewers can use the 'required' information to render the annotation, or just do nothing, whatever they like. | 15:48.36 |
| paulgardiner | Found it: Review 84883 breaks the ruler animation | 15:48.40 |
| kens | The appearance stream is *supposed* to draw the same as if you used the information to render the annotation, but some applications aren't too good at that..... | 15:49.22 |
| mattchz | I should really learn about pDF sometime :) | 15:49.40 |
| kens | One thing you could try is to break the annotation stream, open teh file in Acrobat and save it, which will write a new appearance stream. Then you could open the new PDF and see if it still breaks. I'm guessing it will. | 15:50.15 |
| mattchz | presumably though, given we add both, it shouldnât matter which the app uses? | 15:50.18 |
| kens | Well, the results are 'supposed' to be the same. | 15:50.32 |
| mattchz | ah | 15:50.56 |
| kens | It would have been easier if Adobe had mandated an appearance stream, then viewers wouldn't have to manufacture the dratted things | 15:50.57 |
| mattchz | :) | 15:51.04 |
| Robin_Watts | Ron___: ppt, pptx, txt, csv ? | 15:51.21 |
| kens | Its just that the InkList on that annotation is middling cimplicated (its a relatviely complex annotation) and its possible a viewer might throw a wobbly on it. | 15:51.31 |
| jogux_mac | paulgardiner : presumably it only wasn't broken before that because you couldn't hide the top bar on iOS before that? :-) | 15:51.48 |
| kens | If I was investigating it I'd make 2 copies of the PDF> On one I'd zap the apperance stream, and on the other I'd zap the InkList (empty array) and then opne both on the devcie and see what it does. | 15:52.17 |
| Ron___ | robin: there is an option to allow all attachments, but we probably don't want that. I'll add the ones you thought of.... | 15:52.19 |
| mattchz | kens> I guess it looks like this is a problem at the core left of mupdf, so I should probably leave this to someone more qualified⦠| 15:52.19 |
| paulgardiner | No actually. This seems to include the code that breaks it | 15:52.24 |
| kens | mattchz : It could just be a common bug in the viewer software..... | 15:52.44 |
| mattchz | yeah | 15:52.49 |
| jogux_mac | paulgardiner: huh. odd. but good :) | 15:52.59 |
| kens | The two applications I have some trust in both render it OK, using different approaches, so I'd guess both the Apperance and the InkList are valid | 15:53.19 |
| Robin_Watts | mvrhel_laptop, jogux, pedro_mac, paulgardiner: Fix for the VS solution (adding rebuild and clean targets) on robin/master | 15:53.21 |
| mattchz | kens> ok, cool. Iâll just leave it for someone with pdf expertise to look at then, ta :) | 15:53.57 |
| paulgardiner | jogux_mac: I can easily fix it now, but I'm not sure of the potential side effects | 15:54.00 |
| kens | ok no problem | 15:54.04 |
| mattchz | thanks for looking into it | 15:54.14 |
| kens | Any time I could use a break from this problem :-( | 15:54.29 |
| mattchz | hehe | 15:54.47 |
| | kens> thereâs also this one bug, actually, which may be relevant: http://bugs.ghostscript.com/show_bug.cgi?id=695130 | 15:54.54 |
| kens | Hmm, that's more interesting | 15:55.14 |
| paulgardiner | jogux_mac: could you take a look at: http://intranet.picsel.com/~cvs/showdiffs.cgi?action=diff&colour=1&id=84883&patch=patch.zip.40&style=graphical | 15:55.26 |
| kens | grabs the files | 15:55.32 |
| mvrhel_laptop | thanks Robin_Watts | 15:55.50 |
| paulgardiner | At the line with "/* Update the focusable rect to... | 15:56.08 |
| jogux_mac | paulgardiner: only 40 attempts to get it working spanning over 3 months. I'm sure your easy fix'll be fine ;-) | 15:56.36 |
| paulgardiner | jogux_mac: :-) | 15:56.48 |
| pedro_mac | Robin_Watts: clean targets LGTM | 15:57.04 |
| kens | mattchz : THere's defintiely something wrong with that highlight.... (attachment 2). | 15:57.10 |
| jogux_mac | paulgardiner: okay, I'm looking there. | 15:57.16 |
| jogux_mac | vaguely wonders what's different on the release branch. | 15:57.49 |
| kens | Actually, Ghostscript (which renders highlights from the quad points rather than using the appearance, a rare exception) shows a problem on both annotations | 15:58.18 |
| paulgardiner | jogux_mac: the comment seems to say now we have this new behaviour we need to do X to stop Y, but the code restricts when X is done. X used to be done in more cases. If I change back to doing X independent of app->hasMenuButton all seems well | 15:58.19 |
| jogux_mac | paulgardiner: hmmm. I presume if we change that, you can't top on the area at the top of the screen to retrieve the bar? | 15:58.57 |
| | and/or, doing so gets passed down and causes other things to happen? | 15:59.10 |
| | s/top/tap/ | 15:59.19 |
| paulgardiner | Seems to work here | 15:59.32 |
| jogux_mac | ah. hm. | 15:59.56 |
| | but you could scroll the document to a place where it would be insivible when you unhid the toolbar? | 16:00.12 |
| paulgardiner | I don't think you could. It's forced central when below a certain size | 16:01.02 |
| | The centre of the screen always contains document | 16:01.26 |
| | is within the document | 16:01.37 |
| jogux_mac | I'm not sure I'm having any success parsing that comment | 16:02.24 |
| paulgardiner | I seem to be able to click on the document or within where the menu would be to get the menu back | 16:02.33 |
| jogux_mac | it's definitely saying something about scrolling | 16:02.52 |
| paulgardiner | You cannot pan and zoom such that the centre point of the screen is not within the are of the document | 16:03.03 |
| | s/are/area/ | 16:03.26 |
| jogux_mac | can't figure it out. | 16:05.26 |
| | can you check with docs of different aspect ratios? eg. one pptx and one docx. | 16:05.57 |
| Ron___ | I have to log off. Bye for now. | 16:06.27 |
| kens | mattchz : sadly that particular PDF file has a lto more than a single annotation, which is making it hard to figure out..... | 16:06.58 |
| mattchz | :( | 16:07.16 |
| kens | Its got forms, trasnparenyc, pop up annotations at least 3 highlught annotations.... | 16:07.36 |
| mattchz | kens> nice@ | 16:07.47 |
| | s/@/!/ | 16:07.52 |
| kens | Most of which I think are co-incident..... | 16:07.59 |
| paulgardiner | jogux_mac: seems fine all aspects all orientations | 16:08.25 |
| kens | The OP's deduction seems incorrect though, the QuadPoints seem to be in the correct order | 16:08.27 |
| mattchz | OS X Preview displays no annotations actually. | 16:09.37 |
| Robin_Watts | the ordering of the quadpoints in a highlight annotation is a matter of some controversy I think. | 16:09.50 |
| kens | mattchz : there are 6 annotations, 2xpopup and 4xhighlight, all the highlight annnotations are in the same location..... | 16:10.23 |
| | I have no idea how to even start making sucha PDF file :-) | 16:10.53 |
| paulgardiner | The code and comment definitely do not match. To achieve what the comment says, one would need to alter the if (animateOn) case | 16:11.04 |
| mattchz | Is this the âHello Worldâ file? | 16:11.05 |
| kens | Yes | 16:11.13 |
| | OK so if I remove everything but one annotation, none appear :-) | 16:11.39 |
| paulgardiner | but doing that would cause the rulers to stay hight and be covered by the menu | 16:11.45 |
| mattchz | kens> Start by using MuPDF I guess :) | 16:11.46 |
| paulgardiner | I think I'm going with my fix. | 16:11.54 |
| kens | mattchz : using a text editor | 16:11.55 |
| jogux_mac | paulgardiner: so long and short - the only thing that we can identify that app->fvTop does is move the ruler? | 16:12.15 |
| kens | OK so its only the final popup annotation that does anything useful | 16:12.48 |
| | object 43 0 in the file | 16:13.03 |
| mattchz | wasnât this annotation added in MuPDF? | 16:13.13 |
| kens | has no clue | 16:13.27 |
| Robin_Watts | MyPDF doesn't add popup annotations. | 16:13.35 |
| kens | Actually the last one is a highlight, not a popup | 16:13.52 |
| paulgardiner | jogux_mac: not sure that's all it does, but it definitely does move the rulers so we need it changed and I cannot identify any bad effects from what else it does. Plus the reason for adding the condition I'd remove is flawed | 16:13.55 |
| kens | Potentially the ones that display nothing were added previously and then the highlight added by MuPDF | 16:14.27 |
| mattchz | âI'm using the latest muPDF library available on Git (last clone I made: 02 April 2014). The target platform is Android. | 16:14.32 |
| | When I try to add highlight annotations they are correctly rendered if I use the same muPDF library both for manipulating and rendering the PDF file." | 16:14.32 |
| paulgardiner | Try that again: the condition I'd remove was added for a flawed reason | 16:14.45 |
| Robin_Watts | s/for/for what looks like/ | 16:15.02 |
| kens | mattchz : it could be that the existence of the popup in the same place is confusing somehow | 16:15.17 |
| | Oh interesting, if I remove the appearance stream acrobat doesn't draw the annotation O.O | 16:16.15 |
| mattchz | I didnât even realise a popup annotation was a thing before :) | 16:16.17 |
| jogux_mac | paulgardiner: I can play here if you want to stick a commit up | 16:16.26 |
| | it's puzzling. | 16:16.30 |
| | paulg: such a change should only affect iOS basically? | 16:17.13 |
| paulgardiner | yes I believe so | 16:17.34 |
| kens | mattchz : it does look like the appearance doesn't match the QuadPoints in the example, and not only that, but the QuadPoints look to be incorrect, so manufactured appearances are wrong | 16:17.48 |
| mattchz | confusing :/ | 16:18.18 |
| jogux_mac | so, rons gone: but phpbb warns you if you try to close the window whilst writing a response, discourse doesn't. That sways me to phpbb quite strongly. | 16:18.26 |
| kens | mattchz : yes, really confusing, because normally Acrobat ignores appearances, this time it doesn't possibly because it decides the QuadPoints are bogus | 16:19.07 |
| | You can tell teh difference in this case because the appearance doesn't draw the highlight with rounded end caps | 16:19.40 |
| | also I just changed its coour to green by editing the appearance stream and the acrobat display changed | 16:20.04 |
| | Which is kind of a clue | 16:20.10 |
| mattchz | ah | 16:20.20 |
| kens | I have to admit I can't actually see a problem wiht the data though | 16:21.13 |
| mattchz | kens: oh btw, regarding http://bugs.ghostscript.com/show_bug.cgi?id=694409 | 16:22.09 |
| | can you rememeber how you got the UTF-16 filename onto the android device? | 16:22.20 |
| | It seems to work fine from Mac OS (I guess Mac OS is using UTF-8) | 16:22.32 |
| kens | It wouldn't have been me, I don't have an android device | 16:22.39 |
| mattchz | ah | 16:22.42 |
| kens | OK I have to dash off to take Melanie riding, bye all have a good weekend. I'll stare at the quad poitns some more tomorrow | 16:23.36 |
| paulgardiner | jogux_mac: there, pushed. I'll add some more expanation to the commit message another time | 16:23.56 |
| jogux_mac | mattchz: IIRC, if you use a korean version of windows, it might well right a filename to an sd card using an 8-bit korean specific encoding. (koir? or something like that) | 16:25.16 |
| mattchz | yeah, I guess that could be it. | 16:25.55 |
| henrys | jogux_mac: both of these are hosted right? I didnât get a chance to talk to Ron | 16:26.03 |
| jogux_mac | discourse was | 16:26.15 |
| mattchz | It looks like the app is crashing inside the JNI implementation of the File class, so Iâm not sure thereâs much we can do if so ⦠| 16:26.26 |
| Robin_Watts | henrys: discourse is hosted. phpBB is setup on hosting we control. | 16:26.46 |
| henrys | jogux_mac: phpbb does have hosting as well. Are you sure he didnât use it? | 16:26.57 |
| jogux_mac | I wasn't sure about phpbb, but robin seems to be :) | 16:27.11 |
| | I think hosted phpbb is likely to be preferable (assuming from a reputable provider), surely cheaper to pay someone else to keep it secure. but not my call :) | 16:28.46 |
| | matt: koir would certainly fit with the error | 16:29.06 |
| | mattchz : probably you can write an filename with invalid utf8 from a mac? or at least from linux. | 16:29.52 |
| mattchz | possibly. | 16:30.16 |
| | seems worth a try. | 16:30.24 |
| jogux_mac | basically just stick a latin1 £ sign in and you'll be sorted :) | 16:30.33 |
| Robin_Watts | I am not 100% sure about phpBB being in our own hosting. | 16:30.35 |
| mattchz | :-) | 16:30.52 |
| henrys | jogux_mac, Robin_Watts : Iâd expect godaddy or an amazon instance and it isnât | 16:30.54 |
| mattchz | Are you meaning Ko18-R? | 16:30.59 |
| henrys | Iâll contact him | 16:31.01 |
| jogux_mac | matt : that sounds right. | 16:31.08 |
| Robin_Watts | henrys: amazon instance would be a very poor choice I think. | 16:31.13 |
| jogux_mac | matt : I suspect any invalid utf8 would trigger the bug, anyway :) | 16:31.16 |
| henrys | I really think hosted is a requirement - I did tell him that. Does anyone feel we need to have full control of the BB | 16:31.44 |
| | ? | 16:31.46 |
| Robin_Watts | sorry, it's arxive, and there is a phpbb.arxive.com hosting, so it looks like it is hosted. | 16:32.16 |
| henrys | Robin_Watts: I was considering *possible* choices | 16:32.17 |
| | great | 16:32.39 |
| mattchz | Sounds like FAT FSâs can store names in arbitrary code pages :-( | 16:32.45 |
| Robin_Watts | phpbb.arvixe.com | 16:32.46 |
| jogux_mac | mattchz : that matches my recollection | 16:32.56 |
| Robin_Watts | I vote we abandon phpbb purely because I can't type the hosting name :) | 16:33.05 |
| jogux_mac | mattchz : I recall similar issues with shiftjis | 16:33.06 |
| | robin_watts: :P | 16:33.21 |
| | I would prefer it to be forum.artifex.com. I hate something-companyname.com stuff, you have to go poking to figure out if it's pishing or not. | 16:33.54 |
| Robin_Watts | jogux_mac: likewise. but I suspect that can easily be changed. | 16:34.25 |
| mvrhel_laptop | henrys: question for you | 16:34.31 |
| jogux_mac | yeah, I would presume so. | 16:34.32 |
| henrys | mvrhel_laptop: okay | 16:35.05 |
| mvrhel_laptop | I did not want the fix for customer 532 to slip through the cracks | 16:35.08 |
| | I was not sure who was handling support | 16:35.32 |
| jogux_mac | mattchz : I'm not sure you have to figure out how to correctly convert it into utf8, just make it not crash and displays '?'s :) | 16:35.34 |
| mvrhel_laptop | I can send the patch to them but I am not sure what version they need to apply it to | 16:35.59 |
| henrys | mvrhel_laptop: marcosw is back AFAIK | 16:36.00 |
| jogux_mac | [I'm not sure it's *possible* to correctly convert it into utf8] | 16:36.05 |
| mvrhel_laptop | ok I pinged him a couple times on here and sent him an email . | 16:36.19 |
| mattchz | indeed, but my point was that Iâm not sure it will be possible to do so, as it appears the be the Java directory enumeration functionâs JNI implementation that is crashing ;) | 16:36.35 |
| jogux_mac | mattchz: oh, oops, misread the back trace. | 16:37.39 |
| | that's someone annoying, but surely means a file explorer will also crash? | 16:38.08 |
| mattchz | Iâd expect so. | 16:38.18 |
| henrys | Iâm inclined to put my vote in for discourse but donât feel strongly about it. Anything related to php disturbs me. | 16:38.25 |
| mattchz | I guess weâll see if we can reproduce it. | 16:38.27 |
| | Well, there would be a workaround, and that would be to implement our own native File system enumeration functions. | 16:38.43 |
| Robin_Watts | mattchz: No. That's definitely not our job. | 16:39.14 |
| mattchz | indeed, not saying it would be a good idea. | 16:39.36 |
| jogux_mac | henrys : my objections to discourse are likely to be solvable / solved in due course. (the password field not triggering safari's password generator, it not warning if you try and close tab whilst composing a message). | 16:40.15 |
| | oh, and I'm not sure we can configure it to allow docx etc attachments either. | 16:41.00 |
| mattchz | is this a forum for SOT? | 16:41.52 |
| Robin_Watts | mattchz: It is. | 16:42.02 |
| | jogux_mac: Eh? Ron already made it allow doc, docx, etc. | 16:42.24 |
| jogux_mac | I suspect discourse will be better from a security point of view. | 16:42.29 |
| mattchz | I hope you intend to give customers the level of service theyâve come to expect ;-) | 16:42.31 |
| jogux_mac | robin_watts : oh, he did? I missed that. | 16:42.45 |
| Robin_Watts | He was asking what types to allow and I gave him a few. | 16:43.00 |
| | I closed a tab while composing a message, and when I went back later, it still magically had my part composed message in it. | 16:43.38 |
| henrys | phpbb soundâs like something out of the 00âs | 16:43.44 |
| Robin_Watts | It ain't that modern :) | 16:43.53 |
| mattchz | phpbb historically has had a lot of security flaws. | 16:44.23 |
| jogux_mac | robin_watts : ah, it does say it allows docx now. but I still can't upload one as I'm a new user. d'oh :) | 16:44.52 |
| | robin_watts : oo. okay. that's neat. maybe that objection of mine is not valid then. | 16:45.20 |
| henrys | mattchz: I have no faith in something written in a language that canât be deciphered a week after itâs written by its author | 16:46.38 |
| jogux_mac | henrys : you're think of perl, it's a month for php ;-) | 16:47.04 |
| henrys | perhaps php is slightly less disturbing but not much. | 16:48.53 |
| jogux_mac | :) | 16:48.57 |
| jogux_mac | is happy to try discourse for a bit. | 16:49.13 |
| Robin_Watts | henrys: says the man working in postscript... | 16:49.30 |
| jogux_mac | apparently discourse will be coming out of beta next month | 16:51.06 |
| henrys | discourse has a nice ring to it. Hopefully it will soothe the participants before they post ;-) | 16:54.51 |
| jogux_mac | is that a decision then? | 16:55.39 |
| henrys | like for paulgardiner to weigh in. | 16:56.03 |
| | I donât think we should have ghostscript or mupdf here - stackoverflow works fine, I donât think we need another community | 16:57.27 |
| paulgardiner | I think I'm fairly happy either way. | 16:57.36 |
| jogux_mac | henrys : maybe we should add smartoffice in the hostname? but that sounds sensible to me anyway. | 16:58.31 |
| paulgardiner | I'd like to pick what is most likely to draw others of us to go on and answer questions. :-) | 16:58.44 |
| henrys | paulgardiner: I was waiting for your opinion because I figured you wanted to answer all the questions ;-) | 16:59.32 |
| jogux_mac | vaguely ponders if the forum + bugzilla could share usernames/passwords. | 17:00.11 |
| henrys | jogux_mac: agreed smartoffice in the hostname | 17:00.16 |
| jogux_mac | or even by single-sign-on. | 17:00.21 |
| paulgardiner | You pick one and then I'll leave it to a later date to mention it wasn't the one I'd have chosen | 17:00.25 |
| henrys | Iâll email ron if everyone is good | 17:02.32 |
| Robin_Watts | So, what hostname are we thinking? | 17:02.55 |
| | smartoffice.com is taken. | 17:03.12 |
| jogux_mac | smartoffice-forum.artifex.com is my current not very snappy best suggestion. | 17:03.14 |
| | or possibly discourse-smartoffice.artifex.com | 17:03.55 |
| henrys | oh I assume smartoffic.discoursehosting.net | 17:03.56 |
| | Iâll even spell it right | 17:04.17 |
| Robin_Watts | henrys: We can put it under any hostname we want. Why would we want discoursehosting.net in there? | 17:04.21 |
| jogux_mac | I think having artifex.com in there makes it clear it's "official". | 17:04.38 |
| henrys | Robin_Watts: oh I thought if you used their hosting we were stuck with that | 17:05.02 |
| mattchz | how about forum.artifex.com, then you can use it for other products too? | 17:05.22 |
| | or forum.smartoffice.com | 17:05.37 |
| Robin_Watts | smartoffice.com is not available. | 17:05.45 |
| | henrys: I'd be surprised. | 17:05.49 |
| mattchz | oh, you donât have that domain. | 17:05.51 |
| henrys | mattchz: I donât want to draw folks out of the already popular stackoverflow for gs and mupdf | 17:05.56 |
| jogux_mac | we have picselsmartoffice.com (I think). | 17:05.58 |
| | although I can't see how that helps :) | 17:06.17 |
| mattchz | forum.picsel.com ? | 17:06.35 |
| Robin_Watts | I think we'd ideally like to drop the picsel name, right? | 17:07.18 |
| mattchz | why on earth would you want to do that, with itâs fantastic reputation? | 17:07.34 |
| jogux_mac | mattchz : sarcasm doesn't always come out on IRC ;-) | 17:07.48 |
| Robin_Watts | What am I thinking! Of course... | 17:07.49 |
| mattchz | inserts the appropriate <sarcasm> tags | 17:08.07 |
| henrys | domain name rules are here https://www.discoursehosting.com/ | 17:08.29 |
| jogux_mac | henrys : so something.artifex.com is okay by my reading. | 17:09.03 |
| henrys | jogux_mac: right | 17:09.23 |
| mattchz | iâd personally go for smartoffice.artifex.com | 17:09.33 |
| henrys | I kind of like using discoursehosting.net - the names you are suggesting sound like the site is where you get the product | 17:10.21 |
| Robin_Watts | I'd be tempted by smartoffice-forum.artifex.com, and then have a redirect to that from smartoffice.artifex.com | 17:10.44 |
| | That way, if we ever move the picsel.com site to smartoffice.artifex.com (as I believe is planned), we can use smartoffice.artifex.com/forum to redirect. | 17:11.20 |
| jogux_mac | smartoffice-forum.artifex.com works for me. | 17:11.55 |
| Robin_Watts | henrys: As a user, why would I be remotely interested in the fact that the support forum is a hosted one, or that it uses discourse? | 17:11.57 |
| jogux_mac | plus we'd be locked into discoursehosting.net then. | 17:12.16 |
| Robin_Watts | Having the word 'forum' in the title achieves what you want, I think. | 17:12.21 |
| henrys | Robin_Watts: fine by me | 17:12.38 |
| Robin_Watts | Or.... we could register smartoffice-forum.com | 17:12.53 |
| jogux_mac | robin_watts : I vote against that; I hate it when you have to search to try and figure out if a site is 'official' or not. | 17:13.34 |
| Robin_Watts | jogux_mac: Yeah. | 17:13.57 |
| henrys | also we can get this done now using artifex, we have to get paulgardiner started answering these questions ;-) | 17:14.36 |
| jogux_mac | :) | 17:15.07 |
| jogux_mac | awaits the first "gah I hate all these idiots on SOF" | 17:15.32 |
| Robin_Watts | Gah. So currentLeftMargin, isn't. | 17:16.40 |
| pedro_mac | isnt that a running container-relative offset? | 17:17.15 |
| jogux_mac | and it'll be the CSS left margin, which I can never remember the meaning off. | 17:17.37 |
| Robin_Watts | It seems right in most cases, but when we have a bullet point, then some text, it's to the right of the bullet point, not to the left of it. | 17:18.26 |
| pedro_mac | departs for the drive home. Have a good weekend all | 17:19.47 |
| henrys | The only thing I can think of is that discoursehosting might get us a bit more google juice, does that change anyoneâs mind? Iâll take silence as know and request the agreed upon smartoffice-forum.artifex.com. | 17:22.57 |
| | s/know/no | 17:23.07 |
| jogux_mac | henrys : I'm not sure it necessarily would, unless I missed something | 17:24.54 |
| | paulgardiner: having tested, I'm can't find anything that's worse with your patch cf the current appstore release. | 17:25.35 |
| | there's a lot of stuff there's weird / random, but afaict in the same ways in both builds :) | 17:25.57 |
| paulgardiner | jogux_mac: yes, awful but no more awful. Result! | 17:27.10 |
| jogux_mac | :) | 17:29.09 |
| | it seems 50/50 whether clicking where the toolbar was causes a cell to be focussed or not, but again same in both builds. | 17:29.29 |
| henrys | jogux_mac: I thought google ranking was effected by a reference to your domain as well as an ingoing link to a page, and discoursehosting.net would have more incoming links than artifex. | 17:33.10 |
| jogux_mac | I'm not heard that. | 17:34.15 |
| henrys | that said I donât know how eager I am for folks to find us. | 17:34.34 |
| | okay smartoffice-forum it is. | 17:35.30 |
| Robin_Watts | henrys: pagerank does indeed work on links to a domain. | 17:37.52 |
| | (or did) | 17:37.56 |
| | but AIUI, fred.discourse.net is very differently ranked than george.discourse.net | 17:38.20 |
| | i.e. google juice is not shared between subdomains. | 17:39.19 |
| jogux_mac | paulgardiner: trivial iOS fix on my master, that's been bugging me for a while :) | 17:40.09 |
| | paulgardiner: I'm happy with the ruler fix, if you update commit message like you said, add a link to that review, etc :) | 17:41.17 |
| henrys | Robin_Watts: I havenât read about it recently and itâs always changing. But take wikipedia for example, they must special case that if it isnât getting domain juice. | 17:44.33 |
| Robin_Watts | consider blah.demon.co.uk | 17:45.33 |
| | demon is a big ISP in the UK. everyone that subscribes to it, gets their own website at <username>.demon.co.uk | 17:46.09 |
| | They don't all share google juice. | 17:46.15 |
| | I'd imagine that wikipedia etc is the exception rather than the rule. | 17:46.29 |
| henrys | Robin_Watts: makes sense | 17:46.49 |
| | I do get a lot of yahoo returns from google as well, but I guess their QA forum has established themselves | 17:48.40 |
| Robin_Watts | Regardless of SEO, I think the clarity of having our own domain wins. | 17:53.28 |
| paulgardiner | jogux_mac: LGTM | 17:57.10 |
| jogux_mac | paulgardiner: ta, pushed. | 17:57.35 |
| mattchz | right, hometime. good weekend all! | 18:04.10 |
| jogux_mac | me too. have a good one all. | 18:06.47 |
| mvrhel_laptop | Robin_Watts: are you still around? | 18:17.49 |
| Robin_Watts | I am. | 18:17.54 |
| mvrhel_laptop | I see the rest of the SOT crew is goen | 18:17.57 |
| | do you have a few moments to guide me through some ignorance | 18:18.12 |
| Robin_Watts | I can stumble hand in hand with you through our collective ignorance, sure :) | 18:19.04 |
| mvrhel_laptop | ok. this is with xml parsing so it will probably be easy for you | 18:19.19 |
| | so I have a chart xml content | 18:19.28 |
| Robin_Watts | My exposure to the xml parsing code is almost 2 weeks old now. | 18:19.41 |
| | ok. | 18:19.48 |
| mvrhel_laptop | and the tags if i understand correctly are defined in drawingmlchart-tags.h | 18:20.01 |
| | in drawingml-char.c | 18:20.18 |
| | around line 1399 | 18:20.25 |
| | we are looking to see if there are some optional color properties | 18:20.50 |
| | there is a call sp = NodeMngr_findChildNode(node, OOXML_TAG(DrawingMLChart, spPr)); | 18:21.38 |
| Robin_Watts | I have that here, yes. | 18:22.02 |
| mvrhel_laptop | ok. in this case, node happens to be c:ser tag type | 18:23.07 |
| | let me past the xml some place | 18:23.15 |
| | let me put it in pastebin | 18:23.52 |
| | Robin_Watts: http://pastebin.com/uPxZUCaz | 18:24.53 |
| | so if you go down to <c:ser> | 18:25.15 |
| Robin_Watts | line 23? | 18:25.28 |
| mvrhel_laptop | yes | 18:25.33 |
| | so the node claims to have 2 children | 18:25.41 |
| | this is confusing to me to begin with and we talked about this already | 18:25.55 |
| | yesterday | 18:25.58 |
| | Robin_Watts: in any event we want to see if there is an spPr entry | 18:26.32 |
| | but this is inside dPt | 18:26.42 |
| Robin_Watts | Right. I am slightly surprised by that too, but I can only imagine that it's dropping ones it doesn't understand. | 18:26.52 |
| mvrhel_laptop | not a direct child of <c:ser> | 18:26.55 |
| Robin_Watts | Right. | 18:27.00 |
| mvrhel_laptop | so I thought. ok let me add dPt as a tag | 18:27.11 |
| | and get that first | 18:27.16 |
| | but if I add dPt in drawingmlchart-tags.h the code crashes early on | 18:28.07 |
| | and I suspect this is due to my not understanding how this all works | 18:28.23 |
| Robin_Watts | ah, right, well, that I might understand. | 18:28.38 |
| | Let me look. | 18:28.43 |
| | for each tag we add, it generally goes into a table. | 18:30.49 |
| mvrhel_laptop | ok | 18:31.06 |
| Robin_Watts | so you've added it in the correct alphabetical order, right? | 18:31.34 |
| mvrhel_laptop | yes I added it at... | 18:31.41 |
| | line 29 | 18:31.49 |
| | OOXML_DEFINE_TAG(doughnutChart) | 18:32.06 |
| | OOXML_DEFINE_TAG(dPt) | 18:32.07 |
| | OOXML_DEFINE_TAG(f) | 18:32.09 |
| Robin_Watts | P < o, I think in ascii terms. | 18:32.31 |
| mvrhel_laptop | oh | 18:32.52 |
| Robin_Watts | so dPt should be before delete ? | 18:33.01 |
| | maybe? I assume that's how this works. | 18:33.09 |
| mvrhel_laptop | let me try | 18:33.18 |
| Robin_Watts | That won't solve the crash. | 18:33.27 |
| | This is a chaemleonic header file. | 18:33.37 |
| mvrhel_laptop | ok :) | 18:33.38 |
| Robin_Watts | It gets included several times with different definitions of OOXML_DEFINE_TAG | 18:33.52 |
| mvrhel_laptop | I already had to try to understand the use of ## in the defines | 18:33.56 |
| | that is no fun to follow if you are not in the know | 18:34.14 |
| Robin_Watts | ah, right, in ooxml-tags.h it goes into an enum. | 18:34.34 |
| | and there is a __Last thing there, so we should know the number of tags. | 18:34.58 |
| mvrhel_laptop | yes. I looked over that file and was confused quite a bit | 18:34.59 |
| | ok | 18:35.03 |
| Robin_Watts | and that should automatically be updated. | 18:35.05 |
| mvrhel_laptop | I was wondering about ath | 18:35.12 |
| Robin_Watts | what did you find confusing there? | 18:35.12 |
| | ath? | 18:35.35 |
| mvrhel_laptop | just understanding how it created the id tag from the namespace and the tag name with the ## symbol. I think I understand now | 18:35.52 |
| | ath = that | 18:35.57 |
| Robin_Watts | Ah, right. | 18:36.06 |
| mvrhel_laptop | sorry for the sloppy typing | 18:36.08 |
| Robin_Watts | sorry, ath is something else in here :) | 18:36.16 |
| | So the reason the tags are defined like this, is to make life easy for us. | 18:37.39 |
| mvrhel_laptop | I would assume. but I am missing something... | 18:38.00 |
| Robin_Watts | We define the tag just once, and it can be used both to define the enums, and to initialise the tables. | 18:38.19 |
| | ooxml-context.c reincludes the headers again, with a different definition of OOXML_DEFINE_TAG. | 18:38.54 |
| mvrhel_laptop | darn VS crashed again :( | 18:38.55 |
| Robin_Watts | I was hoping to see somewhere where we weren't accounting for the extra tag, and hence we'd be overwriting the end of a table, but I can't see anything obvious like that :( | 18:39.47 |
| | Are you doing a debug build ? | 18:41.02 |
| mvrhel_laptop | Robin_Watts: good news | 18:41.07 |
| Robin_Watts | </futurama> | 18:41.20 |
| mvrhel_laptop | so moving the thing to the proper alphabetical order fixed it | 18:41.35 |
| Robin_Watts | In a debug build, I would have hoped that you would have seen an assert in ooxml-context.c | 18:41.52 |
| | for the non alphabetical thing. | 18:42.02 |
| | but that's great that it's working. | 18:42.20 |
| mvrhel_laptop | Robin_Watts: its possible it was just an assert | 18:42.27 |
| | that I saw | 18:42.29 |
| | I don't remember now | 18:42.32 |
| Robin_Watts | ah, ok. | 18:42.33 |
| mvrhel_laptop | I am running in debug | 18:42.42 |
| | ok. I think I might actually be able to make some progress on this now | 18:43.17 |
| | thanks for teaching me my ABCs! | 18:43.25 |
| Robin_Watts | hehe :) | 18:43.31 |
| | no worries. | 18:43.34 |
| mvrhel_laptop | although my son wants me to help him with some 3D projective views on his scratch project | 18:44.03 |
| | he had a project that was featured on the scratch web site last week | 18:44.16 |
| Robin_Watts | mvrhel_laptop: Have you been watching castAR? | 18:44.40 |
| mvrhel_laptop | he had 8000+ views on his project | 18:44.52 |
| Robin_Watts | nice. | 18:44.56 |
| mvrhel_laptop | Robin_Watts: no what is castAR. googling.. | 18:45.12 |
| | ok cool | 18:45.36 |
| Robin_Watts | castAR is a seattle based kickstarter project. Couple of ex valve employees have come up with a really cool way of doing VR/AR. | 18:45.48 |
| mvrhel_laptop | is is this the open source stuff | 18:45.52 |
| Robin_Watts | No, it's hardware. | 18:46.06 |
| mvrhel_laptop | ok I see that now | 18:46.19 |
| Robin_Watts | They have glasses with twin projectors on that project images out into the world, and reflect back off a special surface (so that each image only goes the appropriate eye) | 18:46.47 |
| mvrhel_laptop | neat | 18:46.54 |
| Robin_Watts | and they track the glasses position 120Hz to sub millimetre accuracy, so as you move your head, the view updates. | 18:47.23 |
| mvrhel_laptop | Robin_Watts: one more question for you | 18:48.01 |
| Robin_Watts | They have been doing lots of demos, and seattle is a good place to see it. | 18:48.22 |
| | It's supposed to be worth seeing, so if you ever go to a maker faire or anything, watch out for it. | 18:48.37 |
| mvrhel_laptop | Robin_Watts: I will have to look into it more when I get back there | 18:48.37 |
| | I will | 18:48.43 |
| | so <c:ser> has multiple <c:dPt> entries | 18:49.26 |
| | is there any simple way that we can visit each one? | 18:49.56 |
| Robin_Watts | mvrhel_laptop: Urm... | 18:50.22 |
| mvrhel_laptop | if you don't know that is ok | 18:50.44 |
| | I will dig further | 18:50.50 |
| | Robin_Watts: ^^ | 18:50.54 |
| Robin_Watts | In the stuff I've been looking at in the ooxml agent, a handler gets called for each node. | 18:51.01 |
| | or a handler at the start of a node and again at the end. | 18:51.18 |
| | but it's possible your stuff is being processed in a different way. | 18:51.43 |
| mvrhel_laptop | I am in the handler for a pie chart | 18:51.56 |
| | and then there is a whole bunch of get node calls to get at the various parameters of the pie chart | 18:52.20 |
| Robin_Watts | just looking at code now... (oooh, MSVC is ill....) | 18:52.22 |
| | mvrhel_laptop: OK, so you have NodeMngr_findChildNode that finds a given child node... | 18:53.22 |
| mvrhel_laptop | yes | 18:53.32 |
| Robin_Watts | If you use NodeMngr_createChildIterator on the node, you can pass a tagId in, and then the iterator will return all the matching children. | 18:54.05 |
| mvrhel_laptop | ah ok | 18:54.14 |
| Robin_Watts | would that help? | 18:54.15 |
| mvrhel_laptop | cool | 18:54.16 |
| | yes that would help | 18:54.19 |
| | I have to destroy the iterator at the end iirc yes? | 18:54.35 |
| | I remember seeing that someplace | 18:54.39 |
| Robin_Watts | NodeMngr_getNext | 18:54.49 |
| | NodeMngr_destroyChildIterator | 18:54.56 |
| | yeah. | 18:54.58 |
| mvrhel_laptop | cool. ok. that helps a bunch. thanks Robin_Watts | 18:55.06 |
| Robin_Watts | best of luck! | 18:55.12 |
| mvrhel_laptop | thanks. I will need it | 18:55.21 |
| | Robin_Watts: have a good weekend | 18:55.34 |
| Robin_Watts | ta, you too. | 18:55.55 |
| mvrhel_laptop | marcosw_ : are you there? | 18:56.03 |
| marcosw_ | mvrhel_laptop: yes indeed. | 18:56.14 |
| mvrhel_laptop | marcosw_: ok great | 18:56.23 |
| | did you see that I had done a commit for the ICC issue that customer 532 reported | 18:56.40 |
| marcosw_ | no, I haven't looked at the commit logs in a couple of days. | 18:57.10 |
| mvrhel_laptop | there was no bug to close so I just wanted to let you know . I dont know what version they would need a patch | 18:57.10 |
| | also you may want to squash together two commits actually | 18:57.30 |
| | or even more. hold on | 18:57.48 |
| marcosw_ | okay. | 18:58.39 |
| mvrhel_laptop | marcosw_: so http://git.ghostscript.com/?p=ghostpdl.git;a=commit;h=d6be178195f97d7fa1cd2ae010e5d9e89f911e1a and | 18:59.11 |
| | this one | 18:59.21 |
| | http://git.ghostscript.com/?p=ghostpdl.git;a=commit;h=b13d19749b58b2b0263013ad9a516c36e56cdb1c | 18:59.22 |
| | marcosw_: I don't know what version they are running against though. | 18:59.59 |
| marcosw_ | I'll ask them and make a patch applicable to their revision (or just send them a new gsicc_create.c). | 19:00.31 |
| mvrhel_laptop | if there is a problem you could do my last 4 commits which are all in the same file and should go back quite a ways | 19:00.33 |
| | marcosw_ yes, just sending the new gsicc_create.c may be fine | 19:00.52 |
| | and easier | 19:00.58 |
| | thanks marcosw_ | 19:03.11 |
| marcosw_ | I've emailed them asking what version they are using. | 19:03.41 |
| shadowapex | Hi everyone, I was wondering if anyone knew if ghostscript supports any kind of GPU acceleration through CUDA, OpenCL, or something similar. I haven't been able to find anything about it in the depths of the internet. | 19:07.19 |
| Robin_Watts | shadowapex: It does not. | 19:08.04 |
| shadowapex | :< | 19:08.16 |
| Robin_Watts | But such a thing is possible. | 19:08.18 |
| shadowapex | Any plans in the future? | 19:08.27 |
| Robin_Watts | ghostscript has a 'device interface' at its core, and theoretically a device could be written to make use of GPUs etc. | 19:09.15 |
| | We have lots of commercial customers who write their own devices. | 19:09.24 |
| | some of them drive ASICs in laser printers to do rendering etc. | 19:09.36 |
| | It was something we seriously thought about a while ago, but the guy we hired who's job it was left the company for personal reasons before he started on it. | 19:10.27 |
| | We have no development bandwidth left in the company for it to be something we could look at right now though. | 19:11.14 |
| | But if you'd be interested in tackling it... talk to us :) | 19:11.26 |
| shadowapex | Haha, well thank you for the information. | 19:11.49 |
| | I'll try and see if I can convince our company to develop something like that. | 19:13.15 |
| Robin_Watts | shadowapex: which company? if you can say? | 19:13.54 |
| shadowapex | Robin_Watts: GotPrint | 19:16.41 |
| | We do online printing | 19:17.05 |
| | and we deal with a lot of image conversions and manipulations | 19:17.49 |
| Robin_Watts | right. | 19:20.51 |
| | woo hoo. tabs look right now, mostly. | 20:29.47 |
| henrys | marcosw: I notice I got a bug report for something that regressed in April. Did you just miss it or are you way behind on checking stuff? We can recruit eyes. | 21:38.21 |
| marcosw_ | it's not something that's tested by the nightly/weekly regression, I found it while looking for something else. I'm not behind, I also opened a regression bug for something that broke yesterday. | 21:44.28 |
| | ^that's tested^that's _not_ tested | 21:44.44 |
| | never mind, the original sentence was correct :-) I'm glad it's Friday.... | 21:45.34 |
| henrys | marcosw_: okay | 21:55.03 |
| | Forward 1 day (to 2014/06/28)>>> | |
IRC Logs