| <<<Back 1 day (to 2013/07/22) | 2013/07/23 |
mvrhel_laptop | hmm winrt project is broken... | 05:59.57 |
chrisl | paulgardiner: ping | 08:32.25 |
paulgardiner | chrisl: Hi | 08:35.49 |
| Arm build stuff to look at? | 08:36.30 |
chrisl | paulgardiner: good morning - yes, please: could you cast your eye over http://git.ghostscript.com/?p=user/chrisl/ghostpdl.git;a=commitdiff;h=f15df6ff | 08:36.43 |
paulgardiner | Sure | 08:36.50 |
chrisl | Mainly the change in winrtsup.cpp | 08:36.59 |
| Note that the project file changes don't work (yet) | 08:37.12 |
kens | Oh good grief... 8.71 doesn't work,9.07 does, can we have a patch for 8.71 ?..... | 08:47.57 |
Robin_Watts | wintitsup.cpp | 08:48.03 |
paulgardiner | kens: use the diffs between 8.71 and 9.07 to generate a patch and send it to them. :-) | 08:50.17 |
kens | Yeah! 3 years of patches should keep them busy for a while | 08:50.37 |
chrisl | paulgardiner: better still, generate multiple patches, one for each change, and send them the lot! | 08:51.15 |
kens | Fortunately, not my problem, Marcos can try and bisect it | 08:51.16 |
paulgardiner | chrisl: With the message "The fix in here somewhere. You may also like to take on the others" | 08:52.28 |
chrisl | paulgardiner: that would probably indicate to them how we feel about things, yes ;-) | 08:52.55 |
paulgardiner | chrisl: I may be reading this wrongly, but it looks to me like GetSystemTimeAsFileTime gives 10ths of microseconds, whereas GetTickCount was millisecs. Does that matter? | 09:06.05 |
chrisl | paulgardiner: my understanding is that it is only used to ensure a unique file name, so I figured it didn't matter | 09:06.44 |
paulgardiner | Ha. Yes. Silly me. Having written it, you'd think I'd have remembered that. | 09:07.18 |
chrisl | There's a probably a slightly higher probability of a collision due to the value wrapping around, but I doubt it's worth worrying about | 09:08.13 |
paulgardiner | msvc.mak changes assumes the default installation location for VS11, but I'm guessing that's what we do elsewhere | 09:10.48 |
chrisl | paulgardiner: I'm just tweaking that now. The env variable I used for the compiler path isn't set when you build from the GUI :-( | 09:11.47 |
paulgardiner | Pain | 09:12.15 |
chrisl | Yeh, I assumed when you used an nmake project the environment it used was the same as the relevant VS Tools Command Prompt - it isn't, quite :-( | 09:13.09 |
paulgardiner | Some of the make file changes look like they'll alter command lines for other builds, but I'm assuming that's intentional and benign | 09:14.28 |
| biab | 09:15.27 |
chrisl | paulgardiner: it has to, there wasn't sufficient granularity between the CC and CCAX variables for proper cross compiling, so I had fiddle things a bit. | 09:16.57 |
| paulgardiner: I also had to change stuff because several of the object files for the "aux" tools were being shared with the main build - that *really* doesn't work when cross compiling! | 09:18.40 |
| If you look at this revision: http://git.ghostscript.com/?p=user/chrisl/ghostpdl.git;a=commitdiff;h=85157910 | 09:18.52 |
| It removes most of the path assumptions in the original | 09:19.14 |
paulgardiner | Yeah. Look good to me. | 09:30.13 |
chrisl | Cool, thanks. I'll pull the changes onto my windows 7 laptop and make sure I haven't broken the "proper" VS builds before I push it to casper for real | 09:33.41 |
Robin_Watts | invents a new sport; sauna running. | 10:38.49 |
| You either need a sauna with a treadmill on it, and a shower head positioned to leak occasional drops of hot water on you... or a freak british summer. | 10:39.25 |
paulgardiner | :-) I can imagine how bad it is out there today. | 10:40.27 |
chrisl | We already have "sauna squash"....... | 10:40.48 |
kens | will stick to swimming, thank you :-) | 10:41.03 |
Robin_Watts | I will never think of Marcos building cluster nodes the same way again: http://motherboard.vice.com/blog/this-sexed-up-guide-to-building-a-pc-is-simply-the-worst | 10:50.44 |
paulgardiner | Aaggh! Surely one shouldn't hold a CPU like that! | 10:58.36 |
Robin_Watts | tor7: Crumbs. Sorry to head about your uncle. My sympathies. | 11:17.12 |
| s/head/hear/ | 11:17.19 |
| (Sorry, that was a genuine slip, not an attempt to be funny) | 11:17.40 |
paulgardiner | Yeah. Likewise, sorry to hear that. Shocking. | 11:23.33 |
tor7 | paulgardiner: Robin_Watts: thanks. he was only 59 so it was quite a shock. | 13:13.16 |
henrys | sorry to hear about that tor7 if it helps feel free to skip meetings today. | 13:38.57 |
| Congrats on the baby! the news here is full of folks celebrating but I don't know if the coverage exagerrates what's actually going on there. | 14:23.28 |
mvrhel_laptop | henrys: I have to run my daughter to her drawing camp this morning. leaving the house around 8:30 and then I will be back on line around 9:10 or so | 14:25.27 |
| so I will be likely leaving one meeting early and late for the other | 14:25.44 |
henrys | mvrhel_laptop: that's fine | 14:25.55 |
| Robin_Watts: amazing at your latitude - that you share with southern Alaska can have weather like that. | 14:34.28 |
Robin_Watts | henrys: Alaska was beautifully hot when we visited :) | 14:38.02 |
| henrys: The royal baby is merely an excuse to sell newspapers and for daytime TV/rolling news to have a field day. | 14:38.42 |
henrys | I got a bad sunburn sunday working out - spray on sun block - never again. | 14:43.30 |
Robin_Watts | yeah, I went to the "game fair" on sunday and caught the sun. | 14:44.15 |
| Must have been 10s of thousands of people there. was incredible. | 14:44.30 |
| And I bought a shotgun. | 14:44.36 |
henrys | game fair? | 14:44.43 |
Robin_Watts | It's a once a year fair for country pursuits. Shooting/Fishing/Hunting, plus huge sections for dogs and horses, and green energy and anything else like that. | 14:45.50 |
henrys | why does Skype say "This message has been removed", what message? | 14:55.01 |
Robin_Watts | Skype lets you edit the last line you sent. | 14:58.40 |
| Let me try with a message to you. | 14:58.44 |
| You can hit the right hand button on a line a "remove message" | 15:00.05 |
henrys | okay | 15:01.25 |
| meeting time | 15:01.45 |
| even though takani-san is focused on the gs side of things feel free to jump in on the Skype discussion | 15:02.20 |
| does anyone know the underlying PDF technology for good reader? | 15:02.59 |
Robin_Watts | not offhand | 15:03.35 |
henrys | #1 non apple app for the iPad 2010 that has to be quite a revenue stream. | 15:04.43 |
| if there are using mupdf we should find out. | 15:04.52 |
| can somebody investigate? | 15:05.33 |
chrisl | henrys: with comments like: "JPEG2000 image compression used in some PDF files is only supported by iOS 5.0 or later".... it doesn't sound like MuPDF | 15:06.11 |
Robin_Watts | I don't believe it's MuPDF. | 15:06.40 |
kens | Also it handles non-PDF formats, eg MS Word | 15:06.42 |
Robin_Watts | It's not out for android :( | 15:06.56 |
| which means it's much harder to grep for strings. | 15:07.05 |
henrys | sounds like Robin_Watts old institution | 15:07.10 |
chrisl | iOS has PDF reading built-in, doesn't it? | 15:07.21 |
Robin_Watts | I don't believe it's them either. | 15:07.22 |
| It did reflow way before we did. | 15:07.36 |
henrys | and annotations, if was quite a hit at chicago - people like drawing on pdf's with their finger afaict | 15:08.09 |
chrisl | With comments about features only being available in certain iOS versions, I wonder if good reader is using the built-in PDF imaging...... | 15:08.39 |
kens | I suspect it must be | 15:08.48 |
| Or at least, some parts of it | 15:08.56 |
Robin_Watts | http://www.goodiware.com/goodreader.html | 15:09.06 |
| KNOWN LIMITATIONS | 15:09.26 |
| | 15:09.28 |
| GoodReader heavily relies on iOS capabilities. Information about known iOS limitations that affect viewing files can be found here. | 15:09.29 |
kens | Yeah I had read thast | 15:09.35 |
| Not totallyclear if its the PDF stuff that is iOS dependent or 'somethign else' | 15:09.51 |
henrys | right that's where I got the "best app 2010" stuff | 15:09.56 |
| quartz2d does give you a pretty good chunk of a pdf app. | 15:10.33 |
Robin_Watts | In the section that points to, as ken says, it dives right in with PDF limitations. | 15:10.38 |
kens | Yeah but its not strictly PDF, JPEG2000, DRM, £D stuff | 15:11.13 |
| Fonts, MS Office files, etc | 15:11.32 |
| Odd that it doesn't support layers | 15:11.51 |
henrys | it is an impressive list of functionality but I don't have a sense how complete it is. | 15:13.45 |
| paulgardiner: how are signatures? | 15:14.37 |
| and I guess we should talk about the mupdf release? | 15:15.38 |
paulgardiner | henrys: still a way off but getting closer. I've been working on freetext annotations just lately. They involve creating appearance streams with text in, via the pdf-write device. That's something we need for the appearance streams associated with signatures, and is also potentially of use to our main customer | 15:17.23 |
Robin_Watts | henrys: I believe tor7 was planning to do an rc fairly soon, and then go for a release when I got back (mid august). | 15:17.35 |
| but recent events may have pushed the rc back of course. | 15:17.50 |
henrys | okay chrisl do you have any rc date for gs? | 15:19.08 |
chrisl | henrys: no | 15:19.38 |
paulgardiner | henrys: that said, I'm now not sure that freetext annotations will do quite what the customer is after (can't set freetext at an angle and can't make it partially transparent), but in the same way as they are a step towards signature appearance streams, the same techniques can be used for watermark annotations, which is what the customer really wants. | 15:19.49 |
mvrhel_laptop | I have to leave in a bit. Just a quick note that I am making good progress on the windows phone viewer. I will def. have it in the store before the meeting. the branch is named winphone and is in my repos | 15:20.23 |
henrys | mvrhel_laptop: oh cool any intel from MS? | 15:20.50 |
chrisl | henrys: I was going to take the release branch next Monday or Tuesday, then marcosw can start his testing. RC *at least* a week later..... | 15:21.01 |
Robin_Watts | mvrhel_laptop: Anything you want to add to: http://ghostscript.com/~robin/CHANGES ? | 15:21.06 |
mvrhel_laptop | Robin_Watts: no that all looked good. sorry I had not replied to yo on that | 15:21.32 |
Robin_Watts | no worries. | 15:21.39 |
henrys | paulgardiner: that's fine sounds like progress. | 15:22.08 |
mvrhel_laptop | henrys: Nothing yet. I had another contact start looking for me last week | 15:22.09 |
henrys | whatever they're doing wall street doesn't like it. | 15:22.34 |
mvrhel_laptop | yes their revenue dropped | 15:23.26 |
| ok. i will be back a bit late for the gs meeting.... | 15:25.05 |
henrys | hopefully everyone will have a chance to glance at the agenda and make sure things are progressing, other than that let's call it over. | 15:25.05 |
Robin_Watts | henrys: Do you need anything written for the newsletter over and above the CHANGES thing ? | 15:27.09 |
| or can that keep until I'm back? | 15:27.23 |
henrys | Robin_Watts: I think that's fine. I can always make stuff up if I think the newsletter sucks ;-) | 15:28.03 |
Robin_Watts | ok :) | 15:28.13 |
| http://stackoverflow.com/questions/10859173/how-to-disable-all-caps-menu-titles-in-visual-studio | 15:40.15 |
kens | Robin_Watts : I'm amazed there is such a simple solution | 15:42.33 |
Robin_Watts | It's proof that not all the programmers at MS have drunk the UI coolaid. | 15:43.03 |
kens | Indeed :-) | 15:43.20 |
Robin_Watts | "Oh, corporate is insisting that we make menus all caps are they? Fair enough, time for another registry entry" | 15:43.39 |
kens | Exactly :-) | 15:43.55 |
| Now, I still need my coloured icons back.... | 15:46.10 |
| And an MDI interface | 15:46.16 |
Robin_Watts | VS has never had an MDI interface though, has it ? | 15:46.47 |
kens | Well, I mean multiple documents in the main windiw | 15:47.09 |
henrys | wow this might make marcosw job easier, a one computer cluster: http://arstechnica.com/information-technology/2013/07/ibm-unveils-new-mainframe-for-the-rest-of-us/ | 15:54.32 |
Robin_Watts | A dollar per day per linux instance. So $520 a day. | 15:55.46 |
| $200000 a year. Can we stick to the existing cluster please? :) | 15:56.22 |
henrys | that's what it is capable of running | 15:57.20 |
Robin_Watts | henrys: And the price per instance will surely be lowest when it's running all the instances it can manage, right? | 15:58.09 |
henrys | It's unclear to me, can we buy it for 75K and install our own linux... | 15:59.00 |
| ? | 15:59.01 |
| it does say it is cheaper than commodity hardware I don't know if that scales down. | 15:59.47 |
Robin_Watts | We'd also not be testing on x86/amd64 any more, it'd be "some wierdass ibm architecture" | 16:00.31 |
ray_laptop | henrys: how many cpu's do you get for 75k -- I'm looking for the specs | 16:00.33 |
henrys | it says 1/2 the cost of equivalent x86 computing power | 16:01.02 |
ray_laptop | Robin_Watts: right. And how does 4.2GHz on an ibm chip compare to 3+ GHz on an AMD or i7 | 16:01.15 |
Robin_Watts | ray_laptop: Regardless of the speed, how well does it test our SSE code? :) | 16:01.39 |
ray_laptop | henrys: right. sure. If you buy PREMIUM servers from somebody, MAYBE | 16:01.46 |
| considering that each cluster node that Marcos does is < 1K that has to be one heck of a server to equal 75 AMD's | 16:03.00 |
henrys | I just think it would be cool to have the "Artifex" mainframe | 16:03.06 |
Robin_Watts | meeting time? | 16:03.33 |
henrys | yes | 16:04.08 |
| chrisl:so I'll send that patch to luratech and ask them about it. | 16:04.36 |
| alex's patch that is. | 16:04.49 |
marcosw_ | the latest cluster node, fermis, has two Xeon 4 core processors (so 16 threads) and 36 gigs of memory. it's an off lease 1U box that I bought for <$600. | 16:05.09 |
chrisl | henrys: okay, there was two missing | 16:05.10 |
marcosw_ | http://www.theserverstore.com/content/dell-poweredge-c1100-server | 16:05.36 |
henrys | wow | 16:05.55 |
marcosw_ | the only problem is that it's noisy, so miles didn't want it in his office. It's living in my garage :-) | 16:06.33 |
henrys | anyone want to talk about the release. Do we have blockers? | 16:06.40 |
marcosw_ | henrys: no blockers as far as I know, now that the Luratech JPEG2000 bug is fixed. | 16:07.03 |
chrisl | There are none listed - that's nice..... | 16:07.32 |
henrys | looking at the agenda ... | 16:07.56 |
ray_laptop | marcosw: You bought one ? | 16:08.19 |
| oops. nm | 16:08.30 |
| I didn't read far enough back | 16:08.42 |
henrys | ray_laptop:seem like the PSFit documentation could make it in before the release - safe and easy. | 16:08.50 |
ray_laptop | henrys: It's been in | 16:09.21 |
henrys | oh I missed it. | 16:09.33 |
| sorry | 16:09.35 |
chrisl | henrys: I'd like to get the WinRT/ARM changes into the release | 16:09.36 |
henrys | mvrhel_laptop: do you have a large change before the release? | 16:10.20 |
| gs wise? | 16:10.30 |
mvrhel_laptop | good grief. I don't know why the 8.71 customer does not update after all the work I did for them | 16:10.42 |
| henrys: no nothing that is close enough | 16:10.55 |
ray_laptop | I'd like to get the saved-pages changes in. Almost done, just cleaning up | 16:11.01 |
henrys | mvrhel_laptop: yeah I saw that perhaps we need a gentle push back. | 16:11.06 |
| saved pages is only going to affect the system if an option is used, right? | 16:11.41 |
ray_laptop | even if it isn't terribly well tested, by getting it into the release, we'll start tackling the bug tail | 16:12.11 |
chrisl | mvrhel_laptop: do have any sort of feel for what should go in a Ghostscript arch.h file for WinRT/ARM? | 16:12.18 |
ray_laptop | henrys: right. If you don't turn it on, it just operates normally | 16:12.36 |
| I do have to do the documentation on it yet | 16:12.51 |
Robin_Watts | I should try to get the lcms2.5 changes in this week. | 16:13.15 |
henrys | ray_laptop:about saved pages - comment 15 http://bugs.ghostscript.com/show_bug.cgi?id=693121 ?? Is that fixed? | 16:14.34 |
mvrhel_laptop | chrisl: no sorry. I can look at the options that come up in the new version of VS when I change to ARM and that is about it | 16:14.36 |
Robin_Watts | mvrhel_laptop: See the link I posted above about VS2012 (you were disconnected at the time) | 16:15.04 |
chrisl | mvrhel_laptop: Okay, I'm fairly sure it's little endian, it's things like the size of pointers and whether we have a 64 bit integer, etc. Do you have an ARM device for it? | 16:15.42 |
ray_laptop | henrys: no. and opening clist files with 'delete on close' (or unlinking them on linux after open) is totally incompatible with saved pages. | 16:15.47 |
tor7 | henrys: Robin_Watts: sorry, got distracted by programming... yes, I plan on assembling an RC this week. | 16:16.04 |
| (was getting the opengl device to tiger!) | 16:16.17 |
Robin_Watts | tor7: Nice! | 16:16.24 |
tor7 | ugly ugly hack so far though | 16:16.36 |
mvrhel_laptop | chrisl: yes, I am building and running my mupdf apps for both the surface and a nokia lumina phone both of which have an ARM | 16:16.37 |
tor7 | but it runs :) | 16:16.41 |
chrisl | mvrhel_laptop: so we could trying building genarch.exe and running it on the device? | 16:17.10 |
henrys | ray_laptop:well can the bug be updated to reflect the current state of affairs, it sounds like a fix is available so nobody will work on it. Is it something we could make bountiable? | 16:17.30 |
mvrhel_laptop | yes. if we can build gslib then I could write a very simple app to call into it also | 16:17.57 |
ray_laptop | henrys: OK. I'll update the bug with what I've learned so far (specifics) | 16:18.08 |
henrys | thanks ray_laptop | 16:18.24 |
mvrhel_laptop | Robin_Watts: I don't see your link in the logs | 16:19.44 |
Robin_Watts | http://stackoverflow.com/questions/10859173/how-to-disable-all-caps-menu-titles-in-visual-studio | 16:19.57 |
henrys | marcosw_: anything you want to say about the release - bug/customer wise? | 16:20.00 |
mvrhel_laptop | oh I thought it was about ARM sorry | 16:20.11 |
Robin_Watts | mvrhel_laptop: sorry, no. | 16:20.17 |
marcosw_ | henrys: no, i think we are good for the release. We have several customers waiting for it, so sooner is better. | 16:20.40 |
Robin_Watts | chrisl: I can probably make some educated guesses at what should be in a gs arch file for ARM (assuming surface runs on 32bit ARMs) | 16:21.07 |
henrys | kens:any high level output stuff needed for the release? I guess you would have said something but just checking? | 16:21.32 |
Robin_Watts | I suspect you just want to copy the 32bit x86 one. | 16:21.34 |
chrisl | Robin_Watts: sure, that's what I'm hoping. I just don't trust MS to do the sane thing! | 16:21.59 |
kens | henrys I'm just carrying on with the colour stuff for pdfwrite/ps2write. I won't be able to finish it for this release, but hope to make enough progress to talk about it, and allow people to play with it | 16:22.07 |
henrys | marcosw:have you done any benchmarking of the xeon beast to see how it performs relative to the other nodes? | 16:22.27 |
| marcosw_ ^^^ | 16:22.36 |
marcosw_ | henrys: is much faster, 2x to 3x faster than the other nodes:-) except for the updated i7 node that is now a 6 core i7 over clocked to 4.3 GHz, but that cost $1500 to build and it's only slight faster. | 16:23.58 |
henrys | I'm looking the max memory on that baby 144 GB wow | 16:24.06 |
chrisl | Hmm, can't (easily) build genarch.exe for WinRT/ARM :-( | 16:24.06 |
ray_laptop | we could put /tmp in ramdisk (or does linux cache /tmp adequately well) / | 16:25.19 |
| ? | 16:25.21 |
marcosw_ | I originally bought it to be the bugzilla server, but since it's too noisy to live in miles' office I re-purposed one of the other cluster nodes. | 16:25.34 |
ray_laptop | marcosw: is that one online as a cluster node yet ? | 16:25.57 |
marcosw_ | ray_laptop: yes, it's fermis | 16:26.12 |
ray_laptop | marcosw: thanks. | 16:26.24 |
| I was wondering if the xeon was fully compatible with the x86 and AMD's -- I guess it must be | 16:27.04 |
henrys | the Mac Pro has been on the cluster for years | 16:27.27 |
| it is a xeon | 16:27.40 |
marcosw_ | and my macpro is a xeon as well (it's cluster node name is xeon). | 16:27.52 |
ray_laptop | marcosw: so do you have room for extra 1U units. At $349 each, that seems like a good price. Another $100 for a HD | 16:28.33 |
henrys | I think we've only seen os or library configuration differences | 16:28.41 |
Robin_Watts | SSD surely? | 16:28.54 |
| or hybrid at least. | 16:29.05 |
ray_laptop | oops. My wife just called. I have to run home and let her in the house. I came over to the office while she was still over at the church (across the street). She didn't have her keys. bbiab | 16:29.58 |
marcosw_ | Robin_Watts: I tried an SSD but it wasn't noticeably faster as a cluster node than with a traditional hard drive. | 16:30.27 |
henrys | with these huge memory configurations should we be compiling in ram? | 16:30.43 |
Robin_Watts | marcosw_: Interesting. Maybe linux caches the exes well enough then. | 16:31.05 |
marcosw_ | the compile time is ~20 seconds, so I don't think it's worth any effort. | 16:31.12 |
henrys | 1 minute after the meeting ;-) | 16:31.44 |
chrisl | mvrhel_laptop: can you just copy an exe onto the ARM device and run it, or is there more involved? | 16:33.42 |
Robin_Watts | Given you have to sign and deply windows 8 metro apps, I can't believe you won't have to do the same for ARM ones. | 16:34.35 |
mvrhel_laptop | chrisl: in that respect the surface is no different than any other windows device, so yes I could just get it and run it | 16:34.36 |
chrisl | Oh, cool :-) | 16:34.49 |
mvrhel_laptop | Robin_Watts: I can run in desktop mode | 16:34.57 |
Robin_Watts | oh, right. | 16:35.03 |
| WinRT has a desktop mode? | 16:35.17 |
mvrhel_laptop | the surface does | 16:35.29 |
Robin_Watts | With standard WIN32 APIs ? | 16:35.40 |
mvrhel_laptop | well it won't run win32 apps if that is your question | 16:35.54 |
| they have to be built for the ARM | 16:36.02 |
Robin_Watts | right, but I could take a win32/x86 app written in C, and rebuild it for win32/ARM and all the APIs are there? | 16:36.44 |
ray_laptop | I would sure be surprised if they were | 16:37.10 |
Robin_Watts | I'm not expecting the ARM device to emulate x86. | 16:37.26 |
mvrhel_laptop | Robin_Watts: I doubt all the net API calls are available for a developer | 16:37.30 |
marcosw_ | mvrhel_laptop: in terms of release bugs could you look at <http://bugs.ghostscript.com/show_bug.cgi?id=693826>? It's not a blocker, but it would be nice to understand what is going on. | 16:37.32 |
mvrhel_laptop | marcosw_: ok. yes I remember this one. I will take a look at it this week | 16:38.38 |
chrisl | mvrhel_laptop: so, when you have a chance, could you try: http://www.ghostscript.com/~chrisl/genarch.exe | 16:38.43 |
| mvrhel_laptop: invoke it with genarch.exe <output file>.h | 16:39.04 |
mvrhel_laptop | chrisl: ok. It will be later this afternoon (after you are gone). at coffee shop this morning until art camp is over for my daughter. surface is at home | 16:39.43 |
marcosw_ | ray_laptop: and for the logs, could we get <http://bugs.ghostscript.com/show_bug.cgi?id=693121> into the gs9.08 release? | 16:39.45 |
chrisl | mvrhel_laptop: that's cool, I'm not in a desperate hurry - I've done the hard part of the ARM changes :-) | 16:40.23 |
ray_laptop | marcosw: henry and I were just discussing this. No, it won't be in the release (probably) | 16:42.33 |
| marcosw: for one, I need to revamp the clist open in order to make it compatible with saving the page as clist files | 16:43.26 |
| and it still is funky on windows. | 16:44.01 |
kens | OK I'm off, goodnight all | 16:52.17 |
ray_laptop | kens: marcosw: I fixed that 144 dpi segfault. | 16:52.19 |
kens | ray_laptop : good to hear :-) | 16:52.26 |
ray_laptop | a one char change | 16:52.36 |
mvrhel_laptop | ray_laptop: nice | 16:52.40 |
kens | :-O | 16:52.41 |
ray_laptop | classic OBO < instead of <= | 16:53.00 |
| OK. Back to saved-pages.... | 16:53.30 |
henrys | oh I forgot to ask about altona Robin_Watts is that a regression? | 17:02.48 |
Robin_Watts | hmm? | 17:03.01 |
henrys | yes you said you'd have a look at a bug. Let me see if I can find it. | 17:03.38 |
Robin_Watts | oh, I remember, yes. Haven't looked yet, sorry. | 17:04.09 |
| 694311 comment 4 | 17:05.27 |
henrys | yes... | 17:05.38 |
| and if anybody wants to weigh in on comment # 3 - maybe kens has an idea off the top of his head. | 17:06.41 |
| oh he left | 17:07.02 |
| if chrisl or ray_laptop don't have any ideas on the language front I'll try and debug it later. | 17:07.49 |
chrisl | henrys: I don't follow - I thought he was saying that the OpenJPEG 2 branch had these problems, OpenJPEG 1 doesn't | 17:12.52 |
henrys | chrisl: he was but it does sound suspiciously integration oriented, I'll have a look. | 17:13.46 |
Robin_Watts | henrys: I *thought* that mupdf, at the point at which we integrated v2 had no differences to our old v1 one. | 17:14.26 |
| But I will test that. | 17:14.31 |
henrys | chrisl:my understanding is those files do work in mupdf openjpeg 2 which suggests the integration issue. | 17:16.15 |
chrisl | henrys: "Both of these errors are visible in mupdf" | 17:16.35 |
henrys | I'm talking about comment #3 | 17:16.56 |
| chrisl ^^^ | 17:17.56 |
chrisl | henrys: Oh, sorry, Robin mentioned comment #4 | 17:18.03 |
| henrys: if you like, I can take a look at it tomorrow, see what's happening in the PDF interpreter - I could use a few hours break from build hacking! | 17:19.51 |
henrys | chrisl: I would be thrilled if you did exactly that. | 17:20.18 |
| ;-) | 17:20.25 |
chrisl | henrys: you're surprisingly easily pleased! Yeh, okay, I'll look tomorrow | 17:20.43 |
| on that note, have to go - 'nite all | 17:22.05 |
ray_laptop | oh, yuck. The PDF interpreter SMask stuff is a mess, and made all the worse by having JPXDecode having SMask buried inside | 17:22.20 |
| I am pretty sure that GS processes the JPX stream once to extract the SMask and then the second time to process the actual image, but it's been a while since I looked at it. | 17:23.34 |
| but why it would fail to then apply the SMask doesn't occur to me offhand. I'll let chrisl look into it so I can concentrate on the saved-pages finishing, and then look at it tomorrow | 17:25.44 |
| mvrhel_laptop: The andevcon people have your email address wrong. I'll forward it to you. They want you to pre-register for a savings by Friday 7/26 | 17:50.08 |
Robin_Watts | ray_laptop: You'll get about 800 of those mails before the show. | 17:52.52 |
mvrhel_laptop | ray_laptop: sorry about that | 17:55.09 |
monxalo | hi guys :), may i have some insights on the Android code, mainly on how the Hq patch is applied? | 17:59.22 |
Robin_Watts | monxalo: Hi. What exactly do you mean? | 18:09.51 |
monxalo | So, i'm developing a TwoPageView, i got it working by combining 2 bitmaps to display on the screen, now the hard part comes for zooming :). | 18:10.59 |
Robin_Watts | ah, right. | 18:11.21 |
| I suspect you may need paulgardiner to weigh in on this, but he's away for the night now. | 18:11.37 |
| What timezone are you in? | 18:11.41 |
monxalo | GMT+1 (DST) | 18:12.15 |
Robin_Watts | ok, so he should be back in your morning then, he's in the UK too. | 18:13.49 |
| monxalo: So you're combining 2 bitmaps and displaying them as a single 'view' ? | 18:14.15 |
| That sounds like the wrong way to do it to me. | 18:14.42 |
| A purer way (more in keeping with the structure of the code) would be to redo the code that decides where to put views, so as to place them in pairs. That may be the 'adapter'. | 18:15.26 |
| This is why you really need paulgardiner rather than me :) | 18:15.35 |
monxalo | Yes, i tried that before :), merely changing the adapter and combine 2 MuPDFViews, but had too many bugs. | 18:17.10 |
Robin_Watts | What do you mean by "combine 2 MuPDFViews" ? | 18:17.54 |
| AIUI, you should only need to change the Adapter. | 18:18.08 |
| (and tell MuPDF to cache +/- 6 pages rather than 3 or something. | 18:18.31 |
monxalo | join 2 mupdfview into a container (LinearLayout) side by side. | 18:19.19 |
Robin_Watts | ok, now you have lost me. | 18:19.48 |
| It's possible that you were trying exactly the right thing, but my Android-foo may not be strong enough. | 18:20.13 |
monxalo | eheh, no worries :). I'll come back tomorrow. | 18:21.14 |
| Thanks anyway. | 18:21.28 |
Robin_Watts | np. | 18:22.34 |
dogisfat | What is the expected behavior when you have a line inside of a clip path? | 18:23.50 |
Robin_Watts | dogisfat: eh? | 18:24.08 |
dogisfat | Robin_Watts: | 18:24.17 |
| Say I have a rectangular clip path inside of which is a line | 18:24.44 |
Robin_Watts | Generally you see a clip path and a winding rule together. | 18:24.46 |
| Imagine you fill the clip path using the winding rule. | 18:25.11 |
| You then conveptually use that drawn shape as a mask for the contents. | 18:25.30 |
dogisfat | Robin_Watts: So anything inside of the mask should show through, correct? | 18:26.08 |
Robin_Watts | So if you have a rectangular clip path and you have contents which are a line, then you'll see the subsegment of the line that crosses the rectangular clip path. | 18:26.13 |
| yes. | 18:26.15 |
dogisfat | Ok, that is what I thought, I think I am having a problem with my rendering package... bummer | 18:26.52 |
Robin_Watts | monxalo: For the logs: I think it's ReaderView.java you need to play with. | 18:30.48 |
| That seems to calculate where to position pages in the overall scheme of things. | 18:31.01 |
| Forward 1 day (to 2013/07/24)>>> | |