| | <<<Back 1 day (to 2012/09/24) | 2012/09/25 |
| Bhuvnesh | Hey Robin , you there ? or any other muPDF developer around here ? | 05:39.01 |
| mvrhel | good night all | 06:18.24 |
| sebras_ | Robin_Watts: looks like it would be useful to have ghostbot react to the word mupdf whenever the mupdf developers are sleeping... | 07:52.46 |
| | how come they always loing for five minutes and ask questions in the middle of the night..? I'm puzzled. | 07:53.09 |
| kens | US hobbyists asking in their evening ? | 07:53.36 |
| sebras_ | kens: maybe... or .cn ebook-reader workers starting out with a new problem in the morning... | 07:55.03 |
| kens | Their English usually seems reasonable... | 07:55.20 |
| sebras_ | one never knows. looks like Bhuvnesh has been in contact with Mr. Watts at least. | 07:55.32 |
| kens | Ah yes, I saw him too after I'd packed up too. | 07:55.50 |
| sebras_ | true enough. :) | 07:55.52 |
| kens | chrisl any ideas why ssh-add won't workk with my RSA key on a Mac ? Keeps sayign 'bad passphrase' | 07:56.30 |
| | Same key works on Windows and Linux.... | 07:56.45 |
| | Its not desperate, I'm copying the GS sources from my PC instead, its just irritating. | 07:57.49 |
| | I think I need more coffee to face a Mac | 07:58.57 |
| chrisl | kens: I don't know - I usually just copy the keyfiles into .ssh, rather than use ssh-add | 08:06.02 |
| kens | chrisl I did copy them into ~/.ssh | 08:06.21 |
| | You find that's enough ? Its not enough for me on Linux or (apparently) on Mac. | 08:07.00 |
| | On Linux I then run ssh-agent and ssh-add my key | 08:07.16 |
| chrisl | Just copying the files from the .ssh directory on one machine to another has always worked for me | 08:07.59 |
| kens | Interesting, wish it did for me... | 08:08.13 |
| chrisl | From where did you get the key files? | 08:09.51 |
| kens | Off my PC (I can't run my Lilnux VMs in this version of player) | 08:10.17 |
| chrisl | From the msys .ssh directory? | 08:10.47 |
| kens | Nope, from Putty | 08:10.56 |
| | My MingW is using my Putty keys too | 08:11.15 |
| chrisl | Hmm, doesn't putty hold the keys in its own format? | 08:11.26 |
| kens | Hmm, didn't know that. | 08:12.03 |
| | What shoudl hte format look like ? | 08:12.21 |
| chrisl | "-----BEGIN RSA PRIVATE KEY-----" followed by they key, then -----END RSA PRIVATE KEY----- | 08:13.32 |
| kens | Hmm, mine says BEGIN SSH2 PUBLIC KEY | 08:14.10 |
| | The .ppk files do seem to be in a PuTTy format | 08:14.28 |
| chrisl | Well, you really need the private key, or key pair, don't you? | 08:14.51 |
| kens | I thought that's what I had here.... | 08:15.10 |
| | I guess not. | 08:15.27 |
| | I guess I can log into casper and pull them from there | 08:15.39 |
| chrisl | Well, if you can log into casper, you must have your private key there! | 08:16.09 |
| kens | Hmm, assuming I have thee pair there... | 08:16.11 |
| | chrisl I can do that from Windows :-) | 08:16.21 |
| chrisl | IIRC, putty can export your key pairs into the "standard" format | 08:16.53 |
| kens | Like I said, I thought that's what I had | 08:17.12 |
| chrisl | I think you have to specifically tell it to include the private key - I think it defaults to just the public key | 08:17.40 |
| kens | I'll do it again.... | 08:18.12 |
| | Hmm, can't see any obvious way to export the keys | 08:20.48 |
| chrisl | PuttyGen? | 08:22.01 |
| kens | Yeah I'm reading the manual on it now | 08:22.10 |
| chrisl | In there you can "Export OpenSSH key" | 08:22.57 |
| kens | I believe so, I'm juist loading my existing keys | 08:23.10 |
| | computer running slowly because its copying the entire ghostpdl directory | 08:23.33 |
| chrisl | I don't think I ever got putty working, so stabbing in the dark a little! | 08:24.06 |
| kens | I used it to genmerate my keys in the first place, and I use PLink and Pageant for authentication | 08:24.29 |
| | OK well that saved a file, lets see | 08:25.16 |
| chrisl | I gave up and used cygwin instead..... | 08:25.20 |
| kens | OK that file begines ----BEGIN RSA PRIVATE KEY---- | 08:26.38 |
| | So lets see. | 08:26.43 |
| | Hmm, won't talk to me now, its gone into a brown study.... | 08:30.31 |
| | Could be time for the reset button | 08:30.53 |
| chrisl | Might be just slow? | 08:31.16 |
| kens | Slow I can live with, this would be glacial | 08:31.35 |
| | Its probably my fault for copying files to the wrong place, I'm still not used to the way it presents the file system | 08:31.57 |
| kens | presses the reset | 08:32.27 |
| | Aha, that seems to work | 08:35.45 |
| | Lets see if I can login to casper now | 08:35.52 |
| | And indeed I can. Thanks chrisl | 08:36.18 |
| chrisl | NO | 08:36.29 |
| | problem .... even! | 08:36.41 |
| kens | Ah, wondered what teh big no was for :-) | 08:36.54 |
| chrisl | Hit the wrong key - feeling very tired this morning :-( | 08:37.26 |
| kens | Hmm, looks like I may have to use Git to copy files, copying from Windows has left CR/LF pairs in the files, and teh Mac isn't happy about it. | 08:38.01 |
| chrisl | It may have dos2unix on it? | 08:38.29 |
| kens | Nope, I tried | 08:38.41 |
| chrisl | Does it have git on it? | 08:39.00 |
| kens | SHould do I spent 4 hours installing it yesterday | 08:39.18 |
| | and typing git gives me the usual help | 08:39.54 |
| chrisl | Well, probably better doing a git checkout, then, anyway - at least it's there for future use | 08:40.40 |
| kens | Going to try editing autogen.sh with XCode | 08:41.54 |
| | Oh great, I don't have autoconf installed.... | 08:44.39 |
| | Ah back to dependency hell again :-( | 08:50.41 |
| chrisl | What are you using to install the tools? | 08:51.33 |
| kens | Macports | 08:51.43 |
| chrisl | Doesn't that sort out dependecies for you? | 08:52.12 |
| kens | Yes, that's what I mean | 08:52.20 |
| | The reason it took 4 hours to install Git yesterday was because of the plthora of other stuff it insisted on installing | 08:52.42 |
| chrisl | Oh, and macports builds everything as it installs, IIRC | 08:53.06 |
| kens | Yes indeed, everythign is copied as source and built as you go | 08:53.20 |
| | I copied 4GB of data yesterday.... | 08:53.40 |
| chrisl | For git? That seems *way* OTT!! | 08:54.03 |
| kens | Wow, it only installed 2 new dependencies before working on autoconf | 08:54.03 |
| | Well 1.5 GB was for XCode | 08:54.14 |
| | Git (and all teh dependencies) probably only ran to 1 or 2 GB | 08:54.47 |
| | OK so apparently autoconf is ready to go | 08:55.02 |
| | And its actually doing something, so far so good | 08:55.47 |
| chrisl | Oh, macports has "core" packages, IIRC, so most was probably not for git, but for the core. And that core probably has most of the autconf dependencies already in it | 08:56.07 |
| kens | Grr it failed on ./configure this time | 08:56.38 |
| | complaining about ^M | 08:56.49 |
| | I htink I may have to do a fresh checkout | 08:57.01 |
| | OK progress, make is running for GS now :-) | 09:10.39 |
| chrisl | kens: did you remember the workaround for the system/macports libraries? | 09:16.59 |
| Robin_Watts | Is it just me, or do bug reports that say "This needs to be fixed immediately" from free users make you think "what else can I do first?" | 09:17.08 |
| kens | Nope. | 09:17.09 |
| | Robin_Watts : I've flet that way for years :-) | 09:17.26 |
| | chrils waht is the work around ? | 09:17.37 |
| chrisl | Robin_Watts: yes, every time! | 09:17.37 |
| | kens: do: LDFLAGS="-L/usr/lib" ./configure | 09:18.21 |
| kens | Ah, OK best I stop this and restart then | 09:18.35 |
| Robin_Watts | kens: I don't need to do that. | 09:20.26 |
| kens | Oh well, I just started configure again with that | 09:20.41 |
| Robin_Watts | I have to configure --with-lib-iconv=no or something like that, but no LDFLAGS stuff. | 09:21.18 |
| chrisl | Robin_Watts: different solution, same problem | 09:21.38 |
| kens | I'm not supplying any other switches to configure... | 09:21.42 |
| | OK so I should be godd, that's fine then | 09:21.50 |
| Robin_Watts | ok. | 09:21.51 |
| kens | I suppose I shoudl have installed X as well. | 09:22.12 |
| chrisl | You shouldn't need X | 09:22.26 |
| kens | But will I get a display window ? :-) | 09:22.40 |
| | OK make is off again | 09:22.59 |
| Robin_Watts | You will need to install XQuartz | 09:23.02 |
| chrisl | No, but I assume you want to test pdfwrite, and not x11 devices! | 09:23.18 |
| kens | Right now, I'm only bothered by testing pdfwrite | 09:23.25 |
| | In future it would probably be a good idea to see output too. | 09:23.39 |
| Robin_Watts | http://xquartz.macosforge.org/landing/ | 09:23.45 |
| kens | But to be honest I'll be happy to have a working binary today | 09:24.01 |
| chrisl | Robin_Watts: both the Darwin "system" and macports include implementations of iconv, so we either leave it out, or make sure that both the configure tests and the actual build find the same instance. | 09:24.35 |
| kens | excellent, seems I have a working binary | 09:33.18 |
| | Now to run the test. | 09:33.23 |
| chrisl | I have to go collect some cat medicine - bbiab....... | 09:33.44 |
| kens | And (surprise surprise) the files work perfectly for me, just as henry stated | 09:46.41 |
| chrisl | kens: if thre's much more hassle on this one, I can create a release archive for them (either current master, or 9.06 + patch) - but let's not offer that quite yet. | 10:11.15 |
| kens | chrisl I agree completely tehre | 10:11.32 |
| | I'm as convinced as I cna be that they have messed up applying the patch on teh Mac (their inability to build the snapshot inclines me further in that direction) | 10:12.15 |
| | I suspect the reason that they have gone quiet is because they don't want to admit it.... | 10:12.39 |
| chrisl | Their (apparent) inability to type/copy'n'paste into a bugzilla text window worries me a little! | 10:13.10 |
| kens | But it is, just conceivably, possible that there is something else causing a problem. | 10:13.18 |
| | So I'm leaving it open, also to rub in the fact that they haven't responded yet | 10:13.54 |
| chrisl | It seems unlikely, given the apparent problem, that it would work on Windows, but not Mac..... | 10:14.31 |
| kens | Yes, that was what first made me suspicious | 10:14.43 |
| | Anyway, as soon as this build (X support I hope) is completed, I'm going back to linearisation. | 10:15.18 |
| chrisl | And I will go back to trying to work out why the cluster is giving different output than any executable I run myself :-( | 10:17.57 |
| | Hmm, and suddenly my git difftool isn't working right.... | 10:26.12 |
| kens | Good I now have GS and X on Mac. | 10:27.32 |
| paulgardiner | Robin_Watts: ping | 13:53.16 |
| Robin_Watts | pong | 13:53.41 |
| sebras_ | ttl: 7 seconds. | 13:54.05 |
| paulgardiner | Got some more commits if you have a spare moment. | 13:54.39 |
| Robin_Watts | ok. let me look. | 13:55.23 |
| paulgardiner | Ta muchly | 13:55.33 |
| Robin_Watts | 6 of them ? | 13:56.06 |
| paulgardiner | Yes, sorry. But they are smallish, and some trivial | 13:56.30 |
| Robin_Watts | wasn't complaining, just didn't want to miss any. | 13:56.44 |
| | The first one... this one about executing menu items. | 13:57.44 |
| | It strikes me that that scales very badly with multiple languages. | 13:57.55 |
| | Let me phone you. much easier. | 13:58.22 |
| kens | Hmm, interesting. I can build my code fine here on WIndows, but on the cluster it gives me an error on dmprintf passing 5 argumens but only takes 2.... Curious. | 14:01.57 |
| Robin_Watts | paulgardiner: In the launch URL one... | 14:06.48 |
| | In the last chunk of the diff (pdf_js.c) | 14:07.03 |
| | should argc == 1 correspond to bNewFrame = 0? | 14:07.25 |
| paulgardiner | Yes. I'll change that. | 14:08.21 |
| Robin_Watts | And you have some code that checks for overflows by hdr_len = .... if (hdr_len > 512) etc. | 14:09.11 |
| | 512 is a magic number. | 14:09.15 |
| | and I'm really confused by the warn code in the doc.mailDoc commit. | 14:10.30 |
| | Oh! I get it. cunning. | 14:10.42 |
| paulgardiner | Be worth giving 512 a name. I wish we could use snprintf. Maybe we should write our own sometime, or steal BSD's | 14:12.16 |
| sebras_ | paulgardiner: vsnprintf() is already used...? | 14:13.33 |
| paulgardiner | sebras_: Really?! Hmmm. | 14:14.17 |
| Robin_Watts | snprintf is used in pdfapp.c and x11_main.c | 14:16.39 |
| | and pdf_device.c | 14:16.46 |
| | and base_error.c uses vsnprintf | 14:16.56 |
| | paulgardiner: I wonder if you fallthroughtastic switch in the last commit should have a default clause at the top? | 14:18.05 |
| paulgardiner | Oh ok. I thought there was some reason we shouldn't. | 14:18.06 |
| Robin_Watts | default: if (argc <= 0) break; | 14:18.17 |
| | which then falls through to case 6: etc. | 14:18.28 |
| paulgardiner | Ah yes. Maybe. | 14:19.08 |
| | Robin_Watts: Was that last comment about mailDoc? | 14:32.08 |
| Robin_Watts | Yes. | 14:32.28 |
| paulgardiner | I test argc is between 1 and 6 earlier | 14:32.57 |
| Robin_Watts | oh. I missed that. | 14:34.02 |
| | Have you tested what happens in acrobat if you give it argc > 6 ? | 14:34.24 |
| paulgardiner | No. That's a good point. I will, but I can fix that with a later commit if needed | 14:36.09 |
| Robin_Watts | ok. | 14:36.18 |
| | I'm happy. | 14:36.23 |
| paulgardiner | Great. I'll push the alterations in a min | 14:36.38 |
| Robin_Watts | Unless you fancy fixing the > 512 stuff to use snprintf? | 14:36.44 |
| | or are you doing that already ? | 14:36.50 |
| paulgardiner | I wasn't planning to just at the moment. | 14:37.37 |
| Robin_Watts | fair enough. | 14:38.47 |
| paulgardiner | I've used that default: return NULL suggestion in app_launchURL. Yeah, nice. | 14:44.01 |
| Robin_Watts | ? | 14:44.46 |
| | I'll see when you push the new versions. | 14:44.54 |
| kens | Robin_Watts : ping | 14:48.34 |
| Robin_Watts | pong | 14:48.38 |
| kens | What would you expect to happen if pdfwrite produced more pages than before for some files (cluster that is) ? | 14:48.55 |
| Robin_Watts | The md5 sum would change, so it would be reported as a diff. | 14:49.27 |
| | bmpcmp would diff the first n pages and then (I think) give an error. | 14:49.40 |
| kens | Right, I'm seeing an error | 14:49.49 |
| | Failed to oload (reference) image 2 from ... | 14:50.03 |
| | But only on 4 files when I expected rather more. | 14:50.14 |
| paulgardiner | Robin_Watts: pushed | 14:50.15 |
| Robin_Watts | Right. So that's read the first image OK (or maybe images 0 and 1, who knows? :) ) and then failed to load image 2. | 14:50.35 |
| henrys | forms meeting in 10 minutes | 14:51.07 |
| kens | That's what I thought. I'm puzzled as to why the files I expect to show a difference (QL tests) don't. | 14:51.11 |
| Robin_Watts | paulgardiner: So is WARN_LIMIT supposed to be the maximum number of chars it's safe to put into an fz_warn ? | 14:51.49 |
| | kens: No idea, sorry. | 14:52.18 |
| kens | Not to worry, maybe I'll ask marcos to try it at the meeting. | 14:52.40 |
| Robin_Watts | oh, a pdf_warn. | 14:54.19 |
| | hmm. pdf_warn can take 1024 bytes, and you've done WARN_LIMIT as 512 to cope with the fixed parts of the string you use, presumably. | 14:55.07 |
| | I understand why, and it's safe, but it feels wrong to have this global constant published that corresponds to how you use it. | 14:55.38 |
| | Tea... | 14:56.00 |
| paulgardiner | Robin_Watts: Yes, I see what you mean. Maybe should just swap to snprintf. But I thought it was you who pointed out a problem with it. Not C99 or something? | 14:58.21 |
| tor8 | Robin_Watts: are we ready to push the submodule stuff? | 14:58.46 |
| Robin_Watts | tor8: I haven't looked at the cluster implications yet. | 15:02.19 |
| | I may get to it tomorrow. | 15:02.43 |
| | sorry. | 15:02.46 |
| | snprintf isn't C99, but I think we rely on it already. | 15:03.00 |
| | Sorry, it's C99, but not C89, I think. | 15:03.14 |
| henrys | So I'm still a little fuzzy on what gets released mid october, since the viewer will not be ready it seems, how is this stuff demo'd? | 15:03.26 |
| paulgardiner | Ah right. | 15:03.27 |
| tor8 | we use snprintf elsewhere, and have the right voodoo to make it work in windows | 15:03.35 |
| Robin_Watts | henrys: The existing windows viewer will work. | 15:04.08 |
| henrys | any hope of getting in a device that Scott or Miles can carry around? | 15:05.02 |
| Robin_Watts | like a windows laptop? :) | 15:05.13 |
| | We haven't tried an android build with js, have we? | 15:05.43 |
| paulgardiner | Not yet | 15:05.54 |
| henrys | I guess I was more thinking of a samsung tablet or something like that. | 15:05.55 |
| mvrhel | it would be nice to have something for the December show | 15:06.20 |
| paulgardiner | Although I did see that article saying that v8 is easy to link with in android | 15:06.31 |
| Robin_Watts | paulgardiner: How hard would it be to make the android app have the text input stuff ? | 15:06.46 |
| paulgardiner | Not hard if we don't mind sticking with the native widgets. | 15:07.32 |
| Robin_Watts | I think native widgets are perfectly acceptable due to the onscreen keyboard etc. | 15:08.04 |
| paulgardiner | The difficulties would probably be elsewhere: we would probably need the separate rendering of the widgets, and there's handling the fact that everything has to be done on a thread other than the UI one. | 15:08.35 |
| Robin_Watts | to avoid lagginess ? | 15:09.09 |
| henrys | Also I'll be in chicago showing off our wares from october 6 - 10 even if you had something really preliminary I'd go pick up a tablet and install it. But maybe that is too soon. | 15:09.48 |
| paulgardiner | yes. the whole thing goes dead if you render on the UI thread | 15:10.04 |
| Robin_Watts | paulgardiner: What else have you got left to do before the release? | 15:10.28 |
| | (is it feasible for you to change your direction towards making the android app work?) | 15:10.49 |
| paulgardiner | Quite possibly. The remaining tasks have been picked fairly arbitrarily . I don't think we really need any of them. I have just started on submission, which would be a very nice tick box. | 15:12.23 |
| henrys | I think it would be a good direction. We do see lots of folks in the mupdf community doing android stuff with mupdf I think if you had something running we'd possibly get some feedback/testing. | 15:14.09 |
| Robin_Watts | paulgardiner: Hey we have Apps World next wednesday. Can you have it done by then? :) | 15:14.37 |
| | </Majid> | 15:14.50 |
| paulgardiner | :-) | 15:14.55 |
| | I'm certainly happy to see what we can do on android | 15:15.43 |
| henrys | the Chicago show is not important - that aside I wonder if going for an android app is the better course now since we can get something out that has a motivated development community. | 15:17.13 |
| paulgardiner | Makes perfect sense to me. | 15:17.33 |
| Robin_Watts | And the android app is a more shiny beast than the existing windows viewer too. | 15:17.44 |
| | And it's nearer our target market, probably. | 15:18.02 |
| henrys | now I have a perfectly good excuse to buy a samsung tablet. | 15:18.23 |
| paulgardiner | The only thing it would delay is support for form submission | 15:18.30 |
| Robin_Watts | henrys: Asus Transformer Infinity. | 15:18.35 |
| henrys | paulgardiner:do yo have a schedule estimate for submission? | 15:19.28 |
| tor8 | henrys: make sure you get a model you can root and install jelly bean (or whatever is the newest version) on | 15:19.40 |
| | and yes, I agree having the demo on android would be ideal | 15:20.09 |
| Robin_Watts | Transformer Infinity has jelly Bean being rolled out now. | 15:20.34 |
| | oh, actually, that was a month ago. so it should be on the device now. | 15:21.05 |
| paulgardiner | I might have something by next meeting, but I can't be completely sure. If we were hoping to have something on android by the 6th, it would be handy to have both the next two weeks | 15:21.23 |
| | The first 'something' being something on submission | 15:22.06 |
| henrys | paulgardiner:okay so we'll make submission and android the goal if you miss chicago no big deal. | 15:22.50 |
| | what else from the list needs discussion. | 15:23.02 |
| | ? | 15:23.03 |
| paulgardiner | henrys: Is that submission first? | 15:23.23 |
| Robin_Watts | henrys: It strikes me that we should do android first, and submission second. | 15:23.31 |
| henrys | paulgardiner:it seems I could show something in chicago without submission right? | 15:23.54 |
| Robin_Watts | cos an android demo at the show with no submission is more useful than .... what you said. | 15:24.05 |
| paulgardiner | henrys: yes definitely | 15:24.13 |
| henrys | so android first I guess. | 15:24.30 |
| paulgardiner | Ok great | 15:24.36 |
| Robin_Watts | I have nothing else to ask about the list of tasks. All looks good to me. | 15:25.22 |
| henrys | are we set for this week or did you want to discuss more on the list, I see you got printing. | 15:25.43 |
| | s/list/list? | 15:25.51 |
| | I almost forgot to ask tor8 about the viewer? His favorite question. | 15:26.33 |
| paulgardiner | Nothing more I can think of. Printing is there in the library, but the app responds just by printing a message | 15:26.36 |
| henrys | right. | 15:26.45 |
| | by next week I should have some device, Robin_Watts why do you recommend Asus? | 15:28.15 |
| Robin_Watts | henrys: It's by far the best specced tablet I've seen. | 15:28.52 |
| | You saw my one with the removable keyboard, right? The Infinity is the next one up with a full HD screen. | 15:29.22 |
| | It has SD card and USB slots in the keyboard (as well as microSD in the tablet bit) | 15:29.48 |
| | It's got the quad (well 5) core tegra stuff. | 15:30.07 |
| | and 18 hour battery life. | 15:30.15 |
| | (something like 6 hours with just the tablet and another 12 with the keyboard or something). | 15:30.44 |
| | plus a VERY nice IPS screen. | 15:30.49 |
| | Stuff did a review of Android tablets last month, let me dig it up. | 15:31.16 |
| | If you want a 7 inch tab, then the Google Nexus is 'astonishing value' apparently. | 15:32.04 |
| | ok, they looked at: Samsung Galaxy Tab 2 (10.1), Acer Iconia Tab A700, Toshiba AT300-101 Asus Transformer Pad Infinity, Archois 101 XS Gen 10 | 15:33.14 |
| | Asus was 5/5, Archos 3/5, others 4/5 | 15:33.42 |
| | They rated them in order: Asus, Tosh, Acer, Samsung, Archos | 15:34.13 |
| tor8 | henrys: lalala *sticks fingers in ears* | 15:34.32 |
| Robin_Watts | (sorry, 14 hours battery) | 15:35.08 |
| henrys | tor8:does sebras have any interest in helping I'll pay him to stick a fork in your side when you stop typing. | 15:35.27 |
| paulgardiner | :-) | 15:35.51 |
| henrys | Robin_Watts:what the url? | 15:36.16 |
| Robin_Watts | henrys: paper magazine :) | 15:36.40 |
| henrys | by the gods | 15:36.51 |
| tor8 | henrys: you can always pay him to stick a fork in my side. I suspect he'll happily do that even without payment :/ | 15:37.00 |
| Robin_Watts | http://www.stuff.tv/review/asus-transformer-pad-infinity | 15:37.25 |
| henrys | tor8:seriously if you want to give a piece fo the project to sebras that is fine. | 15:37.34 |
| tor8 | I can always ask, but I suspect he hates gui-programming even more than me | 15:38.33 |
| henrys | I was hoping they'd be caught up on the high res - looking forward to that. marcosw did you get a laptop with retinal yet? | 15:41.11 |
| marcosw | henrys: yup | 15:41.20 |
| Robin_Watts | henrys: 1920x1200 on a 10 inch tablet is as near to retinal as you really need, right? | 15:41.49 |
| | henrys: marcosw had it at the meeting. | 15:42.00 |
| | You can't actually use all those pixels as it's unreadable. (and they cheat, cos they aren't real pixels, right?) | 15:42.32 |
| henrys | marcosw:oh I wanted to ask you about X11 and gs and mupdf do they take advantage of the new display. I ssume XQuartz had to be updated | 15:44.07 |
| | Robin_Watts:I don't know, it is noted in the articles last paragraph so I assume the difference is visible. | 15:46.12 |
| Robin_Watts | henrys: The asus has the best screen on any android tablet, I believe. | 15:46.37 |
| | (I've not seen any other 1920x1200 ones, or higher, and most other screens aren't IPS anyway) | 15:47.07 |
| | Oh, the Acer Iconia A700 is 1920x1200 too. | 15:48.03 |
| henrys | the os is ice cream sandwich which I assume jelly bean - 1 | 15:49.01 |
| Robin_Watts | but the Acer is non-IPS. | 15:49.12 |
| marcosw | henrys: There is a new quartz, but that happened because of 10.8, so may not be related to the retina display. I don't think mupdf or ghostscript make use of it, but am not sure how to test to confirm. | 15:49.12 |
| Robin_Watts | http://reviews.cnet.com/8301-19736_7-57406607-251/tablets-with-the-best-screens/ | 15:49.19 |
| | At launch the OS was ICS (which is indeed 4.0). | 15:49.29 |
| | Supposedly 4.1 was being rolled out on August 22nd, so I'd hope it should be there. | 15:49.55 |
| | When I got mine, it was shipped with something < 4.0, and 4.0 downloaded and installed when I turned it on. | 15:50.19 |
| | (mine being the Prime, not the Infinity) | 15:50.31 |
| henrys | marcosw:it would be interesting to see the default resolution of the X11 display by looking at the HWResolution field in the device structure after it is set up but no big deal. | 15:50.51 |
| Robin_Watts | http://reviews.cnet.com/8301-19736_7-57406607-251/tablets-with-the-best-screens/ | 15:51.34 |
| marcosw | one of the cool things about the retina based macbook pro is that it supports 3 external monitors, two via display port and a third via hdmi. The display ports run up to 2560×1600 and the hdmi is 1920x1080. including the 2880 x 1800 from the built in display that's a lot of pixels :-) | 15:52.09 |
| Robin_Watts | marcosw: Those external monitors probably hurt the battery life and portability though, right? :) | 15:53.35 |
| tor8 | Robin_Watts: are there any 4:3 aspect ratio android tablets? | 15:53.47 |
| | or are they all widescreen? | 15:53.54 |
| henrys | Robin_Watts:okay based on that article now I want an ipad how long to do an ipad port? :-) | 15:54.07 |
| marcosw | Robin_Watts: a bit. | 15:54.08 |
| Robin_Watts | tor8: Not ones you'd want to buy, I suspect :) | 15:54.14 |
| | but I have some gaffer tape here you could use to mask a 16:9 one :) | 15:54.33 |
| tor8 | this widescreen madness has got to stop... | 15:54.44 |
| henrys | so kens has is your new system which can't be named? | 15:55.01 |
| | s/has/how | 15:55.08 |
| kens | henrys works well | 15:55.14 |
| | tested customer files, and all is fine for me | 15:55.21 |
| | I updated the bug to say so | 15:55.31 |
| henrys | I bet they aren't getting a good build. | 15:56.31 |
| kens | I suspect they messed up applying the patch | 15:57.18 |
| henrys | 3 minutes until the meeting | 15:57.27 |
| Robin_Watts | henrys: The ipad has the best screen, yes, but it loses out on the fact it runs iOS and has crap connectivity. No SD card or microSD or USB slots. Seriously? What the hell are apple thinking. | 15:57.56 |
| henrys | history shows apple doesn't need much to win - and that screen is probably enough to make it outsell all the others. | 15:59.58 |
| Robin_Watts | I have to say, I used my dads ipad 3 just after it came out, and I really couldn't tell whether it was an ipad 3 or an ipad 2. | 16:00.29 |
| | So it's not the "wow" moment on the ipad that it might be on a laptop. | 16:01.01 |
| mvrhel | good morning | 16:01.21 |
| henrys | interesting it was a wow for me on the laptop but I haven't seen an ipad. | 16:01.27 |
| Robin_Watts | Marcosw's laptop screen did look very nice when he kicked it into 1:1 mode. | 16:01.32 |
| henrys | hi mvrhel | 16:01.34 |
| | I texted ray | 16:01.51 |
| Robin_Watts | henrys: Do you have a dealer for such beasts near you? | 16:02.02 |
| | In this internet age, I often forget about the option of actually going to look at one in a shop. | 16:02.32 |
| henrys | I have a best buy here about a mile away but I don't know if they'll have the latest product. | 16:02.41 |
| | I didn't see a micro center or best buy in your village. | 16:03.08 |
| mvrhel | with a thatched roof | 16:03.45 |
| henrys | ;-) | 16:04.02 |
| Robin_Watts | henrys: I'd have to go to Babury (8 miles) or oxford (25 miles) to find such witchcraft. | 16:04.03 |
| | We barely have electricity as you may remember :) | 16:04.42 |
| henrys | mvrhel:so there was a softmask issue alexcher_ needed to work on which bug was that? | 16:05.14 |
| | I've been buried in the xps driver and have forgotten everything. | 16:05.40 |
| mvrhel | 693115 I think | 16:05.58 |
| alexcher_ | Yes, it's 693115. I'm still working on the problem. | 16:06.36 |
| henrys | alexcher_:okay so you have that and the luratech memory stuff on your plate right? | 16:07.22 |
| | alexcher_:please don't hesitate to contact luratech we are paying them now use their help. | 16:07.53 |
| alexcher_ | henrys: Yes, luratech is quite easy. Is gs_malloc the right function to use? | 16:08.13 |
| Robin_Watts | gs_alloc_bytes ? | 16:08.28 |
| henrys | I assume you can just use bytes and avoid the struct stuff. | 16:08.48 |
| Robin_Watts | (and gs_free_object, I think) | 16:08.48 |
| henrys | I think kens had something for marcosw? | 16:10.01 |
| alexcher_ | OK | 16:10.31 |
| henrys | and we wanted to discuss the unicode patch but ray is not here. | 16:10.39 |
| | Robin_Watts:did you post the patch somewhere? | 16:11.27 |
| Robin_Watts | Yes... | 16:11.34 |
| kens | henrys I'm not sure why my chnage is not triggering some diffs | 16:11.34 |
| Robin_Watts | http://git.ghostscript.com/?p=user/robin/ghostpdl.git/.git;a=commitdiff;h=c9e713f9e36765261b3ac2ed0a86ebd0b0259eb5 | 16:11.53 |
| henrys | that's quite a log message! | 16:12.49 |
| mvrhel | I had one thing for marcosw | 16:12.49 |
| Robin_Watts | henrys: I have no memory, so if I don't explain everything on the commit, I'll never remember what I've done. | 16:14.06 |
| mvrhel | I was wondering if it would be possible to set up a weekly test with all the color options. But more of a targeted set of command line and file combinations | 16:14.07 |
| | I could provide all the command line options and files to him, but I don't know what needs to be done to set it up | 16:14.49 |
| | I can email him. Not sure if he is here this morning | 16:15.08 |
| marcosw | mvrhel: I'm here. | 16:15.16 |
| mvrhel | ok. So could I just provide you with the files and the command lines? | 16:15.52 |
| paulgardiner | Robin_Watts: pushed again | 16:16.05 |
| marcosw | if you send me the list of files and command line options I can figure out what the best way to set it up. Do you want to be able to view the output or just get emails when things change | 16:16.07 |
| henrys | chrisl you'll review Robin_Watts as well, I think you are pretty well versed in this business. | 16:16.24 |
| mvrhel | marcosw: If it is easy to see diffs that would be nice. If it takes a lot of work, then just the notification is fine | 16:16.48 |
| Robin_Watts | I'm sure ray will have opinions too. | 16:16.49 |
| chrisl | henrys: I think I've forgotten a lot of it, but I will look it over, sure | 16:16.56 |
| Robin_Watts | and if Sags reads the logs, it would be good for him to look too, as it was his objections that drove this. | 16:17.22 |
| mvrhel | There are so many command line options (and more coming shortly with the blackpoint stuff) that I need to catch any issues | 16:17.28 |
| marcosw | mvrhel: it's not that big of a deal, I can push them to my web server at home (or to casper). | 16:17.57 |
| henrys | yeah sags was just here - maybe you can add this too the old bug and he'd get cc'd. | 16:18.06 |
| mvrhel | marcosw: ok sounds good. I will try to get you a list this week | 16:18.29 |
| henrys | Robin_Watts:I still don't see how environment variables are postscript interpreter files are handled in this patch. | 16:19.25 |
| Robin_Watts | env vars - no change. What do you see being the problem ? | 16:20.09 |
| | You think that they should be interpreted as utf8 too ? | 16:20.41 |
| | And what do you mean by "postscript interpreter files" ? You mean 'postscript' ? | 16:21.09 |
| henrys | I mean doing open in the postscript language. | 16:21.44 |
| | I would think everything would be consistent but maybe that is wrong. chrisl, kens what does the spec say about the ps file system? | 16:22.33 |
| Robin_Watts | OK. Any strings generated in the 'core' of postscript will ALWAYS be interpreted as utf8 when they get passed out to the gp_fopen function. | 16:22.35 |
| kens | henrys it doesn't really. | 16:22.46 |
| henrys | kens:probably not something we're likely to trip over then? | 16:23.17 |
| alexcher_ | PS file system is flat. Files ate identified bt strings. | 16:23.20 |
| | PS file system is flat. Files are identified bt strings. | 16:23.28 |
| Robin_Watts | Unless you build with WINDOWS_NO_UNICODE, in which case you get historical behaviour of using windows codepage. | 16:23.30 |
| kens | henrys, file access in PostScirp tisn't common | 16:23.38 |
| Robin_Watts | and file access with top bit set chars is even rarer, right? | 16:23.57 |
| kens | much rarer | 16:24.33 |
| | Hens teeth really | 16:24.40 |
| | I would not be concerned by that | 16:24.52 |
| chrisl | Also, I don't think the behaviour is defined using something other than 7 bit ASCII | 16:25.09 |
| henrys | chrisl:that's what I'd think but I wasn't sure. | 16:25.37 |
| Robin_Watts | I could probably make it so that environment vars are subject to the same 'encoding decoding' that command line args are. | 16:25.40 |
| henrys | Robin_Watts:environment vars it seems should be the same. particularly for font and library paths. | 16:26.30 |
| | but we could wait for a bug report. | 16:26.45 |
| | so almost 10:30 don't want to hold you up kens, anything else? | 16:27.55 |
| | I wonder if ray is at the customer, I saw that performance email yesterday. | 16:28.21 |
| kens | I want to change hte behaviour of pdfwrite with partial pages | 16:28.24 |
| | At the moment, if we get marks on a page, but no showpage, we drop it. Acrobat Distiller does not | 16:29.02 |
| | I would like to chnage to match Distiller | 16:29.09 |
| | What does everyone else think ? | 16:29.20 |
| | No dissent ? | 16:30.05 |
| alexcher_ | GS already detects EPS and adds showpage. | 16:30.27 |
| chrisl | Well, I'd prefer to require the showpage, as that's what PS says should happen *but* if distiller works differently, then..... | 16:30.28 |
| kens | alexcher, no EPS | 16:30.35 |
| Robin_Watts | I think we should match distiller. | 16:30.36 |
| mvrhel | I agree about matching | 16:30.52 |
| kens | OK so we are aggreed then. Its mainly the quality logic files where this happens | 16:32.31 |
| | I'll commit a change for this tomorrow | 16:33.01 |
| | It makes linearisation easier | 16:33.07 |
| marcosw | henrys: I have to run. I don't think we need a support meeting this week but if you have quesitons/comments I'll be back on line later today. | 16:34.47 |
| kens | Ned to put dinner on, be back in a bit | 16:34.56 |
| alexcher_ | I've tested 64-bit integers. The time difference is small. The problem is that negative long long can exceed the string size in cvrs operator. Numbers like 16#ffffffff go positive. | 16:36.02 |
| henrys | marcosw:okay | 16:36.13 |
| chrisl | alexcher_: I'm most of the way through making changes to teh interpreter to make that work for the CET | 16:36.51 |
| alexcher_ | There's somewhat similar problem. We have many compiler warnings that the format string doesn't match the arguments. Can we use C99 macros ? | 16:40.36 |
| chrisl | alexcher_: do you know how widely supported the "%jd" format string is for printf and co? | 16:41.43 |
| alexcher_ | chrisl: GCC and and recent MSVC support it. I'd rather use PRIxXXX macros instead. In the worst case we can add them to a portability header. | 16:44.05 |
| chrisl | alexcher_: I couldn't seem to get the PRIxXXX macros to work here on Linux, they didn't seem to be defined - but if I made a mistake, I'd prefer them, too | 16:45.24 |
| alexcher_ | chrisl: See /usr/include/inttypes.h | 16:46.59 |
| chrisl | alexcher_: aha, thanks! | 16:47.53 |
| | Even more confusing I couldn't get them to work :-( | 16:48.18 |
| | I'll look into that tomorrow or Thursday | 16:48.43 |
| alexcher_ | gs doesn't include inttypes.h on most systems. | 16:49.12 |
| chrisl | Sure, also it has different names on some systems.....<sigh> | 16:50.13 |
| Robin_Watts | mvrhel: ping | 17:06.03 |
| Bhuvnesh | Hello folks ! I am really sorry that I bumped in to the chat on a pretty odd time last night. I didn't know that most of you are located in Europe. Brief intro about me : I am a developer working on a e-reader app based in New Delhi, pretty much a noob with C and more in to the Java world. Therefore I'd need little help of you guys with some of the things... I am online right now for next 4 hours, I'd highly appreciate any he | 17:07.52 |
| Robin_Watts | Bhuvnesh: OK. So you're using mupdf. For android? | 17:08.36 |
| mvrhel | Robin_Watts: sorry missed your ping | 17:08.40 |
| Robin_Watts | mvrhel: A quick sanity check if I may... | 17:08.54 |
| Bhuvnesh | yeah... | 17:08.57 |
| mvrhel | working on an example to show black point control for text, graphic, image etc | 17:08.59 |
| | to many options | 17:09.21 |
| Robin_Watts | Before I fixed it so that gscms_create/destroy were called, we were using malloc/free rather than the gs functions for everything allocated in lcms2. | 17:09.39 |
| | when I fixed it, I started having gc problems with the transforms disappearing. | 17:10.01 |
| Bhuvnesh | I am using mupdf and looked in to the sample Android app code..but could not find a method to fetch text on a page, current page number and things like locating the word given rectF coordinates | 17:10.06 |
| Robin_Watts | Bhuvnesh: Give me 5 mins, sorry. | 17:10.20 |
| mvrhel | Robin_Watts: uhoh | 17:10.38 |
| | this is not how we like to start out the day | 17:10.56 |
| Robin_Watts | So I patched around that by passing cache_mem->non_gc_memory into lcms instead of cache_mem | 17:11.01 |
| Bhuvnesh | Watts: I checked with mupdf.c and realized I need to write the JNI connector for the facilities which I need..Yeah Sure Mr Watts I can wait.. | 17:11.16 |
| Robin_Watts | mvrhel: in the gscms_get_link call at line 801 of gsicc_cache.c | 17:11.59 |
| | I missed doing that in the gscms_get_link_proof_devlink call just above it though, which I think is the cause of the customer bug. | 17:12.36 |
| | If I fix that the problem seems to go away. | 17:12.43 |
| mvrhel | oh good | 17:12.53 |
| Robin_Watts | Do you have any objections to that fix. | 17:12.55 |
| | Does that sound wrong in any way? | 17:13.01 |
| mvrhel | yes we def. need to use the same memory type which is all non_gc for color stuff | 17:13.11 |
| Robin_Watts | Fab. So I'll commit that fix. | 17:13.21 |
| mvrhel | if you allocate with one and free with the other, boom | 17:13.28 |
| henrys | I am completely baffled how Guillaume is gettiing the correct result with hsi scheme, it shouldn't work at all. | 17:13.32 |
| | s/hsi/his | 17:13.40 |
| Robin_Watts | mvrhel: No, the problem was not a mismatched freeing. | 17:13.42 |
| mvrhel | ok then I don't understand what you are telline me | 17:14.00 |
| | telling | 17:14.02 |
| Robin_Watts | (Though that may be a problem). | 17:14.03 |
| | The problem, I think, is that the profile was being allocated in gc-able memory. | 17:14.32 |
| | and wasn't being marked during the gc phase, so was being binned. | 17:14.49 |
| | and then we tried to use it, and boom. | 17:14.55 |
| mvrhel | ok. no profile should be in gc memory | 17:15.12 |
| Robin_Watts | Hence I moved it to non-gc memory where it was no longer whipped away. | 17:15.14 |
| | OK. | 17:15.17 |
| mvrhel | nor any link | 17:15.23 |
| Robin_Watts | gs_memory_t *cache_mem = pis->icc_link_cache->memory; | 17:15.39 |
| | Should I make that pis->icc_link_cache->memory->non_gc_memory; then ? | 17:16.03 |
| mvrhel | the link cache itself is a gc'd pointer | 17:16.22 |
| | I think | 17:16.33 |
| | I would need to look that over | 17:16.50 |
| | hold on | 17:17.04 |
| Robin_Watts | From what you're saying, it sounds like cache_mem should really be a non-gc'd memory space itself. | 17:17.44 |
| mvrhel | oh yes. it is in stable memory | 17:17.50 |
| | no | 17:17.55 |
| | oh sorry | 17:17.58 |
| Robin_Watts | stable != non_gc though ? | 17:18.04 |
| mvrhel | hold on one sec | 17:18.15 |
| | I need to see how the link cache is freed | 17:18.24 |
| | and what cache_mem is used for | 17:18.35 |
| Robin_Watts | gsicc_remove_link ? | 17:18.35 |
| mvrhel | hold on | 17:18.40 |
| | let me read the code | 17:18.44 |
| Robin_Watts | shuts up :) | 17:18.56 |
| mvrhel | I see two things that need to change | 17:22.25 |
| Robin_Watts | go on. | 17:24.16 |
| mvrhel | hold on phone now.... | 17:24.39 |
| Robin_Watts | Bhuvnesh: OK, while mvrhel is on the phone... | 17:25.28 |
| Bhuvnesh | Watts: yeah... so could just guide me a little what I need to do and where I need to look to get those things done... | 17:26.04 |
| Robin_Watts | Before we go any further I should at least mention licensing. You're aware that MuPDF is dual licensed, right? It's available both under the GNU GPL license, and from Artifex commercially. | 17:26.07 |
| | If you want to use the GPL version, then you would need to release all your source code under the GPL. | 17:26.39 |
| Bhuvnesh | Watts: Yeah I am aware of that and shall be abiding by the GNU GPL license... | 17:26.45 |
| Robin_Watts | Fair enough. | 17:27.01 |
| Bhuvnesh | so I am looking to build a cross platform module for Titanium | 17:27.31 |
| | and one I build that successfully I'd put it on GNU GPL | 17:27.47 |
| Robin_Watts | Titanium ? | 17:28.04 |
| Bhuvnesh | Appacelrator Titanium...a JS based cross platform library for developing apps on android, iOS ... | 17:28.32 |
| mvrhel | Robin_Watts: so we need to fix gsicc_remove_link to use non_gc memory instead of stable memory. Also, gsicc_alloc_link_entry needs cache_mem (probably should be renamed) is non_gc and also line 597 on the allocation should use non-gc memory | 17:28.47 |
| | gsicc_alloc_link also needs some changes | 17:29.17 |
| | not stable memory but non-gc | 17:29.30 |
| Robin_Watts | mvrhel: can we change gsicc_cache_new to do: result->memory = memory->non_gc_memory ? | 17:29.50 |
| mvrhel | no | 17:30.01 |
| | keep that in stable memory | 17:30.05 |
| Robin_Watts | OK. | 17:30.10 |
| mvrhel | it is a gc'd pointer in the imager state | 17:30.14 |
| Robin_Watts | eh? | 17:30.26 |
| | result is a gc'd pointer in the image state, yes. | 17:30.39 |
| mvrhel | yes | 17:30.43 |
| Robin_Watts | but result->memory isn't | 17:30.48 |
| mvrhel | no but we use that pointer when we free it | 17:31.06 |
| Robin_Watts | Do we? | 17:31.14 |
| mvrhel | we should | 17:31.17 |
| | why else carry it aroudn | 17:31.23 |
| | that is why you add a memory ptr to a structure | 17:31.32 |
| | to make sure you allocate and free with the same memory | 17:31.41 |
| Robin_Watts | Because that's the memory pointer we use for allocating all the links in. | 17:31.47 |
| | That's the memory pointer that is read as 'cache_mem' at various points. | 17:32.03 |
| mvrhel | not after you change it to non-gc for the allcoation of the links | 17:32.04 |
| | that is why I said change the name | 17:32.17 |
| | it is not going to be the cache_mem | 17:32.23 |
| Robin_Watts | The freeing looks to be done in rc_gsicc_link_cache_free | 17:32.25 |
| mvrhel | yes | 17:32.35 |
| Robin_Watts | Which gets a memory pointer passed to it. | 17:32.36 |
| mvrhel | it is probably from the freeing of the imager state | 17:32.54 |
| | which would be the same one | 17:32.58 |
| | that part is all ok | 17:33.26 |
| Robin_Watts | Right, so the sole use of the 'memory' pointer within the structure, is (as far as I can tell) for allocing/freeing the links. | 17:33.26 |
| | Hence if we changed that to be the non gc one, we'd be a long way towards getting what we wanted, wouldn't we? | 17:34.02 |
| mvrhel | I dont want to have a mem pointer in a structure that is different than the pointer that was used for allocating that structure please | 17:34.04 |
| | please. that would be confusing to me | 17:34.15 |
| Robin_Watts | OK. Even if I renamed it? | 17:34.24 |
| mvrhel | I guess if it was named appropriately it would be ok. | 17:34.48 |
| Robin_Watts | OK. Let me add a second memory pointer to the structure, called... link_memory ? | 17:35.31 |
| mvrhel | that would be fine | 17:35.40 |
| Robin_Watts | Ok, so I'll have a bash at this, and put a patch up somewhere for you to cast your eye over after I've tested it. OK? | 17:36.12 |
| mvrhel | sounds good | 17:36.26 |
| Robin_Watts | fab. I'll let you get back to your test file. | 17:36.40 |
| | thanks. | 17:36.43 |
| | Bhuvnesh: OK, back. | 17:37.13 |
| | So, you'll have seen that mupdf is a set of libraries written in C. | 17:37.27 |
| henrys | if somebody sees something in Guillaumes words that I am completely missing let me know. | 17:38.13 |
| Robin_Watts | The android app is basically these libraries with a thin layer on top of them, consisting of just enough C to expose the methods we need to the java, and all the app side stuff is done in the java. | 17:38.18 |
| | henrys: Short version of Guillaumes text: "I'm doing this horrible hack that seemed to work for a while, and now it doesn't. Why?" | 17:38.51 |
| Bhuvnesh | yeah... | 17:39.05 |
| Robin_Watts | So, the C core of mupdf is certainly capable of doing all you want, but if we haven't needed those facilities in our app, the required C->java interfaces won't be there. | 17:39.44 |
| | So, let's run through what you wanted again... | 17:39.59 |
| | You said something about wanting to extract the number of the pdf page from the page? I'm not sure I follow that. | 17:40.17 |
| Bhuvnesh | so lets say I need a method in that thin layer that return me an array of all the words present on the current page | 17:40.22 |
| | ["this","is","part","of","current","page"] | 17:40.55 |
| | how do I go about it ... | 17:41.05 |
| Robin_Watts | OK. The way we'd do that in C is to open a text extraction device, and then run the page through it. | 17:41.28 |
| | That would give you a structure with such details in. | 17:41.45 |
| | You need to write a new C function to do that, and to convert from that structure to a form suitable for passing back to the java (i.e. creating the array from it). | 17:42.22 |
| | Then you call that C function from the java. | 17:42.56 |
| Bhuvnesh | alright I get a bit ... a basic query: What would that text extraction device be ? What C header file I need to look in for that...I know this might be pretty basic but just need little to get off the ground | 17:44.28 |
| Robin_Watts | The text extraction device interface is defined in fitz/fitz.h | 17:45.02 |
| | If you look in apps/mudraw.c you can see it being used there. | 17:45.14 |
| Bhuvnesh | is it fz_device ? | 17:45.51 |
| Robin_Watts | fz_device is the generic device interface. | 17:46.06 |
| | Implementations of fz_device include a draw device, the text extraction device, the trace device etc. | 17:46.26 |
| | fz_new_text_device is the function that makes one. | 17:46.34 |
| Bhuvnesh | alright ...now I get the hang of it ... | 17:46.49 |
| Robin_Watts | So you'd: sheet = fz_new_text_sheet(ctx); | 17:46.58 |
| | page = fz_new_text_page(ctx, box); | 17:47.32 |
| | dev = fz_new_text_device(ctx, sheet, page); | 17:47.47 |
| Bhuvnesh | yeah got that ! | 17:48.08 |
| Robin_Watts | Then you'd run the page through that device: fz_run_page(dev, pdf_page); | 17:48.18 |
| Bhuvnesh | just like what you have done in searchPage() | 17:48.27 |
| Robin_Watts | and that would populate sheet and page for you. | 17:48.29 |
| | indeed. | 17:48.31 |
| Bhuvnesh | but fz_run_page doesn't return anything ...is it fz_new_text_page which gives me the text ? | 17:49.52 |
| Robin_Watts | fz_run_page interprets the pdf page. It calls the device saying things like "draw a line here", "put some text here" etc. | 17:50.39 |
| | The device implements this by storing a record of the text that it was asked to put on the page within the sheet and page structures that it was passed when it was created. | 17:51.17 |
| | By running the page, you populate those structures. Does that make sense ? | 17:51.33 |
| Bhuvnesh | yeah that does ! | 17:52.51 |
| Robin_Watts | fab. | 17:52.56 |
| Bhuvnesh | and is their some global variable in fitz.h which keeps the track of current page in the currently loaded pdf ? just to get the current page number ... | 17:53.41 |
| Robin_Watts | The currently loaded page number is held in the java side. | 17:54.57 |
| | The C side is always told what page to operate on by the java side. | 17:55.26 |
| | Sorry, let me rephrase. | 17:55.49 |
| Bhuvnesh | ok ! my bad, i missed that I guess. | 17:55.52 |
| Robin_Watts | There is no such global in the central C core. | 17:56.05 |
| | There may be in the thin C->java interface. | 17:56.18 |
| | There is a gotoPageInternal call or something like that. | 17:56.34 |
| | Look at either that, or the places that call that. | 17:56.44 |
| Bhuvnesh | yeah I got that now :) | 17:57.00 |
| | one another thing ...lets say I have two points on the screen with their respective X-Y coordinates and the space covered between the 2-D space formed by the start and end point contains some pdf text and I need to extract that test...do I still use the text extraction device ? and if that is the case how do I manage the coordinates based filtering ? | 17:59.14 |
| Robin_Watts | You remember you did: page = fz_new_text_page(ctx, box); ? | 18:00.29 |
| | I think that's what the box is for. | 18:00.44 |
| Bhuvnesh | ohh... | 18:01.19 |
| | I relate it now . | 18:01.42 |
| Robin_Watts | but in any case, the data in the 'page' structure that the text device fills in has coords etc of each char/word/etc | 18:01.45 |
| Bhuvnesh | alright ! so it looks like that at the end of the day a code walk through inside mupdf.c and fitz.h should be good enough to get these things done ... am I right ? | 18:03.36 |
| Robin_Watts | I hope so. | 18:03.47 |
| Bhuvnesh | I too . | 18:04.08 |
| | well that should be good enough lesson for me to try my hands at writing these methods for C-Java interface . Thank you very much for your time and patience . | 18:05.08 |
| Robin_Watts | You're welcome. | 18:05.15 |
| | Let us know how you get on. It's always nice to see what people come up with. | 18:05.41 |
| Bhuvnesh | I hope I should be able to these things now, if needed will ask query or two in future :) | 18:05.54 |
| | sure ! | 18:05.54 |
| | one small thing ... | 18:06.51 |
| | If I need to test these c-java interface methods I guess I don need to go through that long process of ndk-build | 18:07.22 |
| | is there a shorter way of doing that ? | 18:07.48 |
| Robin_Watts | ndk-build should be fast after you have done it once. | 18:08.06 |
| | That is, it only rebuilds the C files if it needs to. | 18:08.27 |
| Bhuvnesh | alright ! Thanks again :) | 18:08.48 |
| Robin_Watts | So if you change only one file and rebuild, you should only need to rebuild that one C file and relink. Something that is as fast as possible. | 18:08.54 |
| | np. | 18:08.59 |
| jen___ | Is there a way to watermark a 10-page PDF document using GS? | 18:25.50 |
| | gs -q -dNOPAUSE -dSAFER -dBATCH -sOutputFile=result1.pdf -sDEVICE=pdfwrite water1.ps original.pdf only marks the 1st page. | 18:26.14 |
| kens | Use a BeginPage or EndPage procedure | 18:26.47 |
| | execute the watermark PostScript as the procedure for the Begin/EndPage | 18:27.02 |
| | If you use BeginPage the watermark will be 'under' the marks from teh input, if you use EndPage it will be over | 18:28.02 |
| jen___ | kens, do you have a link where I can learn more about this procedure? Much appreciated. | 18:28.39 |
| kens | BeginPage and EndPage are part of the standard PostScript Reference, I suspect a search with Google or on StackOverflow will find some likely hits | 18:29.13 |
| | I'm sure I've answered this on Stack Overflow before | 18:29.28 |
| jen___ | kens, reading http://stackoverflow.com/questions/12243044/is-it-possible-in-ghostscript-to-add-watermark-to-every-page-in-pdf | 18:31.46 |
| kens | Yes, but I didn't add any details | 18:32.09 |
| jen___ | I'm not at all with the intricacies of PS... | 18:32.35 |
| | *familiar | 18:33.01 |
| kens | Essentially : | 18:33.03 |
| | << | 18:33.03 |
| | ... | 18:33.03 |
| | insert PS here | 18:33.03 |
| | ... | 18:33.03 |
| | } def | 18:33.04 |
| | >> setpagedevice | 18:33.04 |
| | Damn IRC | 18:33.12 |
| | There should be a "/BeginPage {" after the '<<' | 18:33.34 |
| | jen___ : look at this example: | 18:36.00 |
| | http://www.tek-tips.com/viewthread.cfm?qid=752143 | 18:36.00 |
| jen___ | thanks kens let me look. | 18:37.09 |
| Robin_Watts | mvrhel: I think we can leave the link structure itself in stable memory. | 18:37.27 |
| | It's just the link handle returned from the gscms layer that needs to be non-gc'd. | 18:37.52 |
| | So the important thing is that the gscms calls get the non_gc pointer. | 18:38.08 |
| | And that was exactly the two calls I'd changed. | 18:38.40 |
| | (well, the 1 I'd changed, and the 1 I'd missed that was causing this problem). | 18:38.57 |
| | So I'll just fix that and we're done, unless you can think of anything else that needs fixing. | 18:39.16 |
| kens | OK night all | 18:42.51 |
| jen___ | hello radistao | 18:43.35 |
| | kens just left | 18:43.46 |
| radistao | hi | 18:43.55 |
| ghostbot | hey | 18:43.55 |
| mvrhel | Robin_Watts: ok that sounds great | 19:16.30 |
| Robin_Watts | mvrhel: Thanks, just testing it now. | 19:16.42 |
| henrys | tor8, Robin_Watts: I really don't want bugs without test files cluttering up bugzilla but I don't want to start a fight with zeniko. IMHO things like 693284 should be closed - no test file no bug. Is that going to upset him. | 19:41.15 |
| | ? | 19:42.28 |
| Robin_Watts | lookng | 19:42.33 |
| | So that bug was found by fuzzing? a spin off from 693290 | 19:43.55 |
| | He's give us a patch, which I haven't had time to look into in detail. | 19:45.43 |
| henrys | well I thought there was a test based on the last sentence of comment 0 I don't know much about fuzzing. | 19:46.01 |
| Robin_Watts | Me either. | 19:46.14 |
| | Given that there is a patch awaiting our attention, I'd be tempted to leave it open. | 19:46.43 |
| henrys | fair enough | 19:48.34 |
| Robin_Watts | but yes, we should close bugs that don't have examples or patches on. | 19:48.56 |
| | and we should try to persuade zeniko to give us static patches against our sources rather than autogenerated ones hosted elsewhere. | 19:49.29 |
| | FWIW, Best Buy do carry the ASUS Infinity, supposedly. | 19:53.40 |
| | dinnertime. | 19:53.45 |
| henrys | mvrhel:ping | 19:54.04 |
| mvrhel | henrys: pong | 19:54.15 |
| henrys | on 693055 I think what I want to do is completely disable color management in the pipeline for that devices. | 19:55.12 |
| | presumably the high level output sent to the printer will color managed as appropriate. | 19:55.34 |
| mvrhel | ok. so -dUseFastColor should do the job | 19:55.55 |
| henrys | so then I got to wondering why pdfwrite and ps2write doesn't have the same issues. | 19:56.21 |
| mvrhel | pdfwrite and ps2write dont use any of the icc color management stuff | 19:56.43 |
| | that is the project that I mentioned at our meeting | 19:56.51 |
| henrys | okay I thought you had some of that working. | 19:57.09 |
| | thanks | 19:57.20 |
| mvrhel | no we have it ignoring any of the icc profiles due to the v4 stuff | 19:57.31 |
| henrys | so I guess it would be nice to have a device flag that said don't color manage me. | 19:58.25 |
| mvrhel | henrys: that would be a good idea | 20:05.47 |
| | we really already have that in a way | 20:05.57 |
| | as -dUseFastColor is a device parameter | 20:06.10 |
| | the device is free to set it | 20:06.22 |
| | have to head out for a bit | 20:09.48 |
| Robin_Watts | marcosw: hey | 23:08.18 |
| | tor8 has a commit ready to go that will unleash git submodules on the mupdf source. | 23:08.40 |
| | It's absolutely guaranteed to screw the cluster entirely. | 23:08.57 |
| | So I'll have a go at fixing that tomorrow. | 23:09.12 |
| | Will you be around tomorrow evening to bale me out if I cock it all up? :) | 23:09.24 |
| marcosw | Robin_Watts: I'll be around tomorrow, is that evening in your time zone or mine? | 23:13.20 |
| Robin_Watts | my time zone. | 23:13.28 |
| | ken will be happy. customer seem to have got their build working and the problem has indeed gone away. | 23:14.24 |
| | Forward 1 day (to 2012/09/26)>>> | |
IRC Logs