| <<<Back 1 day (to 2012/10/15) | 2012/10/16 |
mvrhel | good grief. I have a very simple file that crashes that I was sure was from my changes but it turns out the trunk also crashes. | 05:07.43 |
| Robin_Watts: I may send you this one to enjoy in the morning. Some weird memory issue | 05:24.20 |
| night all | 06:03.41 |
kens | chrisl someone on comp.lan.postscript seems to expect our configure script to 'install' Ghostscript, I'm pretty sure it doesn't do that, correct ? | 07:22.19 |
chrisl | kens: make installs it - "make install" | 07:30.06 |
| kens: ah, that's not quite what he's saying. We do ignore the "--prefix" parameter at the moment - the only way to change the install target path is to edit the Makefile after configure is complete. | 07:32.41 |
| kens: Oh, actually, that's the ghostpdl configure, the gs one should work, I think..... | 07:36.14 |
kens | Then I'm confused and *way* out of my depth | 07:45.46 |
| Thanks for penning a reply though :-) | 07:46.38 |
chrisl | Well, we shall see what he comes back with. I don't want to test it because I don't want my Linux install messed up! | 07:47.55 |
kens | chrisl I guiess IO could test it my VM currently only has 8.71 installed | 07:56.41 |
| and its not 100% working so it doesn't matter if I break it | 07:56.54 |
chrisl | kens: I'm inclined to wait and see what chappy says on comp.lang.postscript..... | 07:58.56 |
kens | I'll gicve it a quick try | 07:59.17 |
| Actually, no I won't because I don't have teh current sources on there yet | 08:00.07 |
chrisl | I'm fairly sure is works because it's something I think Nelson Beebe uses. If it turns out theres a problem, I'll clone one of my virtual machines to test/fix it | 08:00.55 |
kens | Fair enough, waiting seems the right thing to do now | 08:01.42 |
oriceon | Hello people | 08:47.22 |
| Someone online? | 08:47.26 |
Robin_Watts | a few of us. | 08:47.40 |
oriceon | Nice | 08:47.55 |
| Someone of us works with xpdf and ghostscript? | 08:48.12 |
| Cause i have an issue regarding png transparency in pdf`s and can`t find a way to handle it | 08:48.38 |
Robin_Watts | gs, yes; xpdf, less likely. | 08:48.43 |
| Describe your problem. | 08:49.05 |
oriceon | So, i have an pdf with rgb colors, black | 08:49.35 |
| and need to convert black rgb to cmyk | 08:49.53 |
| and for this i use two things: | 08:50.00 |
| 1: convert pdf to ps with pdftops that is part of xpdf | 08:50.13 |
| 2: call gs with some parameters and some .ps`s that parse my pdf ps and convert it to cmyk | 08:50.44 |
| untill there all`s good, black is now cmyk but somehow i can`t handle some png image that has transparency | 08:51.10 |
| in new pdf, png images will shown with black background color | 08:51.25 |
kens | PostScript does not have any real concept of transparency, so whatever you are using to handle PNG images must be doing something to convert them into PostScript, this conversion is likely to be the source of your problem | 08:54.10 |
| Its not clear where your PNG images are coming fomr, since PDF also does not include PNG | 08:54.40 |
| (though it does handle transparency) | 08:54.51 |
| However, to me it looks likely that pdftops is the source of your problem, you could try using Ghostscript's ps2write device to do the conversion instead. | 08:55.22 |
oriceon | one sec | 08:55.33 |
| i show you the pdf | 08:55.35 |
| http://imobiliare.topestate.ro/ACCESSRealEstate_revista_A5_10.10.2012_cmyk.rar | 08:58.26 |
| after download change .rar to .pdf please | 08:58.39 |
kens | You mean deompress surely ? | 08:58.50 |
oriceon | i changed extension to rar because i have some htaccess rules for pdfs.. | 08:59.12 |
| only rename it to .pdf | 08:59.20 |
kens | Hmm, OK, I would choose an extension wihtout an association myself but fine. | 08:59.38 |
| Is this *after* you've manipualted it ? | 09:00.10 |
oriceon | this pdf is the pdf generated from gs | 09:00.28 |
kens | OK well that's no real good to me | 09:00.36 |
oriceon | and in first page you see that png transparent for testing pruposal | 09:00.40 |
| and has black bg | 09:00.43 |
kens | I can't tell you much by looking at a broken output | 09:00.46 |
oriceon | did you downloaded .rar? | 09:01.01 |
| did you renamed to .pdf ? | 09:01.06 |
kens | I'd suggest that you make a nice *simple* example file (24 pages is not simple) | 09:01.09 |
| and show me the source file, not the output | 09:01.19 |
oriceon | ah, ya :)... | 09:01.21 |
| i showd my real one | 09:01.26 |
| wait | 09:01.34 |
kens | In fact please let me have both the original PDF and the pdftops output, because that way I can get some idea what's going on | 09:01.51 |
chrisl | <sigh> OS/2 ..... really????? | 09:09.49 |
kens | Yeah, I saw that | 09:10.27 |
Robin_Watts | chrisl: I read that report this morning and thought he'd maybe forgotten WHICH_CMS=lcms2 | 09:10.35 |
kens | I don't have a copy of OS/2 or I could set up a VM, but really, how dead does an OS have to be ? | 09:10.53 |
chrisl | Well, unfortunately, it's been resurrected as e-commstation - or something like that..... | 09:11.14 |
kens | Quick, get Buffy to stick a stake though its heart... | 09:12.36 |
chrisl | Gladly - I can't believe anyone wants to use the dratted thing these days :-( | 09:13.15 |
oriceon | kens: sorry but now i`ve got another problem. I`ll came back tomorrow, hope to find you here | 09:13.37 |
kens | I'll be here, but not for a couple of hours in the morning | 09:13.54 |
| between about 10:30 and 12:30 UK time | 09:14.06 |
| chril, they want customers to *buy* eComStation ? :-O | 09:14.54 |
chrisl | Yeh, and it's not cheap, either: about 250usd | 09:15.22 |
kens | GOod grief.... | 09:15.43 |
chrisl | If it was free, or a sensible price, I'd have setup a VM ages ago | 09:16.19 |
kens | Hmm, they do a demo version: | 09:16.37 |
| http://www.ecomstation.co.uk/html/products.html#TryoutCD | 09:16.37 |
chrisl | It only runs from the CD, and I *seriously* doubt it has any dev tools on it! | 09:16.58 |
kens | Ah, but you can't install it, so I don't think that would work well | 09:17.13 |
kens | wonders what on earth you use to compile on it anyway, gcc ? | 09:17.31 |
| Hmm, possibly OpenWatcom, anotgher dead product ;-) | 09:18.24 |
chrisl | Hmm, dunno - OpenWatcom's documentation seems rather contradictory on the subject..... | 09:22.23 |
kens | THeir last release was more than 2 years ago | 09:22.38 |
chrisl | Well, when was Visual Studio's last release? | 09:23.12 |
kens | 2012 is about to come out | 09:24.05 |
Robin_Watts | 2010 came out about 6 months ago? | 09:24.05 |
kens | 2012 ? | 09:24.16 |
| I downloaded both the ISoOs for the express version | 09:24.37 |
Robin_Watts | Hmm maybe. | 09:24.41 |
kens | But haven't installed tehm yet | 09:24.46 |
Robin_Watts | kens is revelling in all his shiny new disc space :) | 09:25.02 |
kens | *speedy* new disk space :-) | 09:25.23 |
| New PC botts from cold in under 1 minute | 09:25.43 |
| boots* | 09:25.48 |
chrisl | VS2008 came out in 2007, VS2010 came out in 2010, and VS2012 came in 2012 - so two years. OpenWatcom *could* just match that if a new version appears soon! :-) | 09:26.35 |
kens | I was a bit puzzled by the 2012 visual studio express downloads, because there's one for 'desktop' and one for 'Windows 8' so I grabbed both | 09:26.38 |
| chrisl yes, true, but it still seems pretty slow to me. | 09:27.00 |
chrisl | There's been several "calls for more developers" so I suspect they're on their way to a slow death...... | 09:27.38 |
kens | And all those visual studio releases had service packs as well in the interim, which I would be inclined to count as 'releases' | 09:27.45 |
| Yes, I saw the calls for more developers | 09:28.04 |
chrisl | Hmm, can't bmpcmp a 32 bit build on the cluster :-( | 11:22.04 |
Robin_Watts | chrisl: Hmm. | 11:26.05 |
| Do you want a bmpcmp32 option or something? | 11:26.18 |
chrisl | Well, currently (it seems) you can do "clusterpush.pl 32" and get a 32 bit test run | 11:26.59 |
Robin_Watts | yes. | 11:27.15 |
chrisl | But "clusterpush.pl 32 bmpcmp" or "clusterpush.pl bmpcmp 32" results in the cluster doing two runs - one for "32" and one for "bmpcmp" | 11:27.45 |
Robin_Watts | I could try to add a bmpcmp32 option. | 11:28.32 |
chrisl | As I said, it's not urgent, I was just surprised it would do one but not the other - as it were..... | 11:28.57 |
Robin_Watts | Well, if you do: clusterpush.pl gs bmpcmp then it runs gs and then bmpcmp's it. That's a design feature so you can queue both at the same time. | 11:29.33 |
| and you can do: clusterpush.pl gs xps pcl bmpcmp etc too. | 11:29.53 |
| So being able to do 32 is perhaps not suprising. | 11:30.00 |
| unlike my spelling. | 11:30.11 |
chrisl | Yes. The comment at the top of clusterpush.pl is a little confusing: "bmpcmp usage: [gs] [pcl] [xps] [gs] [mupdf] [bmpcmp] [lowres] [32] [$user] | abort" | 11:31.11 |
Robin_Watts | There is a small but non-zero possibility that bmpcmp32 will work now. | 11:34.54 |
Robin_Watts | should be more wary about making cluster changes if marcos is taking time off. | 11:39.01 |
chrisl | I'm running my own script locally - there's not many files to check. | 11:43.41 |
Robin_Watts | tor8: ping | 11:54.31 |
tor8 | Robin_Watts hey | 11:54.47 |
Robin_Watts | We were contacted by a customer asking if MuPDF supported (or could support) transitions. | 11:55.09 |
| So I had a bit of a play over the weekend, and now have some simple changes to the existing pdfapp.c stuff to do it. | 11:55.44 |
tor8 | eww. application level animation stuff? I guess that's what prompted your weekend hack :) | 11:55.50 |
Robin_Watts | If you hit 'p' you go into presentation mode, and then whenever you change page, it uses a transition. | 11:56.14 |
| (basically it keeps the old pixmap around, and blends it with a new pixmap to get the one that's displayed). | 11:56.40 |
| gah. 2 mins. | 11:57.27 |
tor8 | section 8.3.3 of the spec? | 11:57.38 |
Robin_Watts | urm... | 11:58.34 |
| yes. | 11:58.47 |
| I've got code for Blinds/Box/Wipe/Fade | 11:59.11 |
| oops, not Box. | 11:59.15 |
| Split/Push/Cover/Uncover are trivial to add. | 11:59.43 |
| But what we have is enough for a demo. | 11:59.58 |
| I'm not even going to think about the 'animation between different OCG states' stuff (customer has said they don't care about that) | 12:00.30 |
tor8 | yeah. I don't disapprove in general, but on platforms that have native support for animations it'd sort of make sense to use that instead. | 12:00.52 |
| like iOS has this whole core animation framework | 12:01.00 |
Robin_Watts | tor8: Right. The code for this is NOT part of the mupdf core. | 12:01.11 |
| It's effectively a helper class that sits alongside it. | 12:01.22 |
| so you are free to reimplement the effects any way you want. | 12:01.36 |
tor8 | yeah. basically a fz_pixmap crossfade generation helper. | 12:01.43 |
Robin_Watts | indeed. | 12:01.48 |
tor8 | along with the fz_transition struct containing info about what to do | 12:02.02 |
| but not hooked up by default in the core | 12:02.09 |
| if I read your patch (and intent) correctly | 12:02.27 |
Robin_Watts | We read the fz_transition struct every time we read a page. | 12:02.36 |
tor8 | I meant the transition struct and fitz pixmap generation not being hooked up | 12:02.59 |
Robin_Watts | and people can recover it using fz_page_transition (though I'm thinking that should be fz_page_presentation). | 12:03.00 |
| right. | 12:03.03 |
| The one bit of presentation mode we're missing is that we should auto advance after a given Duration. | 12:03.19 |
tor8 | yes, I think fz_page_presentation is better, given the duration field | 12:03.28 |
| is there anything like it in the xps spec? | 12:03.35 |
Robin_Watts | pass. | 12:03.42 |
tor8 | okay, I'll look | 12:03.48 |
| I doubt it though | 12:04.03 |
Robin_Watts | So, can you think of a nice, low impact way to do auto advance after n secs? | 12:04.11 |
tor8 | timer objects are tricky to make cross platform | 12:04.36 |
| on x11 you'd just set a custom timeout to the event loop select() call | 12:04.58 |
| and when you reach it you'll trigger the transfer | 12:05.11 |
| on win32 I have no idea | 12:05.15 |
Robin_Watts | in win32 there is SetTimer. | 12:08.02 |
| I may look into that at some point. | 12:09.30 |
| tor8: So you're not offended by anything I've done so far? | 12:09.41 |
Robin_Watts | lunches | 12:10.28 |
tor8 | not any more than I'm offended by the presence of presentation mode in the spec :) | 12:11.11 |
Robin_Watts | tor8, paulgardiner, henrys: 30 mins to mupdf meeting, presumably? | 14:33.33 |
paulgardiner | yup | 14:33.53 |
| Hmmm, where's my bmpcmp? Is something broken? | 14:45.14 |
henrys | paulgardiner:can you forward miles' questions to Robin_Watts and tor8, or I can do it. | 14:46.54 |
| ? | 14:46.58 |
paulgardiner | no problem | 14:47.13 |
Robin_Watts | paulgardiner: No differences found in your last test. | 14:47.16 |
| hence no bmpcmp process :) | 14:47.21 |
paulgardiner | I must have misread it. Thought it said there were 2 | 14:47.37 |
Robin_Watts | I suspect you're misreading the report with the "Differences since previous clusterpush" section. | 14:47.49 |
| I did exactly that last week, and spent an hour or so debugging the cluster :( | 14:48.05 |
henrys | a common mistake for us all. | 14:48.08 |
paulgardiner | Yeah, that would be it. I could have read ll the words I guess! :-) | 14:49.10 |
Robin_Watts | So is paul answering that mail? Or henry? | 14:56.04 |
henrys | chrisl:os/2 seriously? | 14:56.08 |
Robin_Watts | os half. | 14:56.16 |
henrys | Robin_Watts:I think Paul should respond but if he doesn't want to I'm happy to stumble through it. | 14:57.41 |
paulgardiner | Yes, sure. But makes sense to discuss it now, as you suggested. | 14:58.19 |
henrys | paulgardiner:right | 14:58.45 |
Robin_Watts | I suspect in the question of portability, you should be at pains to point out the that the UI code is android specific and hence not portable. | 14:59.37 |
henrys | I think marketing line needs to be we provide an android UI that can be used as a guide for deployment on other platforms such as: ... | 15:01.02 |
paulgardiner | Yes and each platform has its ownvagaries | 15:01.08 |
Robin_Watts | but then obviously say that the code forms changes to MuPDF are portable. As to how portable the javascript engine itself is, you should say we're largely independent of engine, but v8 is fairly portable. | 15:01.19 |
| yeah. | 15:01.25 |
| In theory BBOS can run android stuff now. | 15:01.36 |
henrys | and you can answer 2 with the space expense. But we are looking at shared libs down the road right? | 15:02.36 |
Robin_Watts | Symbian: No one uses Symbian any more. Any ports to that would be silly. | 15:02.44 |
paulgardiner | Yes | 15:02.44 |
| henrys: I don't think we have access to v8 as a shared lib, but a client might. | 15:03.40 |
chrisl | henrys: my reaction exactly (re. OS/2) :-( | 15:03.55 |
paulgardiner | Nokia uses Qt now | 15:04.01 |
Robin_Watts | I think whether v8 is available as a shared *system* lib is down to the configuration or not. | 15:04.15 |
| (of the deveice O mean) | 15:04.22 |
| s/O/I/ | 15:04.29 |
| paulgardiner: Possibly not any more :( | 15:04.36 |
| Nokia have sold out to the Windows Phone boys in desperation. | 15:05.00 |
| Even stuff like Meego is being sidelined AIUI. | 15:05.19 |
paulgardiner | oh ok. I'm a little out of date | 15:05.42 |
Robin_Watts | I'd be surprised if windows phone uses Qt :( | 15:05.46 |
| paulgardiner: Less so than me, in terms of actually using real hardware, I suspect :) | 15:06.10 |
henrys | anyway this will end up being marketing literature let's not get the bar too high on details or truth ;-) | 15:06.22 |
| right now it doesn't use a shared library but we think in the future it will. | 15:06.43 |
paulgardiner | 3) I don't think we are near a beta. I think we'd need the alpha out for a while before we'd have the feedback that would tell us what would be expected in a beta | 15:06.46 |
henrys | or something to that effect. | 15:06.52 |
| I was ready to call this an alpha though, it is mid october, am I in line? | 15:07.30 |
Robin_Watts | I think when Miles says Beta he means "a public showing of code". So what we mean by alpha. | 15:07.39 |
paulgardiner | I think it's ok to call this an alpha. | 15:07.54 |
| Robin_Watts: ah yes, probably | 15:08.10 |
henrys | doing a press release before February's full release kind of bothers me. | 15:09.39 |
Robin_Watts | A press release announcing the alpha seems reasonable to me. | 15:10.33 |
paulgardiner | What's planned for February? | 15:10.34 |
Robin_Watts | paulgardiner: We release new versions of GS and Mupdf twice a year (feb and August). | 15:10.59 |
paulgardiner | Oh ok | 15:11.08 |
henrys | I prefer a press release to be associated with an actual release. | 15:11.41 |
Robin_Watts | (or at least, we try to - sometimes they slip slightly, but having dates to aim for is good) | 15:11.43 |
| henrys: Well, it's associated with the actual alpha release, right? | 15:11.56 |
| I was assuming we'd be offering a set of source/binary archives as usual? | 15:12.16 |
henrys | Robin_Watts:no I wasn't assuming we'd do a press release with the alpha, I'd like to have something that we can agree is beta before the press release. | 15:14.52 |
Robin_Watts | Well, I guess it depends on what we claim in the press release. | 15:15.38 |
henrys | paulgardiner should probably just defer on that question anyway and we can argue later. | 15:15.39 |
Robin_Watts | Any release without some sort of announcement kind of belies the point of making the release at all, right? | 15:16.07 |
tor8 | for miles point (1), the ios app is very different architecturally from the android app. each platform requires a fair amount of work to make a good UI for. | 15:16.08 |
paulgardiner | tor8: definitely | 15:16.28 |
henrys | right but the android code can be used as a model at least. | 15:17.06 |
tor8 | for point (2) didn't you mention a fairly heavy startup penalty to linking in v8? | 15:17.15 |
paulgardiner | tor8: v8 makes the debug windows build take a long time to start up from within the debugger. | 15:18.19 |
tor8 | henrys: some effort could be spent into hosting some of the duplicate work they both have into common code, but I doubt there's a lot we can do. the UI frameworks are very different, and in android it's java and for ios it's objective-c | 15:18.20 |
Robin_Watts | I thought we should do a press release that says something like: We have been working on forms support. It's now actually usable, but still in a state of flux. The purpose of this release is to gauge interest, and to invite comments from potential users. We'll do a fuller release as normal in february, but this gives interested parties a chance to comment on and affect the development of the APIs. | 15:18.24 |
paulgardiner | tor8: but the android port seems fairly unaffected. | 15:19.06 |
| Robin_Watts: that sounds perfect. | 15:19.33 |
Robin_Watts | It's a stake in the ground that we support forms, it doesn't promise the earth, and hopefully it should be enough to drag some customers out of the woodwork. | 15:20.15 |
henrys | let's back up here: when is beta? | 15:20.27 |
Robin_Watts | We haven't promised a beta at this stage. | 15:20.52 |
henrys | just for the forms code - paul's work. | 15:20.54 |
Robin_Watts | but potentially with the Feb release ? | 15:21.03 |
henrys | paulgardiner:do you have a schedule estimate? | 15:21.17 |
paulgardiner | henrys: very difficult to gage. If we are looking to have a fairly complete solution then we may still be a long way off. We are covering only a small fraction of the dom | 15:22.12 |
henrys | so in the email correct miles that what we have is alpha and use Robin_Watts words and tell him he can do the press release now and let's get back to real work ;-) | 15:24.28 |
paulgardiner | Ok great. I'll do that. | 15:24.50 |
Robin_Watts | If Miles wants a tidied up version of those words, I can do that. | 15:25.18 |
henrys | Robin_Watts:okay let's see what he says. | 15:26.26 |
| so now we can go back to paulgardiner regular todo list? | 15:26.55 |
Robin_Watts | ok. | 15:27.02 |
henrys | paulgardiner:you're good with the email? | 15:27.38 |
paulgardiner | Yeah, fine | 15:27.47 |
henrys | I did have a lock up on the calculator when I entered numbers very quickly but I couldn't reproduce it with any frequency. | 15:28.47 |
Robin_Watts | It's been a while since I saw a full statement of the todo list (ie. which things are done, which are scheduled etc) | 15:29.15 |
paulgardiner | henrys: Oh. That's a new one. I hadn't come across that. What sort of lock up? Did it recover in time? | 15:30.06 |
henrys | paulgardiner:it might have been too slow I rebooted after a bit. | 15:30.32 |
paulgardiner | Robin_Watts: for ages we were listing what was necessary for the alpha. We had submission left as the only entry when we last listed it. | 15:30.59 |
Robin_Watts | OK, so just submission left on that list. | 15:31.19 |
| So... is it worth us laying out another list of tasks? Or should we do that after submission? | 15:32.01 |
paulgardiner | henrys: so you couldn't drag the page around any more when it locked up, I assume? | 15:32.04 |
| Robin_Watts: of course the list always was a bit arbitrary | 15:32.17 |
henrys | paulgardiner:I couldn't do anything not even get back to the main menu - it locked up. | 15:32.36 |
| I'll try to reproduce it again. | 15:32.45 |
paulgardiner | Robin_Watts: I think there's still a lot to do for the android port. | 15:32.48 |
henrys | paulgardiner:performance seemed okay generally though I don't think 4 will be an issue from what I've seen. | 15:35.01 |
Robin_Watts | Well, it's not specifically androidy, right? It's tweaks to the forms/MuPDF APIs to enable any UI to perform better. Plus altering the android port to make use of the new features. | 15:35.32 |
paulgardiner | Robin_Watts: yes exactly so it's useful independently of android | 15:36.12 |
| Also there's some other ui bits that the windows has but not the android app yet | 15:36.42 |
| Having started on the android app, it might be worth making it really usable, plus doing whatever is needed at the library level to support it. | 15:37.28 |
Robin_Watts | Ok, so 1..4 on your latest mail, plus 'submission' and 'windows features to android' is your new todo list then. | 15:37.29 |
| ? | 15:37.37 |
| I personally think that having an app that makes really good use of the new features is really worthwhile. And I reckon android is the best choice for that, because it's the most compelling demo. | 15:38.34 |
paulgardiner | Yes, unless other priorities overtake | 15:38.44 |
| Robin_Watts: Totally agree | 15:38.58 |
Robin_Watts | Oh, while I remember, we had another approach from a potential customer last week, and one of the things they asked about was "Progressive Rendering". | 15:40.12 |
| By which I assume they mean "display as a file downloads". | 15:40.25 |
| tor8 and I discussed that a while ago, and we reckon it's doable without too much major upheaval. | 15:41.16 |
| so I may have a quick look at that at some point. | 15:41.39 |
henrys | I think maybe next week we should talk about "what is beta" to me it means we can do what foxit does in chrome, that was one of the motivations for starting the project. | 15:42.40 |
Robin_Watts | well, that's progressive rendering too then :) | 15:42.58 |
henrys | well that really isn't on paulgardiner though, in terms of covering the dom how much is there to do for example. | 15:44.20 |
Robin_Watts | foxit does submission right? And combo boxes. | 15:44.21 |
| and has a UI for saving out the file. | 15:44.45 |
| Inline editing? | 15:44.58 |
paulgardiner | Yes all those. | 15:45.08 |
Robin_Watts | Anything else? | 15:45.14 |
henrys | so that stuff should show up in the list? | 15:45.33 |
paulgardiner | henrys: the dom is huge. I think we have to work on the assumption we don't need it all. It could take ages, and we might struggle to find test files. We's be making our own and then perhaps the feature would be never used again | 15:46.22 |
| Robin_Watts, henrys: yes, I'll form a new tentative beta list | 15:46.58 |
Robin_Watts | paulgardiner: Do we have a list of all the dom things somewhere? | 15:47.03 |
| I could grep the forms test files, and we could see which ones are actually used. Or has the testing done that for us? | 15:47.35 |
paulgardiner | We have js_api_reference.pdf | 15:47.42 |
Robin_Watts | The testing spits warnings if something undefined is called, right? | 15:48.02 |
| "Undefined method: app.blah called" or something? | 15:48.30 |
paulgardiner | Yes. We aren't doing too badly on covering what's in our test files | 15:49.17 |
| ... accept we currently don't handle blur and focus actions which would probably exercise the dom further | 15:49.55 |
Robin_Watts | ah, right. | 15:50.21 |
| So potentially blur/focus should be high up the priority list in case they uncover some other lurking leviathan? | 15:50.44 |
| hmm. even stub implementations for them might be enough. | 15:51.14 |
paulgardiner | The calls to the actions are fairly easy to add | 15:51.52 |
henrys | I wonder if alexcher's huge corpus of pdf test files had forms examples. | 15:52.05 |
| did anyone look? | 15:52.15 |
| alexcher ? | 15:52.30 |
ray_laptop | iirc, some of the online tax pdf's are forms, but we also have some from customers. | 15:53.24 |
alexcher | henrys: I'm checking. | 15:55.19 |
henrys | paulgardiner:anyway almost time for the next meeting. I'll ask folks to keep an eye open for forms tests and see if we can get your test suite better. | 15:55.19 |
Robin_Watts | fetches tea. | 15:55.36 |
henrys | better meaning more coverage. | 15:55.39 |
paulgardiner | henrys: oh, thanks | 15:55.50 |
henrys | ray_laptop:it would be good for paulgardiner to test with those files can you give him a list. | 15:56.24 |
Robin_Watts | mvrhel: I reckon that pointer *is* being read. Specifically in gsicc_read_serial_icc at line 2122 | 15:56.43 |
henrys | pointers in the clist - no way dude! | 15:57.03 |
mvrhel | Robin_Watts: looking over all of this now | 15:57.07 |
| there are no pointers in the clist | 15:57.14 |
| huh? there there buffer of data is read from the clist | 15:58.51 |
| the pointer was allocated above | 15:58.56 |
| no pointers in the serialized portion of the profile | 16:00.34 |
henrys | so to start out this meeting I dropped a message about using affero gpl in the logs - please read about that and weigh in if you like the legal stuff | 16:00.50 |
| here I'll repeat it so you don't have to go log slogging: | 16:01.28 |
| I'm curious if anyone has given thought to ghostscript or mupdf going | 16:01.30 |
| Affero GPL? We do have one SaaS customer. I think Miles and Scott | 16:01.30 |
| should go after SaaS gpl users as if they were regular software | 16:01.30 |
| companies abusing the gpl, even though the legal loophole exists. | 16:01.33 |
| For now we can at least threaten them with Affero in future releases. | 16:01.36 |
chrisl | henrys: I had a brief look, and what I read said that affero GPL didn't address the saas question - but more reading required.... | 16:01.56 |
henrys | chrisl:the reading I found a little tricky. | 16:02.51 |
mvrhel | looks like it addresses the service provider issue | 16:03.05 |
ray_laptop | Nothing there that helps us with parasites like Oracle, is there ? | 16:03.08 |
chrisl | henrys: yes, not being legally inclined... :-( | 16:03.21 |
henrys | it allows us to demand they post source they link with as I read it. | 16:03.23 |
| but there are other problems with SaaS that can't be fixed. | 16:04.03 |
chrisl | From GNU: "One problem which the GNU Affero GPL does not address is the problem of Software as a Service (SaaS). It is impossible, as far as we know, to address this problem with a software license" | 16:04.38 |
Robin_Watts | AIUI Affero says "If you make a piece of software available to users on a server, then that's the same as you distributing that piece of software to the end users". | 16:04.51 |
| i.e. they have the rights to get everything that links with it so they can make their own equivalent server. i.e. they aren't tied into using yours. | 16:05.30 |
chrisl | FWIW, I certainly don't see any problem - except possibly with the contrib devices (again!) :-( | 16:05.41 |
ray_laptop | it would sure put the folks that have gs converting PDF's on webservers into the need to distribute source or license | 16:06.23 |
| but the 'source' on most web servers is just a little perl script | 16:06.51 |
henrys | chrisl:I think the wikipedia entry is an easier read. | 16:07.01 |
Robin_Watts | mvrhel: That clist_read_chunk call populates profile->buffer | 16:07.08 |
| The block "profile" is indeed allocated earlier. | 16:07.49 |
chrisl | henrys: well, that paragraph I quoted above is pretty unequivocal | 16:07.55 |
alexcher | We can have our own PS to PDF converter that uses the latest version and offers more configuration options. | 16:07.56 |
ray_laptop | Robin_Watts: we need to focus on the writing -- where a non-NULL pointer is written | 16:08.03 |
Robin_Watts | ray_laptop: At the moment, I'd settle for agreement on the fact that a pointer is read :) | 16:08.27 |
mvrhel | Robin_Watts: I dont see how buffer is read though in that function you mentioned | 16:08.59 |
Robin_Watts | If we are supposed to be writing pointers, but only ever NULL ones, that's a different kettle of fish. | 16:09.08 |
mvrhel | only the front part of the structure is read | 16:09.09 |
Robin_Watts | mvrhel: I put a watchpoint on &profile->buffer, and it's hit in that function. | 16:09.30 |
mvrhel | ok. I am not running it now. just looking at the code | 16:09.48 |
| let me work on it then | 16:09.52 |
henrys | chrisl:my reading was the license was ineffective at giving users rights to modify the software, but the license did insist source code be posted, if it didn't do something it would be no more than the GPL 3.0 | 16:10.12 |
ray_laptop | Robin_Watts: if we write a structure that has pointers, that is only a problem if they are non-NULL as NULL pointers won't cause GC or finalize action | 16:10.42 |
mvrhel | oh . I see it change too | 16:10.44 |
Robin_Watts | ray_laptop: I agree. | 16:10.51 |
mvrhel | that makes no sense hold on | 16:10.52 |
Robin_Watts | mvrhel: It's not a structure padding problem is it? | 16:11.01 |
mvrhel | I think it is | 16:11.08 |
chrisl | henrys: I'm not sure I follow - AGPL 3 is supposed to be compatible with GPL 3, so I have thought must allow for modifications to the source | 16:11.48 |
henrys | I didn't have much else for this meeting. Chicago was uneventful, We had very few visitors because we were all the way in the very back of the show. | 16:12.47 |
Robin_Watts | henrys: You should take a coffee machine to the show. That seemed to work well at Apps World. | 16:13.19 |
| You might not want to talk to all the people that arrive on your stand, but... | 16:13.36 |
henrys | chrisl:I guess FSF has some reality foothold. Modifying SaaS software is not going work right? There is one piece of code shared by all. | 16:13.50 |
kens | Robin_Watts : 5D used to have a sofa on the stand :-) | 16:14.11 |
chrisl | henrys: anyway, my point is that it's probably not going to give us an iron-clad way to combat SaaS suppliers, but might be a useful extra string to the bow, if we decide to use it. | 16:14.27 |
Robin_Watts | kens: I took a telly to a show once (and watched the GP on it :) ) | 16:14.35 |
| Certainly I can't see it hurting us. | 16:14.49 |
henrys | the guy selling foot inserts was doing good business anyone fool enough to walk all the way back there was already sold when they got there. | 16:14.49 |
| chrisl:no not at all, just giving miles and scott something to waive around. | 16:15.30 |
ray_laptop | :q | 16:16.04 |
henrys | it isn't clear to me why you would embed gs in an SaaS environment. | 16:16.05 |
chrisl | henrys: this seems a reasonable summary: http://www.gnu.org/licenses/why-affero-gpl.html | 16:16.16 |
henrys | ray_laptop:my sentiments exactly but your probably in the wrong app. | 16:16.23 |
| anything else for the meeting? | 16:16.48 |
alexcher | Please take a look at the bugs 693365 and 693383. | 16:17.28 |
henrys | chrisl:yeah I do find the original wording annoying. We made this for the SaaS loophole and then the last paragraph seems to say it isn't going to help. | 16:18.02 |
alexcher | Perhaps, somebody knows well what's goung on there. | 16:18.03 |
kens | 693383 looks like one for mvrhel | 16:18.44 |
mvrhel | is that the one with the LAB CIEABC image | 16:19.29 |
henrys | agreed but can you give him a bisect? | 16:19.31 |
mvrhel | I think marcos did | 16:19.39 |
henrys | alexcher? | 16:19.39 |
Robin_Watts | 693383 has been bisected. It's in the bug. | 16:19.49 |
henrys | sorry I was looking 365 | 16:20.05 |
mvrhel | just assign to me | 16:20.05 |
kens | Yes which is why I though ti was for Michael | 16:20.18 |
chrisl | henrys: I must admit, I'm not at all clear where the line is between "software running on a server" and "Software as a Service" - or, indeed, if there is a line between them...... | 16:20.49 |
kens | THe clist and transparency one is beyond me. | 16:20.58 |
Robin_Watts | kens: I wonder if that's the same issue that mvrhel is looking at now. | 16:21.26 |
kens | has no idea :-) | 16:21.37 |
henrys | chrisl:one's written in ruby and the other C ? | 16:21.42 |
mvrhel | I have to head to a dr. appt. will be back on line hopefully shortly | 16:22.23 |
henrys | okay I was ready to wrap up early anyway | 16:22.49 |
| let's call it done. | 16:23.10 |
kens | Suits me thanks | 16:23.21 |
| Goodnight all. | 16:23.26 |
Robin_Watts | mvrhel: Yes, it's a padding problem. | 16:23.33 |
chrisl | henrys: it could be as simple as that! I wondered if it was more like SaaS would be something that would have a "local" result. So dynamically creating a PDF to view online is "running on a server", but a system that converts your document to PDF and lets you download it would be SaaS. | 16:23.58 |
ray_laptop | mvrhel: Robin_Watts: so when we get the profile back from the gsicc_serialized_profile we probably need to clear out the rest of the cmm_profile (that has the pointers) | 16:24.04 |
Robin_Watts | At least sizeof(gsicc_serialized_profile_t) = 256 and &profile->buffer and profile differ by only 252. | 16:24.06 |
mvrhel | Robin_Watts: ray_laptop: yes | 16:24.25 |
ray_laptop | sorry phone call (cust 532). | 16:24.39 |
mvrhel | the writing though I thought would have taken care of it though too | 16:24.46 |
Robin_Watts | ray_laptop: Or we need to not use sizeof(gsicc_serialized_pointer) but instead use offsetof. | 16:24.47 |
mvrhel | oh never mind | 16:25.02 |
| I see | 16:25.05 |
| ok. yes it makes sense to clear it out after reading back | 16:25.16 |
paulgardiner | Robin_Watts, tor8: Oh, there's a commit on paulg/master, btw. | 16:25.25 |
mvrhel | that will fix things | 16:25.26 |
Robin_Watts | but we can talk about this later after mvhrel gets back from Drs. | 16:25.27 |
mvrhel | yes. need to head out.. bbiaw | 16:25.37 |
henrys | tor8:have you considered affero for mupdf? | 16:26.40 |
nickelpigga | Hi guys, I have a question about MuPDF which I could not find an answer to in google or MuPDF manual. After I opened a file (from firefox) how can I save a copy of the pdf locally? | 16:32.00 |
Robin_Watts | nickelpigga: Currently you cannot. | 16:32.31 |
| BUT... it's not hard to add. | 16:32.38 |
nickelpigga | to add to source you mean? | 16:32.54 |
Robin_Watts | yes. | 16:32.59 |
nickelpigga | okej, thank you I will download it and try to do it. thank you :) | 16:33.32 |
Robin_Watts | best of luck. Let us know how you get on. | 16:34.00 |
tor8 | henrys: I'm not opposed to it. | 16:55.29 |
Robin_Watts | Woo Hoo! I have auto page advance working on windows. | 16:57.14 |
henrys | what we need is a google search bar in the mupdf viewer ;-) | 17:17.35 |
Robin_Watts | henrys: Or a device that replaces text words with googled for links. | 17:32.46 |
mvrhel_laptop | ok. so sure enough. setting the buffer ptr to NULL during the serial read back fixes the issue | 17:41.09 |
| is that too much of a hack around a structure alignment issue? | 17:41.20 |
Robin_Watts | IMHO< yes. | 17:45.43 |
| It's fixing up something that we've got wrong. | 17:46.03 |
| Why not just get it right in the first place? | 17:46.10 |
| It's just a question of writing/reading the correct number of bytes. | 17:46.23 |
| And I sent an email to you/ray that shows how we can do that. | 17:46.39 |
mvrhel_laptop | ok | 17:47.30 |
| Robin_Watts: ok. I like your approach better | 17:48.06 |
| much cleaner | 17:48.13 |
Robin_Watts | ah, cool. Was worried I was being dogmatic and awkward :) | 17:48.27 |
mvrhel_laptop | well, the other approach, while simpler feels hackish to me | 17:49.05 |
Robin_Watts | yeah. It's a nice proof that that is the problem though. | 17:49.45 |
| sebras: Ahem. I'd forgotten to add a file to the functions commit, so that's why you were seeing missing symbols. | 18:01.29 |
| Sorry about that. | 18:01.32 |
| tor8, sebras: What is the timer stuff in the current x11 thing about? | 18:04.45 |
| Oh, its for the page number display, presumably for WMs with no title bars. | 18:07.29 |
| Bah. Who wrote timersub and timeradd. | 19:22.10 |
nickelpigga | Robin_Watts, took me some time to understand... it is there like you said, I tried with adding 'A' to call pdfapp_save... just X11 part is missing :/ | 20:17.56 |
Robin_Watts | nickelpigga: indeed. | 21:35.20 |
sebras | Robin_Watts: yeah, I'm using a wm without titlebars. | 22:15.13 |
| Robin_Watts: been doing so since 2000 or so. | 22:15.29 |
henrys | gawd I can refinance at 2.8 % interest. with inflation that's like free money | 23:16.22 |
Robin_Watts | sebras: I don't ever see the page number on ubuntu though :( | 23:28.07 |
| henrys: What are the arrangement fees like? | 23:28.41 |
| OK. Auto advance working under X too. Bedtime. | 23:40.11 |
henrys | Robin_Watts:just looking into it. | 23:47.51 |
metta | hello | 23:49.07 |
ghostbot | que tal, metta | 23:49.07 |
metta | I get Shell Script Invocation Error 'stdio.h' file not found when trying to run muPDF as an xcode project | 23:50.14 |
| I assume its because I didn't add the third party dependencies, but I'm not sure how to do that... | 23:50.45 |
| Forward 1 day (to 2012/10/17)>>> | |