| | <<<Back 1 day (to 2013/09/02) | 2013/09/03 |
| vtorri | hey | 05:34.33 |
| ghostbot | bonjour | 05:34.33 |
| vtorri | i would like to use the gs API | 05:34.43 |
| | when should I create an instance (with gsapi_new_instance) ? | 05:35.20 |
| | at the beginning of my program ? | 05:35.30 |
| | or when I open a file ? | 05:35.39 |
| | or when I render a page ? | 05:35.47 |
| | chrisl: do you have an idea about my questions ? | 06:46.17 |
| chrisl | vtorri: not really, not my area. Although, "when you render a page" won't really fly with Postscript...... | 06:48.35 |
| vtorri | you're not a ghostscript dev ? | 06:49.05 |
| | if so, who can answer my questions ? | 06:49.26 |
| chrisl | Yes, I am, but I don't really deal with the "front end" API. | 06:49.34 |
| | TBH, I don't there is an answer - these are design decisions you need to make based on your application | 06:50.14 |
| vtorri | first, as there is no way to know if the library has been compiled with GS_THREADSAFE enabled or not, i can only create one instance at a time | 06:51.21 |
| | and imho, an API should be added to know that | 06:51.50 |
| chrisl | You have a point, yes. | 06:52.59 |
| vtorri | but i would like to know what the instance is used for | 06:53.18 |
| | is it specific to a rendering ? | 06:53.34 |
| | or can it be used for any rendering (in that case, one instance at the beginning of my progrm should be sufficient) | 06:54.00 |
| chrisl | It's the interpreter instance - it contains the entire state of the Postscript interpreter | 06:54.00 |
| vtorri | so i can call gsapi_init_with_args() with the same instance, but different arguments, during all my program life, right ? | 06:55.23 |
| chrisl | No, you don't want to init an already initialised instance | 06:55.49 |
| | If you want my opinion, I would create an instance for a given job, and destroy it at the end of the job. | 06:55.52 |
| vtorri | ok | 06:57.03 |
| | thank you | 06:57.08 |
| chrisl | That also leaves it open for you to scale it up once you can safely use threads | 06:58.01 |
| vtorri | well, as i don't know if, for example, a linux distribution has compiled libgs with GS_THREADSAFE or not... | 06:59.39 |
| chrisl | Sure, but at some point, we'd like to make GS_THREADSAFE the default, the problem is, there are loads of devices that we didn't write, have no way of testing, and the original authors have disappeared or are no longer interested - but people still use. | 07:01.02 |
| Vignesh_ | Hello friends. | 07:22.00 |
| | I want integrate MuPDF reader in my iPhone application. | 07:22.59 |
| | And I'm ready to buy your protect MuPDF. | 07:24.29 |
| chrisl | Vignesh_: you'll need to contact our sales guy: contact details are here: http://www.artifex.com/ | 07:26.39 |
| | Vignesh_: on here, we're just engineers, and generally try to steer clear of the business end of things! | 07:27.05 |
| Vignesh_ | Ok Thanks for your kind info. | 07:27.15 |
| chrisl | NP, if you don't hear back in a few days, then by all means, come back and bug us on here | 07:28.00 |
| Vignesh_ | Sure. | 07:29.26 |
| | I have already implemented MuPDF in my android Demo project. | 07:30.18 |
| chrisl | Cool | 07:30.29 |
| Vignesh_ | May i ask you some questions? | 07:30.39 |
| chrisl | You'd be better to wait a couple of hours, the mupdf devs should be around by then (I'm a ghostscript dev) | 07:31.13 |
| Vignesh_ | Ok sure. | 07:31.49 |
| | And Thanks too.. | 07:32.02 |
| chrisl | You can certainly post your questions, this channel is logged, and we'll make sure the appropriate people see the questions. | 07:32.19 |
| | Vignesh_: and if time zones and such make it difficult for you to hang around, you also also read any answers on the logs: http://www.ghostscript.com/irclogs/current.html | 07:33.15 |
| | vtorri: actually, could you create a bug report about adding a runtime way to check for GS_THREADSAFE, please? | 07:34.51 |
| Robin_Watts | Vignesh_: Hi. Ask away. | 08:18.34 |
| Vignesh_ | Hi robin, Now i am trying to develop offline PDF reader in IOS for my demo project. | 08:38.22 |
| Robin_Watts | ok. | 08:38.38 |
| Vignesh_ | In my Android project, I've implemented MuPDF as offline reader. | 08:38.57 |
| | But in iOS, I am not able to integrate MuPDF with my iOS app demo. | 08:39.53 |
| Robin_Watts | ok. I am not an ios expert, but I will try to help. tor7 may know more. | 08:40.27 |
| Vignesh_ | I have already clone MuPDF latest version 1.3 using gitHub. | 08:40.58 |
| tor7 | Robin_Watts: Vignesh_: on iOS the mupdf library will work out of the box. | 08:41.01 |
| | no source modifications required | 08:41.21 |
| | but you may have some issues with apple's, um, "creative" tool chain for compiling | 08:41.47 |
| Vignesh_ | yeah tor& | 08:42.21 |
| | May i convert MuPDF as Static library in IOS ? | 08:42.54 |
| tor7 | that's how we build mupdf for our iOS app ... as a static library | 08:44.04 |
| Vignesh_ | is it MuPDF 1.3 is a library project? | 08:45.04 |
| tor7 | Vignesh_: you can try the Xcode project in the "platform/ios/" directory | 08:45.28 |
| Vignesh_ | Ya tor7. I have used it in Xcode. It works very well. But how to integrate that one into my OFFline reader project? | 08:46.58 |
| tor7 | what do you mean with "offline"? | 08:48.33 |
| Vignesh_ | I have set PDF in my device. I want to read that via my App with help off muPDF. | 08:50.15 |
| Robin_Watts | Vignesh_: Why is this problem any different on ios than it is on Android? | 08:50.52 |
| | What level were you working at on android? Did you call the C API, or did you just take our viewer app, and tweak it at the java level ? | 08:51.30 |
| Vignesh_ | I implemented in Java level only. | 08:55.14 |
| | As a Viewer app. | 08:55.23 |
| Robin_Watts | Ah, well, you will find ios very different. | 08:55.35 |
| | MuPDF is a C library, and it offers an API at the C level to allow you to open files, load pages, render pages etc. | 08:56.00 |
| | We have built example viewer apps on top of that API for both android and ios, but the shapes of those apps are heavily influenced by the operating system in use. | 08:56.39 |
| | That is to say, the android app looks very different internally to the ios app. | 08:56.56 |
| | So while the android app has an activity, views, adapters etc, the ios app has a different set of classes. | 08:57.46 |
| Vignesh_ | i cann't find the documentation for integrating mupdf in my IOS app. | 08:57.54 |
| Robin_Watts | There isn't any. | 08:58.04 |
| | The public API level for MuPDF is at the C level. | 08:58.18 |
| | All the interfaces within the ios app are considered private and undocumented. | 08:58.38 |
| Vignesh_ | Can we access mupdf APi via objective-c? | 08:58.53 |
| Robin_Watts | If you want to fiddle in there, feel free, but beware of sharp edges and unexpected sharp drops. | 08:59.06 |
| | Vignesh_: I believe objective C can call normal C, sure. | 08:59.22 |
| Vignesh_ | I tried integrating but mupdf library failed to run. | 09:05.23 |
| Robin_Watts | Vignesh_: You'll need to give us more details than that. | 09:06.01 |
| Vignesh_ | Sure. I have debugged in My project. App crash occurred in static renderPage() method. | 09:10.26 |
| | In this method, I got bad access code error like 'MuPDFDemo`fz_device_rgb:' | 09:11.55 |
| Robin_Watts | Vignesh_: Well, there is not enough information there for me to even know where to start, sorry. | 09:24.30 |
| | Vignesh_: Are you a commercial customer or supported customer? | 09:25.40 |
| Vignesh_ | Yes, I am a Commercial customer. | 09:41.22 |
| | And now i got tool chain error in ios | 10:34.47 |
| Robin_Watts | Vignesh_: Without more details we cannot help you. | 11:03.45 |
| | Your best bet is to either open a bug and attach all the details to the bug, or to send all the details to support@artifex.com | 11:04.09 |
| | but with vague things like "now I have a toolchain error", we cannot start to help you. | 11:04.24 |
| Vignesh_ | Sure, i will send documentation regarding bugs and my queries. | 11:08.49 |
| | Now i have done half of document. I will send you soon. | 11:09.29 |
| Robin_Watts | ++tor | 11:24.44 |
| sebras | /help | 11:25.29 |
| | /nick tor9 | 11:25.37 |
| | is there a reason that the mupdf/android app doesn't support http://..../file.pdf? I guess we could do it if we want to..? | 11:26.25 |
| Robin_Watts | sebras: I haven't done a mupdf build of curl. | 11:37.34 |
| | Possibly android has http fetchers we should leverage rather than using curl? | 11:37.59 |
| | in short, yes we probably could do it. | 11:38.09 |
| sebras | Robin_Watts: mmm, I think androids http interfaces is what we would want to use. | 11:39.40 |
| tor8 | Robin_Watts: that's ... curious. | 13:05.57 |
| | the iOS app spews out lots of lock ordering errors | 13:06.15 |
| Robin_Watts | tor8: interesting. | 13:38.44 |
| | which locks in particular? | 13:55.27 |
| kens | Robin_Watts : have you had a look at that interpolation report from customer #531 ? | 14:12.58 |
| Robin_Watts | what where eh? | 14:13.12 |
| kens | support mail "RE: GS 9.07 PDF issue with -dDOINTERPOLATE" | 14:13.30 |
| Robin_Watts | I see it. | 14:13.44 |
| | I have not. I will look now. | 14:13.48 |
| | Thanks for telling me. | 14:14.00 |
| kens | OK I cannot get it to work with bare 9.07, so I'm assuming there was some bug in 9.07 and htey have a patch. For me it works with the current code | 14:14.18 |
| | If you are busy I can try a git bisect | 14:14.36 |
| Robin_Watts | They have a fix for their previous problem. | 14:15.14 |
| | I can try going to 9.07 and adding their fix. | 14:15.38 |
| kens | Yeah I'm assuming it s a new problem, but I don't remember what the previous one was | 14:15.40 |
| | and 9.07 doesn't work with tiffpack for me | 14:15.55 |
| Robin_Watts | leave it with me, at least for a bit. thanks for the offer of help though. | 14:15.56 |
| kens | OK NP | 14:16.00 |
| tor7 | Robin_Watts: mostly "Attempt to take lock 0 when held already!" | 14:18.38 |
| | then errors about "invalid page object" | 14:18.48 |
| | but I get those errors normally too, so that's something else gone wrong | 14:19.10 |
| | Robin_Watts: and the occasional "Attempt to take lock 3 when 0 held already!" | 14:19.46 |
| Robin_Watts | Attempt to take lock 3 when 0 held is trying to take the glyphcache lock during allocation. | 14:20.56 |
| tor7 | it wouldn't surprise me if the GCD thread crap has broken due to some API change or other | 14:21.25 |
| Robin_Watts | I really think it must mean we have 2 threads running at once. | 14:21.41 |
| | I should do a clustertest with the thread debugging forced on. | 14:22.08 |
| tor7 | oh, and yay! "dispatch_get_current_quue" is deprecated. | 14:22.20 |
| | and more yay! "UITextAlignmentCenter" is deprecated. | 14:22.30 |
| | and of course "openssl is deprecated" | 14:22.42 |
| | and the magic adjustments to the scrolling canvas when hiding/showing the tool bar is different in ios 7 beta, with lots of breakage | 14:23.26 |
| | and egads the icons are fugly | 14:23.38 |
| henrys | mvrhel_laptop: you're back! | 14:28.36 |
| | tor7:can you close the xps bugs that were fixed by shelley's changes or ask him to do it? | 14:32.16 |
| Robin_Watts | I haven't actually seen mvrhel_laptop speak on here. He may be jetlagged into unconciousness. | 14:37.45 |
| jitender | hi i need help , i am using mupdf | 14:37.51 |
| Robin_Watts | jitender: go on... | 14:38.17 |
| jitender | i want create my signed application with mupdf library | 14:38.35 |
| | but its create problem | 14:38.43 |
| tor7 | jitender: ios? win8? android? | 14:39.09 |
| jitender | can i create signed application with mupdf library in android | 14:39.13 |
| | android | 14:39.25 |
| Robin_Watts | jitender: You can. | 14:39.57 |
| | In fact, if you follow the instructions in platform/android/ReadMe.txt it tells you how to build our viewer, and (if you have a key setup) then you can sign that. | 14:40.38 |
| jitender | when i create an normal apk i works well but when i craete signed apk my application crash when i use the code of mupdf library | 14:40.47 |
| Robin_Watts | jitender: So you're putting the mupdf library into your apk? | 14:41.12 |
| jitender | yes | 14:41.24 |
| Robin_Watts | and is it crashing at the point at which MuPDFCore loads the libmupdf.so ? | 14:41.44 |
| | or at the point at which you start calling into it ? | 14:41.52 |
| jitender | yes | 14:41.53 |
| | yes sir | 14:41.59 |
| Robin_Watts | Which point is it crashing at? The point at which the lib is loaded, or the point at which the lib is first called? | 14:42.48 |
| jitender | when iam using the code of mupdf library in my signed application it crashes | 14:43.35 |
| Robin_Watts | and we're running in cricles... | 14:44.00 |
| jitender | yes sir | 14:46.32 |
| | is there any condition when i use the library of mupdf with my app for publishing to market | 14:48.42 |
| | ???? | 14:48.46 |
| Robin_Watts | jitender: Yes. | 14:49.00 |
| | MuPDF is licensed in 2 ways. | 14:49.09 |
| jitender | what condition?? | 14:49.10 |
| Robin_Watts | Firstly it's licensed under the GNU GPL. | 14:49.22 |
| | or rather the GNU APGL v3, sorry. | 14:49.34 |
| jitender | u can say if i have no licence i can't publish my app in market with mu pdf | 14:50.24 |
| Robin_Watts | There are many restrictions in that license, but the basic upshot is that if you want to use MuPDF under that license, you need to license your entire app under the GNU AGPL. | 14:50.25 |
| | Licensing your app under the GNU AGPL will involve you giving away your source code. | 14:51.34 |
| | If you're not prepared to abide by the terms of the GNU AGPL then your choices are 1) license MuPDF commercially from Artifex, or 2) not to use MuPDF. | 14:52.12 |
| jitender | when i check my app standlone its work fine but when i check it for publishing its not work | 14:52.18 |
| | is there any licence issue????? | 14:52.36 |
| tor7 | jitender: is your app open source? | 14:52.51 |
| Robin_Watts | jitender: If your app is licensed under the GNU GPL, then you can use MuPDF without a license issue. | 14:53.12 |
| jitender | no sir | 14:53.17 |
| Robin_Watts | Otherwise, you must get a commercial license from Artifex. | 14:53.26 |
| tor7 | jitender: then you need a commercial license. | 14:53.27 |
| jitender | what amount i have to pay for licence????? | 14:54.05 |
| Robin_Watts | jitender: It varies according to your needs. | 14:54.20 |
| | but it's a non-trivial amount for an individual. | 14:54.51 |
| jitender | i need it for only pdf page to image | 14:54.55 |
| Robin_Watts | jitender: It sounds like you should send a mail to scott.sackett at artifex.com saying you're interested in licensing mupdf for use in your app. | 14:56.25 |
| | He'll send you back a huge scary list of questions. | 14:56.42 |
| | You answer the questions as best you can. | 14:56.51 |
| | The idea is to give him an idea of exactly what your app does, and what proportion of it's functionality depends on MuPDF. | 14:57.21 |
| | Then he'll give you a price. | 14:57.27 |
| jitender | kkkk | 14:57.42 |
| | sir | 14:57.44 |
| | how many year licence is valid???? | 14:58.16 |
| Robin_Watts | it would be a per year license fee. | 14:58.34 |
| jitender | ok sir | 14:59.05 |
| | thanx for ur interest | 14:59.16 |
| Robin_Watts | good luck. | 14:59.28 |
| mvrhel_laptop | I am here | 14:59.31 |
| | First day of school so a bit chaotic here | 14:59.40 |
| Robin_Watts | mvrhel_laptop: Hi. Do you have that color polarity thing to hand? I'd like to get that in before the staff meeting if possible. | 15:00.28 |
| mvrhel_laptop | Robin_Watts: yes. I will commit it today | 15:00.38 |
| | had a couple compiler warnings I wanted to fix first | 15:00.47 |
| | with it | 15:00.49 |
| Robin_Watts | mvrhel_laptop: ok, ta. | 15:01.00 |
| mvrhel_laptop | are we doing a meeting today? | 15:03.02 |
| | I am guessing not | 15:03.08 |
| Robin_Watts | mvrhel_laptop: I think henrys said just to make sure we all look over the workflowy/agenda thing to get our stories straight before the meeting. | 15:04.00 |
| mvrhel_laptop | gotcha | 15:04.12 |
| Robin_Watts | did you have anything you wanted to discuss? | 15:04.26 |
| | anything mupdfy come up in japan? (that you can talk about in public) | 15:04.41 |
| mvrhel_laptop | no not really. I did have the android app act up on me on the nexus 10 though. It was very confused when opening documents from the file explorer app | 15:05.13 |
| henrys | mvrhel_laptop: did you see my email about the tech agenda? I thought you might have stuff to add from the trip. | 15:05.40 |
| Robin_Watts | mvrhel_laptop: you'll have to show us that at the meeting. | 15:05.51 |
| mvrhel_laptop | henrys: yes. I will review that today | 15:05.57 |
| | Robin_Watts: yes. I will bring the nexus with me | 15:06.28 |
| henrys | mvrhel_laptop: also can you leave the large nexus in chicago with me, or I could get a new one, I think the small one sucks for demos. | 15:06.33 |
| | ? | 15:06.34 |
| mvrhel_laptop | henrys: I can do that. I think it would be nice though if we did get one other. I do like having an android device to keep up with all that paulgardiner is doing and testing it out | 15:07.25 |
| Robin_Watts | LG have a nice new one out. | 15:08.00 |
| henrys | mvrhel_laptop: I think I'll just buy another, is there something different I should buy for testing purposes? | 15:08.04 |
| Robin_Watts | 8" with full HD. | 15:08.10 |
| | (or at least they've just announced it). | 15:08.33 |
| mvrhel_laptop | cool | 15:09.16 |
| henrys | samsung galaxy? | 15:09.49 |
| Robin_Watts | god no. | 15:09.57 |
| | tor7 has one of them, I think. | 15:10.21 |
| tor7 | I have one of the original galaxy tabs. it's dreadful. | 15:12.55 |
| | zero battery life | 15:13.01 |
| henrys | the toshiba excite lineup with stylus looks interesting. | 15:14.51 |
| | price and performance though nexus 10 looks strongest. | 15:18.08 |
| Robin_Watts | I think the transformer infinity is still holding up well. | 15:19.20 |
| | the 10 had a lack of expansion ports, if memory serves. | 15:19.37 |
| mvrhel_laptop | yes that is the only downside of the 10. but it does make a nice demo | 15:20.39 |
| Robin_Watts | supposedly there will be a new Nexus 10 this year. | 15:23.05 |
| | http://www.knowyourmobile.com/google/google-nexus-10-2/20987/new-nexus-10-2-2013-be-manufactured-asus | 15:23.24 |
| henrys | the transformer does look good. Having a variety of devices is good for us, so maybe another 10 is a bad idea. | 15:25.55 |
| Robin_Watts | if there is to be a new 10, that might be good. | 15:27.23 |
| henrys | hopefully I can get it locally. | 15:31.06 |
| tor7 | are there any low-res 10" tablets? the nexus ten sometimes has trouble driving all those pixels at smooth frame rates. | 15:31.29 |
| henrys | I have this locally, no wait: http://www.bestbuy.com/site/searchpage.jsp?_dyncharset=ISO-8859-1&_dynSessConf=5151998713977932719&id=pcat17071&type=page&ks=960&st=asus+pad&sc=Global&cp=1&sp=&qp=&list=n&usc=All+Categories&nrp=82&fs=saas&iht=n | 15:35.01 |
| Robin_Watts | henrys: Apparently there is a 2013 refresh of the transformer infinity | 15:36.42 |
| | http://www.trustedreviews.com/news/asus-transformer-pad-infinity-gets-tegra-4-upgrade | 15:37.01 |
| | 2560x1600 panel - is that low enough res tor? | 15:37.19 |
| | :) | 15:37.21 |
| henrys | Robin_Watts: I need something for Chicago | 15:38.07 |
| Robin_Watts | henrys: oh, for chicago, steal the nexus 10. | 15:39.06 |
| | at least we know that works :) | 15:39.12 |
| henrys | mvrhel_laptop: so I take the 10 and ship it back to you immediately after the show? | 15:40.08 |
| mvrhel_laptop | henrys: that is fine. | 15:40.30 |
| henrys | okay great | 15:41.02 |
| mvrhel_laptop | bbiab | 15:41.16 |
| henrys | Robin_Watts: that article is from 6/3 it may already be in the stores | 15:42.36 |
| Robin_Watts | http://www.knowfirst.info/forums/showthread.php?t=45889 | 15:43.48 |
| | 25th August had it just being unveiled. | 15:44.01 |
| | oh, that's a repost, sorry. | 15:44.39 |
| ray_work | henrys: For shows, I've suggested to Miles that a large screen would be good. Many (most) of the tablets have HDMI out. If he doesn't want to ship one, he can probably just rent one for the show. | 15:45.14 |
| tor7 | Robin_Watts: that would be a pretty sweet netbook... if you could run something other than a phone OS on it... | 15:45.23 |
| Robin_Watts | ray_work: the transformer prime infinity will output 4K via hdmi :) | 15:45.56 |
| ray_work | it wouldn't have to be a super large screen, but visibility would be a LOT better | 15:46.00 |
| Robin_Watts | And if we get it working on the gamestick, we can just use that. | 15:46.29 |
| ray_work | Robin_Watts: I wonder if doing 4K would slow down the page display rate, or if mupdf is the primary limiting part | 15:47.02 |
| henrys | ray_work: well that isn't really the sort of demo's I've doing at the shows, folks want to walk around with a mobile device and see the product as it would be used. I don't know if Robin_Watts or mvrhel_laptop show experience is different. | 15:47.41 |
| Robin_Watts | henrys: IME at most shows, I'd wave the device at people, open a PDF, flip through a few pages, zoom, and people would have seen enough. | 15:48.19 |
| ray_work | henrys: for the couple of shows here in LA, I brought in my 24" monitor and it caught a lot more attention than the lame-o (IMHO) static booth with a couple of people standing in it | 15:50.12 |
| henrys | ah "in my experience" not "I mother earth" ;-) | 15:51.34 |
| ray_work | henrys: but I agree that once someone is interested enough to want a demo, letting them hold it and "drive it" themselves is better. That's step 2 where you set the hook ;-) | 15:51.49 |
| henrys | ray_work: should I add it to the agenda? | 15:53.30 |
| ray_work | BTW, has anyone played with --saved-page or --saved-pages-test ? I'd really like to get some more time on it. | 15:53.33 |
| henrys | the large screen that is. | 15:53.41 |
| ray_work | I think it's mostly up to our marketing dept. We don't really need a technical discussion unless enough of us think it's important and we want to gang up on Scott and Miles. | 15:55.01 |
| | otherwise, less agenda items is better :-) | 15:55.13 |
| | I don't really count on much from shows anyway | 15:55.47 |
| | oops. Have to run an errand. bbiab. | 15:56.06 |
| henrys | Robin_Watts: I don't see any refreshed infinities in the wild yet. | 16:01.40 |
| Robin_Watts | no, I haven't seen any news articles with firm release dates or pricing yet. | 16:02.03 |
| sebblonline | Hi I wrote a few days ago concerning memory leaks in mupdf. I notice some leaks with memento but only with some specific PDF files. | 16:03.45 |
| Robin_Watts | sebblonline: Ok, so there may be some cases we've missed. Could you let us have those PDF files? | 16:04.05 |
| sebblonline | one log looks like this for example: http://pastebin.com/xHjVhhw6 | 16:04.16 |
| Robin_Watts | sebblonline: With the PDF file, that's probably an easy fix. | 16:04.44 |
| | without it, hopeless. | 16:04.50 |
| sebblonline | ok, i understand | 16:04.58 |
| Robin_Watts | (well, not hopeless, just much harder :) ) | 16:05.08 |
| sebblonline | i try to find one, that I can give you :) | 16:05.15 |
| Robin_Watts | sebblonline: Thanks. | 16:05.22 |
| ray_work | probably one of the third party libs (my guess) | 16:38.03 |
| sebblonline | this is an online catalogue, which can be downloaded too: http://www.mitsubishicarbide.com/EU/de/product/epaper/index.html | 16:39.28 |
| ray_work | but the fact that it's pdf_obj thingys doesn't really suggest that | 16:39.40 |
| sebblonline | the seventh icon at the bottom to download | 16:39.43 |
| | it's 300MB! | 16:40.06 |
| chrisl | ray_work: I just noticed that jbig2dec uses heap memory, it doesn't hook a Ghostscript memory manager - I'm thinking we should change that? | 16:40.42 |
| ray_work | sebblonline: must have LOTS of images in it | 16:40.45 |
| sebblonline | yes it does! | 16:41.05 |
| | thats what I assumed is the reason for the leaks | 16:41.24 |
| ray_work | chrisl: definitely, but most of our customers use luratech, so it isn't super critical (IMHO) | 16:41.36 |
| sebblonline | its the only document we have atm with these problems | 16:41.45 |
| ray_work | sebblonline: well, at least that's good news | 16:42.03 |
| chrisl | ray_work: <sigh> the luratech code seems to use heap memory, too :-( | 16:43.07 |
| ray_work | chrisl: It does ? I thought it hooked our memory manager. My mistake. That's much more critical to fix. I must have fixed it for cust 532 and then forgot to port it to our code | 16:44.19 |
| | chrisl: If you want, I'll fix the luratech, since I know I have it working for cust 532. (they only have about 1M heap memory, and even luratech uses more than that for JPX images) | 16:45.13 |
| chrisl | ray_work: yeh, that's good. I'll look at sorting jbig2dec on the plane | 16:45.46 |
| ray_work | chrisl: OK. Do you want to open a bug (so I don't forget it) ? | 16:46.21 |
| chrisl | ray_work: will do | 16:46.40 |
| | ray_work: oddly, the luratech JPX code does use the GS memory manager.... | 16:48.28 |
| ray_laptop | chrisl: Oh, so it's just the jbig2 decoder I need to fix ? | 16:48.56 |
| Robin_Watts | ray_work: No, it's definitely a mupdf leak (not the thirdparty) | 16:49.00 |
| ray_laptop | chrisl: maybe I did fix it (but just for the JPX) :=/ | 16:49.22 |
| | Robin_Watts: yeah, I just like to blame the other guys :-) | 16:49.43 |
| chrisl | ray_laptop: it's the jbig2 interface code - everything is in the luratech code to use memory manager callbacks, it's just the callbacks in sjbig2_luratech.c just use malloc/free | 16:50.17 |
| ray_laptop | chrisl: then that'll be a a realtively easy fix :-) | 16:51.17 |
| chrisl | Yep, it shouldn't be a problem - the jbig2dec one is going to be more involved, but not mentally very taxing (hence a good candidate for a flight!) | 16:52.01 |
| ray_laptop | glad I volunteered for the easier one :-) | 16:52.52 |
| chrisl | ray_laptop: :-) Bug 694559 | 16:53.15 |
| kens2 | OK goodnight all | 16:59.33 |
| Robin_Watts | sebblonline: Downloading the document now. | 17:07.32 |
| | You see the leaks when rendering all pages? or just page 1? | 17:07.49 |
| sebblonline | I used the basic win32 example included in the mupdf lib. The leaks appear, when I at least change 1 page | 17:10.53 |
| | so only opening the first page and closing doesnt give the leaks | 17:11.07 |
| ray_laptop | there's 1233 bogus xref entries that show up with gs: **** Warning: considering '0000000000 XXXXX n' as a free entry. | 17:12.10 |
| Robin_Watts | sebblonline: Possibly a stupid question... | 17:12.21 |
| | mupdf has a store (a cache) of objects etc that is only cleared when you destroy a context. | 17:12.43 |
| | Presumably you're generating the memento output after closing the context? | 17:13.05 |
| | ok, leaks are on page 2. | 17:13.31 |
| sebblonline | hm, I didnt change the example of mupdf | 17:14.38 |
| ray_laptop | pages 1, 2, 505 and 1233 use transparency | 17:14.43 |
| Robin_Watts | sebblonline: Ah, ok. | 17:15.47 |
| sebblonline | so i think the memento output is created, after context is destroyed? | 17:15.52 |
| Robin_Watts | yeah., | 17:16.02 |
| | Looks like these are SMasks named in gstates that are being leaked. | 17:16.15 |
| ray_laptop | whoever created this file (using pdftk) was not very careful. It has something I don't often see: a non-embedded font with an embedded font name: FTERWO+ArialMT | 17:16.16 |
| | Robin_Watts: We have lots of files that have that | 17:16.50 |
| henrys | mvrhel_laptop: when you get a chance have a look at http://bugs.ghostscript.com/show_bug.cgi?id=694435 comment #3 | 17:17.05 |
| bigjohntoday | i run html2ps http://yahoo.ca | ps2pdf - yahoo.pdf and i get the nocurrentpoint error | 17:27.16 |
| | ghostscript v9.1 | 17:27.29 |
| Robin_Watts | 9.10, presumably. | 17:31.21 |
| | bigjohntoday: Run html2ps http://yahoo.ca > yahoo.ps | 17:32.26 |
| | Then go to bugs.ghostscript.com and open a bug and attach yahoo.ps to it. | 17:32.49 |
| | Then when we get a chance we can look at it. | 17:33.04 |
| | I strongly suspect the answer will be that html2ps is generating broken postscript. | 17:33.19 |
| bigjohntoday | okay let me go do that thanks | 17:33.55 |
| | that was weird... if i redirect to ,ps file then utilize a .ps file in the ps2pdf it works | 17:35.20 |
| | but a pipe kills it | 17:35.23 |
| Robin_Watts | ah. | 17:36.21 |
| bigjohntoday | okay | 17:36.38 |
| | i got a different one i'll post | 17:36.44 |
| | heh | 17:36.47 |
| sebblonline | bye | 18:00.01 |
| mvrhel_laptop | henrys: looking at comment. let me look at the code now | 18:21.55 |
| henrys | mvrhel_laptop: thanks sorry to mug you as soon as you get back | 18:23.45 |
| mvrhel_laptop | np | 18:24.28 |
| | henrys: so a question for you | 18:27.29 |
| | actually, let me just run the test and see for myself | 18:27.53 |
| henrys | mvrhel_laptop: I don't think pcl has a 5 component device by default like with gs you can use tiffsep | 18:29.12 |
| mvrhel_laptop | henrys: no I dont think that is the issue | 18:29.26 |
| | it is odd that it is resetting the profile | 18:29.53 |
| | gsicc_init_device_profile_struct should not be getting called after the profile has already been set | 18:30.32 |
| | I need to understand where else the profile is getting set | 18:31.00 |
| henrys | mvrhel_laptop: they call it in their device. | 18:32.28 |
| mvrhel_laptop | oh | 18:32.35 |
| | but the languages do need to set a default profile if one has not been set | 18:33.12 |
| | I understand your question now | 18:33.18 |
| | right now the languages don't specify a profile. They simply tell gs to initialize one. | 18:35.09 |
| | in case one has not been set. | 18:35.35 |
| | and gsicc_init_device_profile_struct will not blow away the current one | 18:36.45 |
| | so I am a little confused now | 18:36.49 |
| | the call should not unload the properly installed profiles | 18:37.19 |
| | there is a check and if the curr_profile != NULL but we called with profile_name == NULL we simply return | 18:38.01 |
| | hmm hold on | 18:38.31 |
| | oh I see | 18:38.54 |
| | crap | 18:39.01 |
| | I need to make one change | 18:39.19 |
| | henrys: and I am going to have to test this with several cases to make sure it is OK | 18:39.39 |
| henrys | mvrhel_laptop:that isn't consistent with the other device parameters. take resolution the device provides a default and if you set -dHardware 600 the library parameter craziness sets it. the language have nothing to do with it. -r is just a convenience that the languages parse ... | 18:39.55 |
| | -dHWResolution I mean. | 18:40.13 |
| mvrhel_laptop | so you will keep your current call in the languages, but I will be doing a change in gsicc_init_device_profile_struct | 18:40.21 |
| | henrys: fair enough. but gsicc_init_device_profile_struct has to be called by someone at sometime to set the device to some default state. | 18:42.05 |
| | different devices will have different defaults | 18:42.22 |
| | with the resolution, all devices have the same default | 18:42.49 |
| | I suppose we could go through all the devices and change them so that the have their default explicitly set | 18:43.29 |
| henrys | you already do that in gsdparam.c, just like all the other parameters no? | 18:43.56 |
| | absent other information isn't the default simply a function of the number of components? | 18:46.29 |
| Robin_Watts | mvrhel_laptop: Sorry to stick my nose in, but all devices do NOT have the same default resolution. | 18:47.01 |
| | pdfwrite has a default of 720, display has 96 etc. | 18:47.13 |
| mvrhel_laptop | right | 18:47.16 |
| | ok | 18:47.18 |
| | bad example | 18:47.20 |
| henrys | Robin_Watts: oh thank you I missed that. | 18:47.51 |
| | mvrhel_laptop: my guess is simply deleting the calls fixes the problem, you've got the initialization in put_params and get_params. Of course reading gs code is treacherous. | 18:48.50 |
| mvrhel_laptop | henrys: that is what it is looking like in gsdparam.c but let me double check with a run | 18:49.26 |
| henrys | and understanding why the customer doesn't have a problem with pdf and ps is curious - I can look at that later. | 18:50.31 |
| bigjohntoday | odd i posted an issue i was having on bugs.ghostscript.com on pdf writer | 18:57.35 |
| | but now that i go and take a peak its not there... | 18:57.43 |
| henrys | bigjohntoday: maybe we fixed it stranger :-) | 18:58.45 |
| | things have happened | 18:58.54 |
| | mvrhel_laptop: the customers code is absolutely awful I never know what to do about that, you don't want to be harsh but they're likely to get in trouble if I can easily read the bugs. | 19:01.05 |
| bigjohntoday | henrys, i just put it up a few minutes ago, but if it was solved wouldn't i still see it with noes? | 19:01.42 |
| | notes* | 19:01.45 |
| henrys | bigjohntoday: you should have gotten an email confirmation if you used a real address, no? | 19:02.23 |
| bigjohntoday | yes i used a real address | 19:02.55 |
| | i don't have a confirmation email | 19:03.02 |
| | odd i must say | 19:03.52 |
| | its one of those nocurrentpoint erros | 19:04.00 |
| | i think i finally found it | 19:05.22 |
| henrys | yes! my bike has been shipped back from canada safely, there were some concerns: The driver crashed going up there and seemed a bit nutty. | 19:07.17 |
| mvrhel_laptop | wow | 19:07.32 |
| henrys | I'm from colorado my bike costs more than my car. | 19:08.22 |
| | bbiab | 19:22.49 |
| mvrhel_laptop | Robin_Watts: I pushed the polarity stuff | 19:23.39 |
| | please let me know if you have any problems with it | 19:24.10 |
| Robin_Watts | mvrhel_laptop: For the logs. Thanks. | 22:14.53 |
| | Forward 1 day (to 2013/09/04)>>> | |
IRC Logs