| <<<Back 1 day (to 2014/08/08) | 2014/08/09 |
amish | hi | 02:35.10 |
ghostbot | Welcome to #ghostscript, the channel for Ghostscript and MuPDF. If you have a question, please ask it, don't ask to ask it. Do be prepared to wait for a reply as devs will check the logs and reply when they come on line. | 02:35.10 |
amish | Have a real old version of mupdf 0.9 - am looking for the commit that fixes a rendering problem (it looks like its fixes in newer versions) | 02:42.42 |
| An example pdf is here https://github.com/amishpshah/test/blob/master/Pagesfromtxt9780544070657.pdf | 02:42.54 |
| mupdfDraw renders it like this - https://github.com/amishpshah/test/blob/master/pdf.png | 02:43.46 |
| Seems like the shadow is moved on top of the image | 02:45.08 |
robin_watts_mac | .- | 02:47.17 |
| amish: Hi. | 02:47.24 |
| Are you a customer of ours? I saw a support mail earlier about the same issue, I thnk. | 02:47.49 |
| I'm on holiday at the moment hence I haven't had a chance to look into it. | 02:49.05 |
| The simple answer is to use a newer version of MuPDF. | 02:49.23 |
| We've fixed SO much since 0.9 that really, upgrading is the right kind of thing. | 02:49.53 |
| And really, the API hasn't changed that much that it should be hard to update. | 02:50.20 |
| In order to find out which commit fixed it, we'd need to bisect it. | 02:52.13 |
| The guy who usually does this kind of stuff is Marcosw, and he's on holiday too at the moment. He has a set of mupdf binaries for each revision, going back in history to make it a much faster process. | 02:53.03 |
amish | hi robin, yeah, I'm a customer from Kno | 13:34.27 |
kens | amish robin is on vacation at the moment (Easter Island I think). | 13:34.51 |
| Henry has said he'll take a look at idenfiying the commit, but with such an old version of MuPDF you would be much beter off updating. | 13:35.22 |
amish | yeah, except we've made a bunch of changes to extract other information | 13:35.57 |
| from mupdf for these products | 13:36.11 |
kens | Well, that wil make life more difficult...... | 13:36.15 |
amish | I'm not too worried about the interface, just porting all this other code | 13:36.39 |
kens | Its possible that we may not be able to simply create a patch for a 4 year old version of the code, it may rely on other changes. | 13:36.54 |
| But we'll have a look anyway | 13:37.05 |
amish | With the back-to-school season around the corner | 13:37.07 |
| there is no way I'm going be able to get the time to do this | 13:38.03 |
| thanks | 13:38.06 |
| I know its a kind of longshot | 13:38.18 |
kens | If I get time after what I'm doing I'll look into it, but it likely to be Monday our time before I can get a MuPDF developer to look at it (I work on GS, not MuPDF) | 13:38.51 |
amish | FRankly, I know its a kind of longshot | 13:38.57 |
kens | Its worth a try, it shouldn't take me long to bisect the reopsitory | 13:39.16 |
| Its just a case of trying to figure out if I cna apply what I find :-) | 13:39.29 |
amish | and I'm impressed you are trying to find the patch for me | 13:39.31 |
| ha | 13:39.44 |
kens | I have to finish the bug I'm looking into first, just trying to figure out if I can fix this particular bit of silliness from a free user, or tell them to go away and do it differently..... | 13:40.18 |
amish | let me know if you need anything from me. I tried to cut the book down to the smallest possible case. But I'd be happy to send you the entire book, or other books where we see the issue as well. | 13:43.33 |
kens | From what I understand of the problem, a simple file should be OK | 13:43.53 |
| OK I came up with a way to do this stupid stuff from the free user, on to MuPDF where life is at least sane | 13:55.45 |
henrys | kens:oh thanks for doing that⦠hereâs a really nice tip for bisect that keeps you sane: http://stackoverflow.com/questions/15407075/how-could-i-use-git-bisect-to-find-the-first-good-commit when looking for a fix instead of a bug | 14:10.34 |
kens | henrys I may have to give up. I can't seem to build a MuPDF of that vintage | 14:10.57 |
| Still trying, but.... | 14:11.13 |
henrys | kens: oh the third party stuff changed to a sub module in that period I wonder how that will mess us up. | 14:12.09 |
kens | Its messing me up :-( | 14:12.20 |
| I can't seem to get the thirdparty stuff checked out | 14:12.35 |
| Any clues to how I use submodules ? | 14:13.02 |
henrys | kens: Iâm going swimming but Iâll look at it where you leave off⦠ârecursive on the website | 14:13.47 |
| ? | 14:13.48 |
kens | SOrry, the funny Mac quotes make that unreadable :-( | 14:14.12 |
| What about recursive on the website ? | 14:14.24 |
henrys | mupdf.com git clone ârecursive | 14:14.45 |
kens | Umm, does that work against a particular checkout ? | 14:15.10 |
| I have a feeling this is a can of worms..... | 14:15.21 |
henrys | kens:Iâll give it a go later but it may have to wait for tor. | 14:16.40 |
kens | I think that may be the case, I'm getting lost in that maze of twisty passages feeling..... | 14:17.00 |
henrys | kens: I didnât know I was making funnny quotes | 14:17.03 |
kens | You're on a Mac right ? I think it shows up in the irc logs too | 14:17.19 |
| Yes it does# | 14:17.29 |
henrys | kens: I thought I fixed that I wonder if it got reset | 14:17.56 |
| anyway bbiab have a nice weekend! | 14:18.16 |
kens | I'm sure you did, but its been doing it again for a while, all you Mac users write funny :-) | 14:18.16 |
| enjoy your swim | 14:18.22 |
kens | has already finished mine | 14:18.28 |
krt | hi, i am trying to get mupdf 1.5 to fdroid, an "appstore" for free/libre/opensource software. | 14:22.36 |
| is it safe to assume that versionCode for 1.6 will start with 60? | 14:22.55 |
| (this is the android version) | 14:23.14 |
kens | krt sorry but the MuPDF developers aren't here at the moment. I'm not able to answer the question but someone will read the logs on Monday | 14:23.29 |
robin_watts_mac | hi | 14:23.38 |
ghostbot | Welcome to #ghostscript, the channel for Ghostscript and MuPDF. If you have a question, please ask it, don't ask to ask it. Do be prepared to wait for a reply as devs will check the logs and reply when they come on line. | 14:23.38 |
kens | OK so I'm wrong :-( | 14:23.50 |
| How is/was Rapanui robin ? | 14:24.03 |
robin_watts_mac | just waking up. Let me check. | 14:24.04 |
| wet. | 14:24.11 |
kens | LOL | 14:24.15 |
robin_watts_mac | krt: versionCode? | 14:25.25 |
| Where are you finding that? | 14:25.31 |
| oh, you mean the android version? | 14:25.49 |
| No. | 14:25.49 |
| The android apk version numbers bear no resemblance to the app version. | 14:26.12 |
| mupdf/fitz/version.h is where you should be looking | 14:26.31 |
| All we will promise about the android apk version numbers is that they will be monotonically increasing with new releases we do. | 14:26.55 |
| Android releases do not strictly correspond to mupdf releases. | 14:27.17 |
krt | okay | 14:27.36 |
| will android releases be tagged like ios ones or should i refer to the default version tag (e.g. "1.5")? | 14:28.49 |
kens | robin_watts_mac : I'm trying to bisect a really old version of MuPDF for amich and having trouble. Probably something to do with the thirdparty libraries. I cna checkout an old version of MuPDF< but it won't build for me (on Windows). Any ideas or shall I poke Tor on Monday | 14:34.03 |
robin_watts_mac | kens: I would try 1.2 first, as I think that postdates the thirdparty change. | 14:45.25 |
| Third party changed from being a snapshot of dirs to being submodules around about 1.2 | 14:46.12 |
kens | Right. I don't suppose you know a way to identify a SHA for that ? | 14:46.41 |
robin_watts_mac | If you have to go back to before 1.2, then there are various thirdparty.zip's on mupdf.com I think. | 14:46.44 |
kens | If the old version was a dir snapshot I'm kind of surprised it doesn't build for me | 14:47.11 |
robin_watts_mac | http://mupdf.com/downloads/archive/ | 14:47.16 |
| the old version had thirdparty/jpeg-8 etc. | 14:47.36 |
| where the new version has thirdparty/jpeg | 14:47.43 |
kens | Ah right, that'll be quicker. I wonder if I have any of these installed already | 14:47.47 |
| Seems I have 0.6-0.9 installed and then 1.3 | 14:48.12 |
| LOL 0.6 is *really* wrong :-) | 14:49.26 |
| 0.9 is OK | 14:49.49 |
| So I'm already back before 1.2 | 14:50.03 |
| 0.8 is OK | 14:50.24 |
| 0.7 is wrong. | 14:50.46 |
robin_watts_mac | amish is using 0.9, I think | 14:51.09 |
kens | And I know it was after december 2010 (because that's what amish is using) | 14:51.14 |
| He says their last commit is December 2010 | 14:51.25 |
robin_watts_mac | I think pre-0.9 we didn't evn | 14:51.28 |
| en attempt transparentcy. | 14:51.34 |
kens | Well that would certainly explain the result | 14:51.49 |
| 0.8 works for me here and is dated March 2011 | 14:52.16 |
robin_watts_mac | I was sure they were newer than that. | 14:53.58 |
| oh, gawd. | 14:54.04 |
kens | I can't be certain, its the dat that they were installed on my PC | 14:54.19 |
robin_watts_mac | when did we introduce the context and error handling? | 14:54.30 |
kens | hasn't a clue | 14:54.45 |
| The dates on my binaries are self-consistent (0.7 predates 0.8 predates 0.9 etc) | 14:55.18 |
| They are 'sort of' 6 months apart | 14:55.43 |
robin_watts_mac | OK. 0.7 is when we first did transparency. | 14:55.51 |
kens | OK | 14:56.13 |
robin_watts_mac | August 2010. | 14:56.16 |
kens | Right, but this file definitely doesn't work with 0.7 | 14:56.30 |
| And Amish says it doesn't work for them with code from December 201 | 14:56.48 |
| 2010 | 14:56.52 |
| But 0.8 (March 2011) does work | 14:56.59 |
robin_watts_mac | Applying any patch post 1.0 is going to be very hard to apply. | 14:57.05 |
kens | So I really do need to bisect | 14:57.06 |
robin_watts_mac | So you need o bisect between 0.7 and 0.8. Not too bad. | 14:57.20 |
kens | Except that I can't get the December source to build under WIndows | 14:57.37 |
robin_watts_mac | Got to go - osrry! | 14:57.56 |
kens | NP enjoy :-) | 14:58.07 |
krt | robin_watts_mac: regarding android versioning, will the versionCode in the AndroidManifest.xml will be updated? The base version has been at 50 since 2013 | 15:03.20 |
jhabjan | hi, any idea where latest GSView source code can be found? Is http://git.ghostscript.com/?p=user/ghostgum/gsview.git;a=summary the latest or ? | 15:29.11 |
kens | jhabjan,pretty much yes,it hasn't been updated in ages. We are working on a new Windows viewer which works in a radically different way | 15:31.44 |
jhabjan | kens: ah, ok | 15:32.56 |
kens | NP | 15:33.11 |
tomty89 | is this even newer http://pages.cs.wisc.edu/~ghost/gsview/get50.htm :O | 15:33.36 |
kens | Is it not the same ? | 15:33.56 |
| Oh I see yes, it is newer | 15:34.09 |
| 5.0 is the latest there is | 15:34.16 |
jhabjan | tomty89: thanks | 15:34.19 |
kens | I wonder why we don't have 5.0 on our site..... | 15:34.37 |
tomty89 | it's windows only, i guess | 15:35.30 |
kens | Yes it is | 15:35.35 |
| The new one will be also | 15:35.42 |
jhabjan | 5.0. does not work for some pdf's | 15:36.27 |
kens | Hmm in what way ? | 15:36.37 |
jhabjan | does display the pdf at all, I assume it has to do with '(r) file pdfopen begin' | 15:37.54 |
| as I had a similar problem in Ghostscript.NET, and after changing that with '(r) file runpdfbegin' the pdf was displayed properly | 15:38.26 |
kens | Maybe. But its GS that does all the work for GSView. | 15:38.27 |
| TBH neither of you should be using that low level stuff :-) | 15:38.49 |
jhabjan | hm... as far as I know, gsview handles pdf opening via ps | 15:39.14 |
tomty89 | opdfread.ps? heh | 15:39.27 |
kens | It sends PostScript ot GS to run the PDF file in the GS PDF interpreter. WHich is the same as what you are doing, and the same way the GS itself works with PDF file. | 15:39.57 |
| tomty89 : opdfread.ps is the prolog for the ps2write device | 15:40.14 |
jhabjan | kens: yep | 15:40.20 |
kens | Point is, neither of you should really be using these PDF constrcuts. One day I would like to rejjig the PDF interpreter to make a defined API. | 15:40.59 |
robin_watts_mac | krt: We update the versionCode when we build | 15:41.23 |
jhabjan | kens: I get your point :-) but, that was the only way some thing could be accomplished.. | 15:41.40 |
kens | I imagine I'll get to it the same day Gatwick air traffic control get round to dealing with wild boars | 15:42.10 |
jhabjan | :) | 15:42.42 |
robin_watts_mac | kens: it was all sunshine when we had our very bumpy landing. | 15:47.04 |
| Then it was sunshine as we walked into the terminal. | 15:47.11 |
kens | And then it rained ? | 15:47.17 |
| Ther gods obviously don't like you | 15:47.24 |
robin_watts_mac | Then 5 minutes later when we walked out, the entire pacific ocean dropped onto our heads. | 15:47.27 |
| Rained biblically on thursday. | 15:47.45 |
kens | :-( | 15:47.45 |
robin_watts_mac | Rained on and off yesterday. | 15:47.55 |
kens | Presumably it'll stop when you leave | 15:47.58 |
robin_watts_mac | Forecast to rain a bit today. | 15:48.03 |
| forecast to clear up tomorrow as we leave. | 15:48.10 |
kens | THere yoiu are then | 15:48.17 |
| Maybe you'll get some pictures out the window..... | 15:48.29 |
robin_watts_mac | They have 1 flight a day in, and yesterdays was delayed by 6 hours or so. | 15:48.33 |
| TBH, if it's like yesterday, we'll be OK. | 15:48.48 |
kens | That's quite a lot | 15:48.48 |
robin_watts_mac | heading out now. bbl. Best of luck with the bisect. | 15:49.49 |
kens | Not getting anywhere, cna't build a thing | 15:49.59 |
| Have a good day though | 15:50.03 |
| henrys I'm unable to build old versions of MuPDF. I'm not certain why, but I keep getting compilation errors with openjpeg. So I'm giving up. From the discussion above you can see the fix was sometime between 4th December 2010 and 3rd March 2011 | 15:58.26 |
kens | heads off to cook, bye folks | 16:00.59 |
mvdan | robin_watts_mac: update the version code how? you haven't been updating it in the android manifest file | 16:27.39 |
| robin_watts_mac: also, what will the version code for 1.6 be? | 16:27.48 |
| since your published version code has been 50 for so long, we built 1.2 with the vercodes 50-52 long ago and are now stuck unable to update to 1.5 | 16:28.16 |
robin_watts_mac | mvdan: Whenever we decide to do an appstore release, we look at the google play developer console, and pick the next 4 integers, and use those values. | 21:13.48 |
mvdan | robin_watts_mac: and that is fine, but please commit the updates to the AndroidManifest.xml file | 21:14.11 |
| and if you tagged android releases like you do with ios, that'd be great too | 21:14.22 |
| robin_watts_mac: it would also help if you used version codes that bear resemblance to the version names | 21:17.39 |
| so e.g. version 1.6 could have version code 160 | 21:17.58 |
robin_watts_mac | mvdan: The problem is that I've done builds in the past, only to find a cock up while uploading to google play. | 21:18.13 |
| And I can't reupload another with the same version. | 21:18.22 |
| hence I've had to jump to the next set of 4. | 21:18.30 |
mvdan | I don't understand | 21:18.47 |
robin_watts_mac | I have to do builds for arm, arm7, mips, x86. | 21:19.07 |
mvdan | yes | 21:19.09 |
| and what is the problem? | 21:20.14 |
robin_watts_mac | and then I upload to google play. | 21:20.34 |
| but if I then realise there is a problem with one of them, I can't selectively delete it and just upload a new version. | 21:20.34 |
mvdan | you can do a bugfix release, but it has to be a newer version code | 21:21.08 |
robin_watts_mac | and the numerical ordering of arm/arm7/mips/x86 is important. | 21:21.21 |
| so I need to upload all 4 again. | 21:21.21 |
| right. | 21:21.25 |
mvdan | sure. | 21:21.27 |
| then if you had more digits of space you'd be safer | 21:21.36 |
robin_watts_mac | so I can't use 160 for 1.6, 170 for 1.7 etc. | 21:21.38 |
mvdan | but you can use 161/162/163/etc for 1.6, and so on for 1.7 etc | 21:21.56 |
robin_watts_mac | Not if we do more than one android release per mupdf release. | 21:22.23 |
mvdan | then throw in more zeros | 21:22.40 |
| 1601 et al | 21:22.47 |
robin_watts_mac | I'm not gonna do that. | 21:22.51 |
| We should tag versions | 21:23.04 |
| and I will try to do that in future. | 21:23.11 |
| And I will try to push the versionCode updates as I do that. | 21:23.27 |
mvdan | do whatever suits you with version codes | 21:23.30 |
| thanks! | 21:23.32 |
| one last thing | 21:23.39 |
| the reason I am asking about the pattern of vercodes is because we built 1.2 with the vercode 52 | 21:24.00 |
| so we right now can't build 1.5 | 21:24.05 |
| we could though build it with 53-55, but that would overlap with future releases | 21:24.22 |
| could you skip to e.g. 60 on your next release please? | 21:24.30 |
robin_watts_mac | I can't remember what version codes we're currently on on mupdf. | 21:24.37 |
| I think it's over 54. | 21:24.45 |
mvdan | I think it's 51-54 | 21:25.09 |
robin_watts_mac | just checking | 21:25.18 |
mvdan | anyway, if you could give us up to 59 we would be able to build & publish 1.5 on fdroid | 21:25.46 |
robin_watts_mac | Let me check what we are on now. | 21:26.04 |
mvdan | or if you plan to release a new version soon, we'll just use that. | 21:26.20 |
robin_watts_mac | Easter Island has seriously sucky internet. | 21:26.23 |
| We release twice a year. | 21:26.34 |
| I think in future we plan to put the apk's as built for google play on our download site. | 21:27.11 |
mvdan | I think they are already, but hidden somewhere | 21:27.30 |
| aren't they on http://mupdf.com/downloads/? | 21:28.11 |
| looks like i was wrong, 50-53 | 21:28.22 |
robin_watts_mac | ok, so those ones. | 21:32.35 |
| I will put a note into the docs to try to use 60+ next time. | 21:33.01 |
| How are the versions you build different to our versions? | 21:33.14 |
mvdan | we build from your source, so the apks are nearly the same | 21:34.03 |
robin_watts_mac | Is there a reason you can't just use our apks ? | 21:34.23 |
mvdan | yes, that we build from source | 21:34.32 |
robin_watts_mac | mvdan: http://twiki.ghostscript.com/do/view/MuPDF/AndroidReleases | 21:37.42 |
| That twiki page is our release instructions. | 21:38.09 |
| If you make an account and put a note on there, we can avoid overlapping version numbers. | 21:38.34 |
mvdan | any chance you could do it? I'd rather not have an account hanging there | 21:39.39 |
robin_watts_mac | mvdan: I have just added a request that the next one be >= 60 to that list. | 21:43.22 |
| but the point of you having an account is that you could update the request each time you do a build. | 21:43.36 |
mvdan | robin_watts_mac: many thanks! | 21:43.48 |
robin_watts_mac | np. | 21:43.59 |
mvdan | well, the point of you skipping a handful of codes is that we can catch up with you | 21:44.02 |
robin_watts_mac | ok. | 21:44.12 |
mvdan | and will be able to use the correct version codes starting at 1.6 | 21:44.14 |
robin_watts_mac | ok. Pop in here and talk to us if you ever have problems again. | 21:44.32 |
mvdan | robin_watts_mac: thanks! | 21:45.14 |
| Forward 1 day (to 2014/08/10)>>> | |