| <<<Back 1 day (to 2014/04/01) | 2014/04/02 |
Ailter | Good day. I'm trying to use mupdf on an adroid. I'm using the muPDFCore file provided, I create a new pdf from a stream and successfully read page count, but drawPage results on blank images. | 07:35.14 |
chrisl | Ailter: the mupdf devs probably won't be around for an hour or two yet | 07:36.40 |
Ailter | Snap. Thanks. | 07:37.02 |
ivanshmakov | Not quite on-topic, but: âFrom git://git.ghostscript.com/mupdf ! [rejected] master â master (non-fast-forward)â.âWhat was that?? | 08:47.50 |
kens | Sounds like you have changes in your local repository, so you can't rebase to the master (caveat, I'm bad at Git) | 08:48.41 |
| what Git command did you execute ? | 08:48.56 |
ivanshmakov | kens: $ git fetch master:master.âAnd Iâm pretty sure that thereâre no local changes in that repository, for I use it only for tracking upstream, and never commit anything there myself. | 08:50.04 |
kens | Hmm, we never normally use git fetch | 08:51.01 |
| Normally we'd use git pull --rebase | 08:51.11 |
| You'll have to wait fomr someone better versed in Git than me to answwer the question I'm afraid | 08:51.49 |
ivanshmakov | kens: Thatâs a --bare repository; it has no âworking copyâ for $ git pull to update. | 08:51.55 |
kens | ivanshmakov : ah well, you didn't say that :-) | 08:52.09 |
| Still though, it will need someone better at Git than me to answer your question | 08:52.24 |
ivanshmakov | kens: Could you please check if rev. c119c22d5bba is still on the master branch? | 08:52.43 |
kens | I can't, no, I don't have a working MuPDF repository at the moment | 08:53.00 |
kens | is a Ghostscript developer | 08:53.10 |
Jogux | ivanshmakov : you can browse repository here if that helps: http://git.ghostscript.com/?p=mupdf.git;a=summary | 08:54.14 |
ivanshmakov | kens: ACK, thanks. | 08:54.23 |
Robin_Watts | ivanshmakov: We did a force push on the mupdf repo about a week ago. | 08:58.51 |
ivanshmakov | Robin_Watts: Curiously, what was the reason? | 08:59.19 |
sebras | morning tor! | 09:02.18 |
Robin_Watts | ivanshmakov: We broke all our regression tests, so had to take extraordinary measures to fix it. | 09:10.40 |
ivanshmakov | Robin_Watts: ACK, thanks for the information. | 09:11.23 |
Fabio | Hi | 09:22.54 |
ghostbot | salut, fabio | 09:22.54 |
Fabio | I'm developing an Android app by using muPDF | 09:23.12 |
| I think I've found a bug about highlight annotations | 09:24.00 |
Robin_Watts | Fabio: Tell us more. | 09:25.02 |
Fabio | By adding an highlight annotation I use the internal native method: addMarkupAnnotation | 09:25.16 |
| the paramters are the page number, the quadPoints and the markup tye (highlight in my case) | 09:25.54 |
| unfortunately if I specify the quadpoints (PointF array) in the order: | 09:26.37 |
| 1) left bottom | 09:26.57 |
| 2) right bottom | 09:27.02 |
| 3) right top | 09:27.16 |
| left top | 09:27.19 |
| then I save muPDf is able to show the highlighting in the correct way | 09:27.49 |
| instead Adobe Reader and other PDF viewers are not showing a rectangle because they want the order: | 09:29.02 |
| 1) left bottom | 09:29.06 |
| 2) right bottom | 09:29.14 |
| 3) left top | 09:29.22 |
| 4) right top | 09:29.27 |
Robin_Watts | Are you usng the latest version? | 09:29.39 |
Fabio | I should be 1.3 | 09:31.50 |
Robin_Watts | I would try the git version. IIRC paulgardiner fixed something to do with the ordering of highlight points. | 09:32.43 |
Fabio | Do you mean that the bug correction has been made after 1.3 release? | 09:36.42 |
| We cloned it from git the 6th of September 2013 | 09:37.06 |
Robin_Watts | Fabio: Yes. | 09:39.07 |
| The 1.4 release is imminent | 09:39.12 |
Fabio | Ok. I'll try the last version then I'll report to you if the bug still exist | 09:41.09 |
| thanks a lot for your help | 09:41.18 |
sebras | tor8: finally! I greeted you about and hour ago, anticipating that you'd log in. :) | 09:54.32 |
tor8 | morning sebras | 09:54.42 |
Jogux | amusing. if you search for office in the ios appstore (uk at least), smartoffice appears in the results before microsoft office. | 10:41.05 |
pedro | maybe you get points for being the incumbent ;) | 10:43.24 |
tor8 | Robin_Watts: maybe we should add a -J (disable javascript) option to mupdf viewer? | 10:50.22 |
| for people who are obsessed with security | 10:50.36 |
| Jogux: pedro: I think you get penalized for being free... | 10:51.49 |
| s/you/apps/ | 10:52.26 |
pedro | tor8: interesting; does make some sense I guess | 10:56.04 |
tor8 | I've never seen free apps on the very top | 11:01.30 |
| paulgardiner: ping | 11:18.22 |
| (or should that be morning?) | 11:18.27 |
arse | hello ! | 11:19.13 |
| can I have some help about muPdf libs ? | 11:19.32 |
| please ^^ | 11:21.23 |
tor8 | arse: we can't help you (or even know if we can help you) if you don't actually ask your question... | 11:25.39 |
arse | Ok ! | 11:38.56 |
| I'm developping an application (a pdf viewer) and I have a problem : | 11:40.17 |
| After I have built the MUPDF and importing it to the Eclipse, every time I try to run it on emulator or any android device it produce this error in the logcat _see the attachment_ error is: java.lang.ExceptionInInitializerError. | 11:40.24 |
| Specially when a try to put a path to the asset : Uri uri = Uri.parse("file:///android_asset/PDF_Demo.pdf"); | 11:41.04 |
| Actually, when I tried to debug step by step, I reached to "MUPDFCore" class and in that line "System.loadLibrary("mupdf");" it give me the Error "java.lang.ExceptionInInitializerError" | 11:41.43 |
tor8 | arse: have you followed all the steps in platform/android/ReadMe.txt? | 11:46.52 |
arse | yes ! And It works when I import the project | 11:50.48 |
| But when I try to put it in my projet, I have this problem | 11:51.32 |
| The file chooser works but when I choose a pdf, it just crashes | 11:52.21 |
tor8 | well, the error is obviously about not finding the native code shared library (libmupdf.so) | 12:01.49 |
henrys | kens:odd that customer is using pxlcolor | 12:02.07 |
tor8 | so I'd start looking to why it can't find that library. | 12:02.28 |
kens | henrys, it seems peculiar to me, but they are an odd customer :-) | 12:02.30 |
arse | ok tor8 | 12:02.48 |
tor8 | arse: look for tutorials about writing your own native code libraries, and see where they differ from how your project is laid out | 12:03.10 |
| but if you don't have a libmupdf.so in your build, that's a big problem :) | 12:03.25 |
arse | mmh ok, but do you have a link to this tutorial | 12:03.40 |
| please* | 12:03.57 |
tor8 | arse: start reading here https://developer.android.com/tools/sdk/ndk/index.html | 12:04.54 |
arse | ok thanks ! | 12:05.07 |
| I've already dowload it, and use it | 12:06.19 |
| where must I find the file libmupdf ? | 12:06.42 |
Robin_Watts | libmupdf is built as a result of the ndk-build steps. | 12:07.46 |
arse | in the android folder ? | 12:07.53 |
paulgardiner | tor8: pong | 12:07.56 |
Robin_Watts | Step by step instructions can be found in platform/android/ReadMe.txt | 12:08.03 |
tor8 | paulgardiner: I've looked over your 3 iOS commits on paul/master | 12:08.19 |
| the airprint one is very short! it took me a while to figure out that you hadn't actually missed a file :) | 12:08.42 |
arse | I know it, but i would check if I arlready have the file or not | 12:08.48 |
tor8 | (oh, and there'sa tab/space error in there) | 12:08.52 |
| paulgardiner: what happens with that code if you try to print a non-pdf file? | 12:09.22 |
Robin_Watts | Jogux: Did you see my comments about your #include idea last night? | 12:09.25 |
tor8 | I'd imagine normal image files would work, but CBZ and XPS would probably fail with an error | 12:09.41 |
| Robin_Watts: are the four commits on robin/master that I reviewed yesterday good to go? | 12:10.07 |
Robin_Watts | yes. | 12:10.23 |
| I've pushed them I think. | 12:10.35 |
tor8 | Robin_Watts: fab. do you know if michael's win_desktop branch is supposed to land before release? | 12:10.46 |
Jogux | robin_watts : yes. I both agree with your comments and still want #include. but I have no intention of implementing it for now anyway :-) | 12:11.02 |
arse | tor8 : I've found 2 libmupdf in the android project generated. Where must I put them ? | 12:11.03 |
Robin_Watts | tor8: I don't know. | 12:11.09 |
tor8 | Robin_Watts: ah, yes, you have. my bad. | 12:11.10 |
Robin_Watts | Jogux: OK, sounds reasonable. | 12:11.23 |
paulgardiner | tor8: I was originally intending to disable that, but now I'm looking at changing to the use of images (or I would be if I hadn't had to have a tooth filled, taken my blood pressure and then been packing ready for tomorrow) :-) | 12:11.31 |
Robin_Watts | If you get prompted to code it, then presumably there would have been a reason :) | 12:11.35 |
Jogux | robin_watts : (the thought came up as the SO+ build is just the SO build + 2 defines) | 12:11.40 |
| the SO file is like 7 lines though, so I don't feel awful about the duplication due to not having #include. | 12:12.26 |
Robin_Watts | arse: Before we go any further, what are you planning to do with mupdf? | 12:13.13 |
tor8 | paulgardiner: alright. I'll leave that one out of the tagged release then, and you can wrap it up when you get back. where are you two off to? | 12:13.34 |
Robin_Watts | Southern spain. | 12:13.51 |
tor8 | just as spring is arriving! I almost envy you :) | 12:14.17 |
arse | I need to make a pdf viewer in my app. This app will contain some pdf, and the user will choose one of them, and read it. | 12:14.32 |
Robin_Watts | arse: OK. Will your app be released under the GPL ? | 12:14.50 |
arse | (sorry for my english, I'm french ^^) | 12:15.01 |
Robin_Watts | no problem. | 12:15.13 |
arse | What's the GPL ? | 12:15.22 |
Robin_Watts | You should google for GNU GPL and read up on it. | 12:15.36 |
arse | ok ! | 12:15.43 |
paulgardiner | tor8: sounds sensible. Actually, the airprint one might not cause any problems other than displaying an error for xps because there is a test for whether the printer can handle the file format. What's there probably works for jpg and some other things we can read. But still maybe best left out. Your call | 12:15.55 |
Robin_Watts | It's (one of) the software licenses that MuPDF is released under. | 12:15.57 |
| Essentially, if you use MuPDF under the GNU GPL, then you agree to release your app under the GNU GPL too. | 12:16.21 |
| Apart from anything else that means you must give *all* the sources away. | 12:16.36 |
tor8 | paulgardiner: yeah. I'd be wary of making a big headline about supporting print and it not working for everything, is all. | 12:17.10 |
paulgardiner | yeah | 12:17.24 |
chrisl | NOTE: GPL does not mean "giving your source away", but making it available - it's not public domain, you are still hold copyright | 12:18.00 |
| "you still hold the copyright" | 12:18.21 |
arse | My app will be not be released under GPL : it will be a commercial app | 12:20.07 |
| (thanks for your explications :) ) | 12:20.22 |
Jogux | it also doesn't stop you charging for the app; but the source must be available. | 12:20.25 |
Robin_Watts | arse: OK, so you cannot use MuPDF under the GNU GPL. | 12:21.55 |
| You *can* get a commercial license for it (from Artifex), but that will cost money. | 12:22.14 |
arse | Ok and how much it cost ? | 12:22.33 |
Robin_Watts | I can't give you a figure. | 12:22.45 |
arse | ok ! | 12:22.49 |
Robin_Watts | All our licenses are tailor made. | 12:22.57 |
arse | ok ! | 12:23.09 |
Robin_Watts | Contact sales at artifex.com and explain to Scott what you want to do. | 12:23.23 |
| He'll send you back a huge list of questions - just answer them as best you can (lots won't apply to you). Then he'll get you a quote. | 12:23.53 |
arse | Ok ! A question : So if i would like to do an app under GNU GPL, I cannot use MuPDF ? | 12:24.10 |
Robin_Watts | If you want to do an app under GNU GPL, then you can use MuPDF just fine. | 12:24.34 |
arse | Ok ! | 12:24.46 |
| I will contact Scott ! | 12:25.02 |
Robin_Watts | Fab. | 12:25.07 |
tor8 | Robin_Watts: paulgardiner: care to look over tor/master? especially the CHANGES update and see if I missed something important. | 12:25.34 |
arse | And about my frst problem, can you help me or Scott will give me the solution ? | 12:26.41 |
Robin_Watts | arse: Depending on the deal you take with scott you may or may not get support as part of the deal. | 12:28.20 |
arse | Ok ! | 12:28.28 |
| So see you soon ! | 12:28.34 |
| And thanks for all | 12:28.39 |
| Bye ! | 12:28.43 |
Robin_Watts | arse: No problems. | 12:28.43 |
| tor8: looks OK to me. | 12:29.23 |
| All the commits look OK to me. I can't really comment on the ios ones, but then as they are paul's originally, and you've OKd them I think we can count them as reviewed. | 12:30.02 |
tor8 | Robin_Watts: thanks. now to see if I can get an android sdk that works on one of my chromebooks so I can test the android app + gmail thing... | 12:31.07 |
Robin_Watts | tor8: let me see if I can test it here. | 12:31.30 |
Jogux | I'm syncing from points -> marcosw's garage again; I set a bwlimit this time so hopefully no issues but let me know if anyone sees any. (It turns out the dbserver image in the directory with the ATS vms was actually the non-ATS db server, and the ATS db server was in the folder with all the ops VMs. so long story short, now transferring the right one ;) ] | 12:34.54 |
norbertj | henrys: I'm ready to commit my jpeg fix, but have to wait until I'm at home. The company-firewall just shut me down this morning, so have to wait for them to fix it again. But at home I can do the commit. | 12:57.12 |
| henrys: git commit -a (that's the command I think?) | 12:57.31 |
| henrys: testing here showed no regressions. | 12:58.03 |
henrys | norbertj: that will work usually you do git add file1 file2 then git commit | 12:58.32 |
| norbertj: I stay away from -a to avoid committing unintended stuff. | 12:58.53 |
tor8 | henrys: norbertj: git commit -p is a nice alternative to -a that asks you for each diff hunk if you want it included or not | 12:59.19 |
norbertj | henrys: will try that one. | 12:59.44 |
Jogux | norbertyj: if you like/prefer guis, git gui or sourcetree are good. but there's a bit of a learning curve, moreso with sourcetree. | 13:01.09 |
tor8 | Jogux: I've never tried sourcetree, but I'm a diehard git-gui fan :) | 13:01.31 |
kens | uses git gui mostly | 13:01.49 |
Jogux | I switch between sourcetree and the command line. sourcetree makes some things hard or difficult :( | 13:02.27 |
| occasionally git gui too. | 13:02.44 |
kens | some thigns git gui can't do, or are decidely non-intuitive, that's when I hit the command line | 13:02.57 |
norbertj | Jogux: I'm fine with git commandline. But I'll give sourcetree a try. That works on Win7? Or Linux only? | 13:03.23 |
Jogux | sourcetree is much more complete than git gui, eg. switching branches, viewing history. | 13:03.29 |
| norbertj : sourcetree is windows & mac os | 13:03.48 |
kens | rarely views history, more an Italian driving style | 13:04.03 |
norbertj | Jogux: yes, just googled it. | 13:04.06 |
Jogux | did the linux -> mac os conversion a few years back. | 13:04.08 |
| iphone dev made me take on mac os, but I actually prefer it now. | 13:04.33 |
| kens : :) | 13:04.50 |
henrys | Jogux: make a gui client that doesn't support linux just seems wrong to me ;-( but it is a nice client. | 13:12.26 |
| git gui that is | 13:12.43 |
Jogux | henrys : yeah. it's weird. the windows and mac os versions of sourcetree don't even seem to have much code in common (the two behave differentely in quite a few cases) :-S | 13:15.12 |
| git gui works on all git platforms btw, apologies if I implied it was mac/win only :) | 13:15.31 |
henrys | Jogux: 2 language C# and Objective C - you think most of these things would be done in java ... | 13:21.19 |
Jogux | or C++, or mono, or... there are quite a few solutions other than writing from scratch on each platform! | 13:22.04 |
henrys | Jogux: yea strange choices | 13:23.38 |
spanners | I use SourceTree | 13:25.34 |
| It does occasionally go mad and gobble 1G+ of RAM though. | 13:25.49 |
Robin_Watts | I use command line for everything, except for when I want to remove hunks from a commit, when I use git gui. | 13:26.27 |
| Jogux is clearly using a git client with very wide windows. | 13:31.03 |
Jogux | robin_watts : I think, more accurately, I'm using a git client that supports wrapping :-) | 13:31.24 |
Robin_Watts | Well, basic git doesn't :( | 13:31.40 |
| git log for instance is much harder to read. | 13:32.35 |
Jogux | so it is. | 13:32.51 |
| sourcetree doesn't appear to have an option to make it wrap, but you can manually tell it to. how annoying. | 13:33.06 |
Robin_Watts | Jogu: Should appstore-ios perhaps be appstore-ios-sot2 ? Or is there only 1 ios appstore product? | 13:34.43 |
| sot2/appstore-ios ? | 13:34.53 |
Fabio | hi Robin | 13:35.32 |
Robin_Watts | hi | 13:35.38 |
Fabio | I downloaded the last version of muPDF | 13:35.51 |
| and I built it by ndk-build | 13:36.04 |
| unfortunately I have to give you good and bad news | 13:36.28 |
Robin_Watts | Jogux: Otherwise those commits all look good to me. | 13:37.19 |
| Fabio: OK... | 13:37.47 |
Fabio | 1) Now the highlight appereance is the same if we add highlight annotations by muPDF library and by Adobe Reader | 13:37.59 |
Jogux | robin_watts : I have a appstore-ios and an appstore-ios-smartoffice-plus | 13:39.23 |
| currently | 13:39.34 |
Fabio | however if I try to select the highlight rectangle in Adobe Reader it shows me that quadpoints are still in the wrong order ... when the highlight annotation is selected it should show a dashed rectangle | 13:39.36 |
| sorry I think we are crossing two conversations | 13:39.53 |
Robin_Watts | This is really one for paulgardiner. Paul, you here? | 13:39.55 |
Jogux | there's also an smartoffice for good; do we have an ongoing relationship with good and should be updating? | 13:39.59 |
Robin_Watts | Jogux: I am not sure. | 13:40.23 |
paulgardiner | Sort of. I'm in and out packing | 13:40.29 |
| Just reading back | 13:40.46 |
Jogux | robin_watts : it's down as a free app, so I presume we must receive money via good somehow. | 13:40.50 |
Robin_Watts | paulgardiner: Fabio reported a problem with the ordering of quadpoints in highlight annotations. | 13:40.58 |
Jogux | (as you need to pay good money to activate it) | 13:41.01 |
Robin_Watts | and I told him to try the latest version. | 13:41.12 |
| He has done so, and it's better than before, but there are still problems. | 13:41.31 |
Fabio | If I can send you a PDF file you can directly see my problem | 13:41.44 |
Robin_Watts | Jogux: Let me ask Miles. | 13:42.07 |
Jogux | robin_watts : if we do want to update it, I presume there must be some test logins or something that good can provide. | 13:42.44 |
| as otherwise I have no idea how we'd check it works | 13:43.05 |
| that was quite a diversion from "is that filename correct". | 13:43.32 |
paulgardiner | Fabio: okay, if you can upload it somewhere | 13:43.34 |
Jogux | robin_watts: would you prefer appstore-ios-so and appstore-ios-so+ ? | 13:43.49 |
Robin_Watts | That might be clearer? | 13:44.26 |
Fabio | OK Paul ... let me know how to do it ... can I upload a file by means of this IRC channel? ... I'm using the web interface | 13:44.55 |
Robin_Watts | Fabio: Go to bugs.ghostscript.com, and open a bug there. Mark it as being for MuPDF. | 13:45.20 |
| You can attach the file there. | 13:45.35 |
Fabio | ok | 13:46.06 |
| thanks a lot | 13:46.08 |
Robin_Watts | Fabio: Both paulgardiner and I are away from tomorrow until next wednesday. | 13:46.31 |
Fabio | ok ... In the meantime I'll open the bug ... then I'll be waiting answer directly on bugzilla | 13:47.47 |
Robin_Watts | pedro: Is it just me getting older, or are specs getting less and less clear. | 13:49.50 |
| pedro: The ECMA OOXML specs are just unreadable. | 13:50.21 |
kens | Try jpeg2k | 13:50.38 |
paulgardiner | Fabio: I don't completely understand the described symptoms | 13:52.07 |
| Also it would be useful to have a description of what problem you were seeing with the older version | 13:53.57 |
Robin_Watts | So all sizes in an ooxml document are given in 20ths of a point. Obviously! | 13:54.07 |
Jogux | examines the good build command. It supports he-il and ar-ar (which other builds don't), but doesn't support vi-vn or th-th. I have no idea if that's random, intentional, of what :-S | 13:57.02 |
pedro | the default SOT build on appstores probably needs a bit of thoguht too in terms of language support. Any good reason not to include soem CJK fonts for example? | 14:06.48 |
pedro | transposes too many characters :( | 14:07.18 |
Jogux | ped: -language=ja,jp,zh-cn etc /doesn't/ make it include cjk fonts?! | 14:08.13 |
| robin_watts : just pushed with fixed commit messages + that rename | 14:10.08 |
Robin_Watts | Fab. | 14:10.35 |
Jogux | no, ignore that, I've made a mess of it | 14:10.46 |
pedro | jogux: ddn't think the android build used those options -if it does already then cool :) | 14:11.11 |
Jogux | p> the last android appstore build was tgvbuild.py --android -define=ALIEN_ANDROID_LVL -define=UE2FILEVIEWER_THEME_TABLET_HIDPI -language=en-gb,ko-kr,ja-jp,zh-cn,zh-tw,vi-vn,th-th --dispman-ue2fileviewer -define=ALIEN_ANDROID_RELEASE app/simple | 14:11.40 |
pedro | ah, ok - hadn't seen that in the release spreadsheet | 14:12.15 |
pedro | sees it now | 14:13.19 |
Jogux | now you've made me look I may as well create a release config file for that too :-) | 14:13.40 |
ray_laptop | uh-oh. Looks like peeve[sd] are down. I'll have to head over to the office and find out what's up. | 14:14.28 |
| marcosw: I left the copy to the external USB drive running last night. I hope it wasn't a power out, because then I won't know how much it got correctly. | 14:15.32 |
| I timed it at about 16 to 18 hours (must be a pig slow drive) | 14:16.20 |
Fabio | I Robin and Paul | 14:25.20 |
| I've just reported the bug on bugzilla | 14:25.41 |
ray_laptop | Fabio: congratulations | 14:25.57 |
Fabio | I hope my explanation is clear :-) | 14:26.54 |
paulgardiner | Just looking | 14:27.08 |
kubilayrd | hey guys | 14:30.17 |
| what's up | 14:30.20 |
Robin_Watts | Fabio: So you're saying that the quadpoints are bad, but the appearance stream is good. | 14:30.36 |
kubilayrd | I was wondering if ghostscript has a width-only option like mupdf | 14:30.58 |
| equivalent to... mudraw -o out%d.png -w 1536 | 14:31.15 |
Robin_Watts | With mupdf that will set the width to the required value, and calculate h from it to keep the aspect ratio correct. | 14:31.56 |
| I'm not sure gs has the same facility. | 14:32.09 |
kens | You cna set the widht and height, but not just the width | 14:32.33 |
paulgardiner | Fabio: strange, the spec says "counterclockwise order" and has a diagram with pt3 being top right | 14:32.51 |
kubilayrd | How about dpi? Does -r300 always returns the same or similar resolutions? | 14:33.08 |
kens | its 300 dpi | 14:33.23 |
Fabio | Robin if you open the attachment 1 you can see that the highlighting is correctly rendered with both muPDF and Adobe Reader | 14:33.27 |
kubilayrd | for example two documents, 300x400 and 900x100, will -r100 return similar resolutions? | 14:33.48 |
| at least close to each other | 14:33.51 |
kens | No, it will return images at 100 dpi | 14:34.03 |
Fabio | however if you select it Adobe Reader is not correctly displaying a rectangle | 14:34.07 |
kubilayrd | kens: will it change resolutions? | 14:34.26 |
Fabio | maybe the problem is in Adobe Reader that is not following the specifications | 14:34.27 |
kens | If you ask for -r100 you get 100 dpi | 14:35.09 |
kubilayrd | kens: Can I convert two different pdfs to same resolution, at least close resolutions? | 14:36.12 |
kens | kubilayrd : you don't seem to be using 'resolution' correctly | 14:36.26 |
Fabio | you are right | 14:36.45 |
| I've re-seen now the Text Markup Annotation paragraph | 14:37.14 |
kubilayrd | kens: the problem is that I can't give width and height both, ratios differ among files | 14:37.16 |
Fabio | Adobe Reader is not following the specifications | 14:37.31 |
| let me test other cases | 14:39.06 |
Robin_Watts | kubilayrd: Use MuPDF :) | 14:39.25 |
tor8 | kubilayrd: do you mean size (as in how many pixels wide and tall) when you say resolution? | 14:39.27 |
| gs doesn't have the same automatic calculation of resolution to fit a given size that mupdf does | 14:39.45 |
Jogux | robin_watts : right, fixed version really pushed. and I added the android appstore config (but not tested that as I don't have the build environment setup yet) | 14:39.53 |
kubilayrd | Robin_Watts: I already am, just want to make sure about performence cost | 14:40.09 |
Jogux | robin_watts : I won't push the ios 7.1 stuff to golden yet; I'm not confident in it. | 14:40.26 |
kubilayrd | tor8: I mean a static width, like 1500 pixels and a height according to ratio | 14:40.34 |
tor8 | kubilayrd: mupdf -w300 -h300 will calculate the resolution automatically to fit the page dimensions (given in millimeters) to a specific pixel size | 14:40.40 |
kubilayrd | tor8: No way to do this in GS? | 14:41.07 |
tor8 | kubilayrd: with mupdf and gs you can give the resolution (in dots per inch) with the -r option | 14:41.10 |
| gs doesn't have the ability to compute the resolution automatically to fit a given page dimension to a given number of pixels | 14:41.37 |
| now, if you know the page size (and I think you can ask GS for the page size) you can do that computation yourself and then pass in the value with the -r option | 14:42.01 |
kubilayrd | tor8: Where? | 14:42.14 |
| tor8: I mean, to make calculation | 14:43.34 |
tor8 | if the page is 792 points wide (the usual metric in postscript/pdf), divide that by 72 to get 11 inches, then if you want to fit it to say 1500 pixels, divide 1500 pixels by 11 inches and there's the -r you need (136 in this example) | 14:43.47 |
kubilayrd | tor8: In shell? | 14:44.35 |
tor8 | kubilayrd: actually, it may be possible to do in gs with -dPDFFitPage and -dFIXEDMEDIA and -sPAPERSIZE or something like that. kens would know better. | 14:45.41 |
| but that doesn't preserve the aspect ratio IIRC | 14:46.04 |
kens | is a little busy at the moment | 14:47.57 |
| -dFitPage preserves the aspect ratio of the content, but fits to a specific size (widthxheight), unmarked areas are whiter | 14:48.26 |
tor8 | kubilayrd: if you use gs with the bbox device you can get the page sizes out for each page if you want to script it from the shell | 14:49.20 |
| gs -q -dBATCH -dNOPAUSE -sDEVICE=bbox adesso8.pdf prints the bounding boxes on stderr | 14:49.55 |
kens | pdf_info gives sizes | 14:49.55 |
kubilayrd | Or should I just use mupdf if it doesn't change the performance and quality? I still couldn't make sure | 14:50.24 |
tor8 | kubilayrd: it would probably be easier to just use mupdf :) | 14:51.33 |
kens | MuPDF probably has better performance anyway | 14:51.54 |
kubilayrd | tor8 kens: Does it work on its own? Even GS is not installed? | 14:52.18 |
kens | yes | 14:52.52 |
tor8 | yes, mudraw is a single binary, no extra dependencies program. | 14:53.34 |
kubilayrd | kens tor8: Okay guys, thank you. I will go with mupdf | 14:54.10 |
norbertj | Robin_Watts: do you know if there is something wrong with peeves. My ssh connection to peeves suddenly died this morning (at work), and now at home it too doesn't work. | 15:05.05 |
Robin_Watts | norbertj: I think ray said it was having problems. | 15:05.29 |
| I can't connect, it would seem. | 15:05.53 |
| indeed, if you look on the web dashboard, peeves and peeved are shown as down. | 15:06.18 |
norbertj | Robin_Watts: ah, thanks. Pinging casper works but peeves not. So I just have to wait until it is fixed. | 15:06.20 |
Robin_Watts | norbertj: Yes. | 15:06.27 |
| Have you got it so that you can connect from work to casper yet? | 15:06.36 |
norbertj | Robin_Watts: casper not yet. Have to make an official request for that. This monday it suddenly worked (FileZilla and PuTTY-ssh), so I could work from peeves (clusterpushing), but now I wanted to commit my fix, and the connection was lost :( | 15:08.14 |
Robin_Watts | norbertj: I would make the official request :) | 15:08.37 |
| peeves and peeved have poorer connectivity than casper does. | 15:08.51 |
norbertj | Robin_Watts: yes, and perhaps cancel the one I made this afternoon about the peeves-connection problems. | 15:09.08 |
Jogux | we casper doesn't run anything on port 443 atm right? I wonder if an sshd could be started on there and if norbertj's firewall may default to letting https connections through... | 15:10.06 |
| experience suggests that the sshd I run on 443 is accessable from behind most corporate firewalls :-) | 15:11.11 |
Robin_Watts | I would be reluctant to stop us being able to use 443 for https at some point in the future. | 15:12.03 |
Jogux | yeah, my VPS has multiple IPs for that reason. | 15:12.32 |
Robin_Watts | Especially given that norberts IT dept have let him get through to peeves, so they should let him get through to casper. | 15:12.44 |
Jogux | robin_watts : were you happy enough that my changes upto the ios7.1 stuff go to golden? | 15:27.33 |
Robin_Watts | Jogux: You've pushed them already, right? | 15:27.51 |
Jogux | I've only pushed to my private repos so far (I hope) | 15:28.07 |
Robin_Watts | oh, well, I was happy enough for it to be pushed to golden. | 15:28.49 |
Jogux | cool, ta. | 15:30.02 |
Robin_Watts | Jogux, pedro: Interesting. We have Layout_TextTransform's that can do stuff like "convert to upper case". | 15:37.42 |
| So small caps may be doable as an extension of that. | 15:37.57 |
pedro | Robin_Watts: they're trivial character conversions though, aren't thy? For small caps we'd have to be transforming glyphs I'd have thought? | 15:38.51 |
Robin_Watts | pedro: Yeah, probably need a different style or something. | 15:39.08 |
| spanners: Do you remember, when doing Font_calculateWidths, did it calculate the width for each char independently, or did it allow for kerning? | 15:50.21 |
| it allows for kerning. | 15:57.37 |
| bang goes that theory. | 15:57.41 |
| And using the SwissA font in Libre office doesn't change the rendering either. | 16:13.34 |
spanners | kerning's optional, but usually used | 16:20.12 |
Robin_Watts | We are breaking a line a word earlier. | 16:21.31 |
spanners | I had a prototype which allowed small caps to be dynamically generated int the font engine. | 16:21.41 |
Robin_Watts | because our measurement of the line shows that including the last char would overflow by 672. | 16:22.05 |
| (that's 672 65536ths of an inch) | 16:22.16 |
spanners | SwissA vs ? | 16:22.44 |
| Arial? | 16:22.49 |
Robin_Watts | SwissA vs Arial. | 16:22.52 |
| And the metrics for SwissA now exactly match Arial, kerning and all. | 16:23.11 |
spanners | What about the device metrics? Some stuff uses them instead of the proper/true metrics. | 16:23.52 |
Robin_Watts | spanners: Bear with my ignorance here. The device metrics = the tables of 'pre scaled' metrics held in some fonts? | 16:24.34 |
spanners | that's right | 16:24.40 |
Robin_Watts | Those aren't in ef1's at all, right? | 16:25.12 |
spanners | oh, hmm | 16:25.46 |
Robin_Watts | Let me see if such tables exist in the swissA.ttf | 16:25.49 |
spanners | In any case, no they're not in ef1s | 16:26.30 |
Robin_Watts | not being in ef1s shouldn't affect the libre office test. | 16:26.54 |
kens | Goodnight all | 16:27.09 |
Robin_Watts | but I didn't update the tables, so it's possible that the swissA.ttf has the old values in. | 16:27.16 |
spanners | I suspect I'm missing some data here | 16:27.40 |
| Libre Office renders the same for Arial and SwissA - right? | 16:27.56 |
Robin_Watts | yes. | 16:28.35 |
| No hdmx or ltsh tables in swissa.ttf | 16:31.22 |
spanners | ok | 16:31.32 |
| Have you run PSO with Arial vs SwissA? | 16:31.53 |
Robin_Watts | spanners: I have not. Let me try that. | 16:32.42 |
norbertj | Robin_Watts: at home I'm now on casper (using PuTTY-ssh). | 16:32.54 |
Robin_Watts | norbertj: Give me a couple of minutes, and I'll be right with you. | 16:33.14 |
| spanners: Copying arial.ttf as swarr.ef1 and rerunning gives identical results. | 16:35.56 |
spanners | so it could be epage's interpretation of the font data | 16:36.40 |
Robin_Watts | yeah. | 16:36.48 |
spanners | perhaps a scaling issue | 16:36.50 |
Robin_Watts | perhaps. | 16:37.00 |
spanners | though it does use 2.14 units to minimise any scaling artifacts | 16:37.24 |
Robin_Watts | I might try a quick test with freetype to see if that gives equivalent results. | 16:37.35 |
spanners | sensible | 16:37.49 |
Robin_Watts | spanners: Thanks. | 16:37.52 |
spanners | np | 16:37.55 |
Robin_Watts | I think you're right about SmallCaps being best done in the font engine. | 16:38.08 |
| OK, norbertj, I am with you now. | 16:38.43 |
spanners | I've got a screenshot of it working somewhere. :-) | 16:38.52 |
Robin_Watts | spanners: Would your code be on a picsel server somewhere? :) | 16:39.12 |
spanners | If I have it, it'll be in storage right now. :-| | 16:39.49 |
| (Lots of stuff in a lockup while we move home) | 16:40.05 |
Jogux | s> might it have been in your central home dir? | 16:40.25 |
Robin_Watts | norbertj: Inside your windows checkout... git remote add cluster regression@ghostscript.com:/home/regression/cluster/gitbridge/ghostpdl | 16:40.37 |
spanners | feigns memory loss until a brown envelope is produced | 16:41.47 |
| seriously, it might be in a backup tar on the home server box | 16:42.15 |
| but i'll need to look later on | 16:42.28 |
| tbh, i think it was a crappy quality patch anyway | 16:42.39 |
| but at least illustrated the right approach | 16:42.54 |
spanners | scoots | 16:43.08 |
norbertj | Robin_Watts: done | 16:43.27 |
Robin_Watts | spanners: Thanks. | 16:43.29 |
| norbertj: OK, now we need to add the cluster key to your local machine I think. Just trying to remember :) | 16:43.48 |
| norbertj: OK, download /home/robin/privatekey.ppk from casper to your local windows box. | 16:45.27 |
| actually, hold on that. | 16:47.11 |
| norbertj: OK, can you check that you can ssh to regression@ghostscript.com: using your existing key ? | 16:53.12 |
norbertj | Robin_Watts: I'm missing the public_key | 16:54.30 |
Robin_Watts | norbertj: Sorry, telling you to download privatekey.ppk was a mistake. Just delete that. | 16:54.58 |
| Use the same private key you've been using to connect to casper anyway to get you in to regression@ghostscript.com: | 16:55.33 |
| norbertj: Presumably you are running Pageant on your windows box? | 16:55.47 |
| If not, do so. Arrange for Pageant to run at startup. Drop your key into there. | 16:57.24 |
| That way you can just do: plink ghostscript.com: and it will automatically use the right key. | 16:57.57 |
norbertj | Robin_Watts: ok, got the key for casper dropped in (on my home pc) and now from a git-bash: ssh regression@ghostscript.com ? | 16:59.08 |
Robin_Watts | Personally, I use plink rather than ssh. | 17:00.03 |
| because the ssh that comes with msys git has some problems (or did when I last looked) | 17:00.25 |
| But if you have it working with ssh, we can go with that. | 17:00.43 |
norbertj | Robin_Watts: my putty-installer did not contain the plink. Have to get it. | 17:01.18 |
| Robin_Watts: it did, it was just not in the start-menu. | 17:02.18 |
Robin_Watts | Does plink regression@ghostscript.com: work? | 17:03.06 |
norbertj | Robin_Watts: yes, | 17:06.19 |
Robin_Watts | norbertj: OK. | 17:06.51 |
| So the first thing to do now is to copy gs/toolbin/localcluster/gitpush.sh to be somewhere on your path. | 17:07.19 |
| Now, if memory serves, your windows username was njj or something? not "norbert" anyway, right ? | 17:08.04 |
| You need to arrange that your MSYS git shells start up with GIT_SSH=plink.exe and CLUSTER_USER=norbert | 17:09.42 |
norbertj | Robin_Watts: njj at work, norbert at home | 17:10.13 |
Robin_Watts | The easiest way to do that is to edit (or create) a .bashrc in C:\Users\njj | 17:10.14 |
| ah, ok. | 17:10.29 |
norbertj | Robin_Watts: done. | 17:12.20 |
Robin_Watts | Then: git config alias.cluster \!gitpush.sh | 17:12.47 |
| Once you've done that, you should be able to do: "git cluster" and it should behave pretty much exactly like "clusterpush.pl" does on peeves. | 17:13.54 |
| The one difference is that git cluster will not send newly created files up to the cluster to be tested until git knows about them. | 17:14.38 |
| So if you ever add new files, you need to remember to git add them first. | 17:14.50 |
norbertj | Robin_Watts: echo $CLUSTER_USER = norbert but still "Unable to determine username; please set CLUSTER_USER | 17:18.33 |
Robin_Watts | how odd. | 17:20.11 |
Jogux | just to check: you're running both commands in the same window? | 17:20.45 |
norbertj | yes | 17:21.49 |
Jogux | oh, try export CLUSTER_USER | 17:22.47 |
chrisl_r500 | Yes, bash needs the "export" | 17:23.32 |
norbertj | Jogux: that did the trick, I think. It's doing something. | 17:26.23 |
Jogux | cool :-) | 17:26.40 |
norbertj | Robin_Watts: git cluster seemed to do something (made a clusterstamp?? which is gone again?) and is quiet now. | 17:29.59 |
Robin_Watts | norbertj: Right. | 17:30.13 |
| If you look on the web dashboard your job is running. | 17:30.24 |
| Essentially, I: git stash, make 2 files (clusterdatestamp and clustercmd), git stash, git push that stash, git stash drop, git stash pop | 17:31.12 |
| so the net effect on your local machine is that nothing changes. | 17:31.22 |
norbertj | Robin_Watts: yes, and how do I add parameters (like pcl, xps or lowres) | 17:31.46 |
Robin_Watts | but the current state is pushed up to a repo on casper. | 17:31.47 |
| That repo watches for stuff coming in, and clusterpushes it for you. | 17:32.04 |
| git cluster gs | 17:32.09 |
norbertj | Robin_Watts: ok, I'll see now if I can get my jpeg fix in. | 17:32.37 |
Robin_Watts | Oh, I forgot to say. You can: git cluster abort (or clusterpush.pl abort) to stop jobs. | 17:32.38 |
Jogux | robin_watts : if you have time, there's three trivial build fixes + a crash fix in my git tree | 17:32.45 |
Robin_Watts | Jogux: Sure. | 17:32.51 |
norbertj | git cluster abor | 17:32.57 |
Jogux | robin_watts : is that policy that pete/I should always ask you or Paul for reviews? | 17:33.05 |
Robin_Watts | If we're about, ask us. | 17:33.19 |
Jogux | (or that pete/I can do each others; or can do each others if you're not available) | 17:33.20 |
| right | 17:33.23 |
Robin_Watts | If not, I think you can review each others :) | 17:33.33 |
Jogux | :) | 17:33.38 |
| next policy question: I have a bug that says "x crashes". It now doesn't crash, but the rendering is nowhere near msword (I think due to unsupported features). Do I close the bug? If I do, should I open another one? | 17:34.24 |
| robin_watts : continue to ignore the ios 7.1 stuff btw :-) | 17:37.08 |
Robin_Watts | Jogux: I suspect we have bugs open for most of the unsupported features at the moment. | 17:38.28 |
| Unless it shows something hugely different to everything else, close it. | 17:38.54 |
| Maybe if it's a good example of a particular rendering bug, add it as an example on one of the others? | 17:39.16 |
Jogux | robin_watts : there's 21 open bugs :-) none of which mention charts; so I'm tempted to close, and open one about charts. | 17:39.19 |
Robin_Watts | Ah, well, that sounds good. | 17:39.30 |
Jogux | there seem to be about 4 billion chart types; fixing that is going to be tedious :-S | 17:40.16 |
Robin_Watts | Jogu: LGTM | 17:44.11 |
Jogux | ta | 17:44.16 |
| is there an easy way to push just some commits to golden? | 17:44.25 |
Robin_Watts | git push golden hash:master | 17:44.33 |
Jogux | awesome | 17:44.45 |
| pushed :) | 17:45.27 |
Robin_Watts | How is ATS going? | 17:45.53 |
Jogux | I have the right db server now (hopefully); waiting on mace trying to boot it | 17:46.14 |
| or more concisely, "slowly" :( | 17:46.21 |
tor8 | mvrhel_laptop: this new universal windows app stuff looks interesting | 17:47.31 |
mvrhel_laptop | tor8: where is this? | 17:47.53 |
Robin_Watts | Was the non-ATS db server the bugzilla one ? | 17:48.02 |
Jogux | robin_watts : yes. | 17:48.08 |
tor8 | http://wmpoweruser.com/microsoft-introduces-universal-windows-apps/ | 17:48.10 |
mvrhel_laptop | oh for the hone | 17:48.12 |
| phone | 17:48.14 |
Jogux | robin_watts : so will be able to boot intranet at some point. | 17:48.20 |
Robin_Watts | cool. | 17:48.25 |
tor8 | mvrhel_laptop: looks to be phone, tablet and win8 tiles | 17:48.33 |
mvrhel_laptop | finally.... | 17:48.38 |
Robin_Watts | Was spanners "thomas" at picsel? I remember him as being dpt, but i can't see a dpt homedir. | 17:48.58 |
tor8 | mvrhel_laptop: is there something you want to get in before I stamp the 1.4 release? the win_desktop branch has a lot of stuff that hasn't been merged yet. | 17:50.09 |
mvrhel_laptop | tor8: I will get that stuff merged later. | 17:50.48 |
tor8 | mvrhel_laptop: alright, just wanted to check. | 17:51.04 |
mvrhel_laptop | beating on a dumb ui sluggishness problem with my text selection stuff.. driving me crazy right now | 17:51.23 |
| brb. need to reboot | 18:11.27 |
Jogux | robin_watts : I think he was dpt. I wonder if it got archived off. | 18:15.15 |
| robin_watts : there's a 253G image of the homeserver too irc; not sure what's on that vs the homeserver dump on the NAS. | 18:16.09 |
norbertj | Robin_Watts: so I did a "git commit -p" (this is to my local-master), | 18:24.13 |
| Robin_Watts: how do I get it in the remote-master ("git push" ?) | 18:24.33 |
Robin_Watts | norbertj: You should be able to push to the golden repo. | 18:25.26 |
norbertj | Robin_Watts: git push golden hash:master ? | 18:26.04 |
Robin_Watts | We are trying to have a policy that (as far as possible) commits should be reviewed. | 18:26.33 |
| So it's probably best if you push your commit to your personal repo on casper, and then one of us looks it over. | 18:26.57 |
norbertj | Robin_Watts: offcourse. | 18:26.57 |
Robin_Watts | Then the reviewer can push it. | 18:27.11 |
| but in theory you can just push. | 18:27.39 |
norbertj | Robin_Watts: so how to push to casper? I have no repo there yet (no publickey) | 18:30.32 |
Robin_Watts | norbertj: You can ssh to casper, right? | 18:31.00 |
| so ssh to casper. Then do: | 18:31.10 |
norbertj | Robin_Watts: yes, I'm on it now. | 18:31.13 |
Robin_Watts | mkdir repos | 18:31.17 |
| cd repos | 18:31.21 |
norbertj | Robin_Watts: did that, but the git clone complained about the Permission denied (publickey) | 18:31.51 |
Robin_Watts | git clone --bare /home/git/ghostpdl.git ghostpdl.git | 18:32.07 |
norbertj | Robin_Watts: working ... | 18:32.28 |
| Robin_Watts: done. | 18:34.57 |
Robin_Watts | norbertj: OK, so on your local windows PC... | 18:35.18 |
| git remote rm origin | 18:35.26 |
| git remote add origin norbert@ghostscript.com:/home/norbert/repos/ghostpdl.git | 18:35.48 |
norbertj | Robin_Watts: done. | 18:37.19 |
Jogux | would I be correct to assume I should use US English in source code? | 18:38.54 |
| (having a color vs colour moment for a variable name here) | 18:39.12 |
Robin_Watts | norbertj: Now "git push origin" | 18:39.18 |
| Jogux: Smart office is UK english throughout, I believe. | 18:39.35 |
| We should probably stay consistent for now. | 18:39.42 |
| or rather "proper english" :) | 18:40.02 |
Jogux | robin_watts : apart from styling where we followed the (US) standard names :-S | 18:40.06 |
| hehe | 18:40.06 |
Robin_Watts | Jogux: Hmm. OK, so follow the historic trends :) | 18:40.40 |
norbertj | Robin_Watts: have to do a 'git pull' first. | 18:41.00 |
Robin_Watts | git pull --rebase | 18:41.08 |
| otherwise it will do a merge, and we don't want that. | 18:41.36 |
norbertj | Robin_Watts: I did a "git pull" followed by a "git pull --rebase" , that does it? | 18:46.42 |
Robin_Watts | norbertj: I fear not. | 18:46.54 |
norbertj | Robin_Watts: so how to undo? | 18:47.18 |
Jogux | it doesn't; git rebase origin/master probably | 18:47.19 |
| possibly should be golden/master, or it may not matter in this case. | 18:47.55 |
Robin_Watts | To be safe, let's do: git branch safety | 18:48.07 |
Jogux | likes to life dangerously. (almost, I have time machine backing up every hour :-) ) | 18:48.33 |
Robin_Watts | then: git stash (hopefully shouldn't need to do anything) | 18:48.41 |
| git reset --hard HEAD~1 | 18:48.53 |
| Then: git pull --rebase | 18:49.11 |
Jogux | robin_watts : one more trivial crash fix if you have time. | 18:50.11 |
| (in my private repo) | 18:50.17 |
Robin_Watts | I'd also strongly recommend: git config alias.logg "log --graph --oneline --decorate --all" | 18:50.19 |
| Jogux: I live more dangerously when working on local git repos where I can see what's going on :) | 18:51.30 |
Jogux | :-) | 18:51.35 |
Robin_Watts | norbertj: How goes it ? | 18:54.08 |
norbertj | Robin_Watts: tried "git push origin" again, but it still complains: | 18:54.17 |
Robin_Watts | norbertj: What does it say? Not a fast forward? | 18:54.41 |
norbertj | Robin_Watts:! [rejected] master -> master (non-fast-forward) | 18:55.05 |
Robin_Watts | git push -f origin master | 18:55.18 |
norbertj | Robin_Watts: that did it, (what did it do?) | 18:56.14 |
Robin_Watts | norbertj: Hmm. Doesn't seem to have done it. | 18:56.58 |
| git remote -v and paste the output please? | 18:57.09 |
norbertj | cluster regression@ghostscript.com:/home/regression/cluster/gitbridge/ghostpdl ( fetch) cluster regression@ghostscript.com:/home/regression/cluster/gitbridge/ghostpdl ( push) origin norbert@ghostscript.com:/home/norbert/repos/ghostpdl.git (fetch) origin norbert@ghostscript.com:/home/norbert/repos/ghostpdl.git (push) | 18:57.43 |
Robin_Watts | norbertj: git logg HEAD~5 | 18:58.04 |
| norbertj: I can't see any sign of your fix in the remote. | 19:00.06 |
| perhaps you were up to date when you pulled? | 19:00.20 |
norbertj | Robin_Watts: probably, but why did it complain? | 19:00.41 |
| Robin_Watts: emailed the 'git logg HEAD~5' output | 19:01.32 |
Robin_Watts | norbertj: OK. git reset --hard safety | 19:02.35 |
norbertj | Robin_Watts; the remote is on casper:~/repos/ghostpdl.git ? | 19:02.36 |
Robin_Watts | then: git push origin master | 19:02.54 |
norbertj | Robin_Watts: njj[ghostpdl](master): git reset --hard safety HEAD is now at 9c06c7d modified: gs/base/sdctd.c njj[ghostpdl](master): git push origin master Counting objects: 13, done. Delta compression using up to 4 threads. Compressing objects: 100% (7/7), done. Writing objects: 100% (7/7), 760 bytes, done. Total 7 (delta 6), reused 0 (delta 0) To norbert@ghostscript.com:/home/norbert/repos/ghostpdl.git c74e146..9c06c7d | 19:04.14 |
Robin_Watts | Ok, I see the commit in the remote repo. | 19:05.09 |
| http://git.ghostscript.com/?p=user/norbert/ghostpdl.git;a=summary | 19:05.18 |
| The commit looks OK to me, but the commit message needs improvement. | 19:05.50 |
| I'd do it for you, but I have to go pack. Am going on holiday tomorrow morning. | 19:06.09 |
| If I get a chance I will try and tweak it later. Sorry! | 19:06.32 |
norbertj | Robin_Watts: I will remember next time I make a commit to also add a message, but 'git commit -p' did not ask for it. | 19:07.01 |
| Robin_Watts: henrys is also online? Perhaps he can fix it. | 19:08.53 |
henrys | norbertj: oh something went bad with the commit message? | 19:10.04 |
spanners | Robin_Watts: I was dpt. I very likely emptied my homedir when I exited. | 19:10.21 |
| https://dl.dropboxusercontent.com/u/2347900/poxel/font/smallcaps.png | 19:10.25 |
| I can't see the patch for that locally. | 19:10.41 |
norbertj | henrys: yes, I did a 'git commit -p' , but that did not ask for a message. | 19:10.45 |
henrys | norbertj: funny I haven't seen mail for the commit | 19:11.06 |
Robin_Watts | norbertj: This is why "git commit -p" is a bad idea. | 19:12.00 |
norbertj | henrys: see 12 lines up http://git.ghost.... | 19:12.26 |
Robin_Watts | You should: "git add filename" for each file you want in the commit. | 19:12.36 |
| or "git add -u" to add all changed files. | 19:12.54 |
| Then "git commit" will fork an editor for a message. | 19:13.05 |
| henrys: the commit is on norberts personal repo. | 19:13.36 |
henrys | Robin_Watts: oh so no problem | 19:13.55 |
| Robin_Watts: well at least it isn't on master ... | 19:14.24 |
Jogux | norbertj: if you want to tweak the commit message for the most recent commit only, it'd be git commit --amend, then git push --force origin | 19:17.02 |
Robin_Watts | norbertj: If you mail me a commit message, I'll try and fix it later. | 19:18.08 |
spanners | Robin_Watts: I found my old homedir archive, untarred, grepped, no sign of that patch. | 19:21.38 |
Robin_Watts | spanners: Would you be interested in some contract work to recreate it? :) | 19:22.57 |
norbertj | Robin_Watts: henrys: just adjusted the commit-message. | 19:23.14 |
Robin_Watts | (not my decision, but I can mention it to people if you're interested :) ) | 19:23.27 |
henrys | norbertj: you've added vcproj/readme? | 19:25.23 |
henrys | really doesn't like this add -u or commit -a business⦠say what files you want to add individually and then commit | 19:26.40 |
norbertj | not intentionally. | 19:27.11 |
Robin_Watts | henrys: git add -u, then git diff --cached to check that the set is right works for me. | 19:27.43 |
spanners | Robin_Watts: possibly | 19:28.22 |
henrys | norbertj: do you want me to commit the patch for you or I can help you through the git. Let me know. | 19:29.59 |
| bbiam | 19:30.23 |
norbertj | henrys: can you remove the vcproj/readme ? | 19:30.25 |
Robin_Watts | norbertj: If you want to do it, run "git gui". | 19:30.59 |
| This will open a window with various panes in. | 19:31.10 |
| In the bottom right one, click "amend last commit" | 19:31.21 |
| Then everything will change to show the last commit. | 19:31.40 |
| The bottom left pane will show the list of files that changed in that commit. | 19:31.49 |
| Click on the icon for the vcproj/readme to get rid of it. | 19:32.05 |
| Then click the 'commit' button. | 19:32.12 |
| Then: close git gui | 19:32.24 |
| Then: git push -f origin master | 19:32.31 |
| The -f flag to push means "force". i.e. your commit will be pushed onto the remote repo, even though it's not just adding new commits on the end. | 19:33.09 |
| You need to specify -f as otherwise git is afraid you might accidently be losing work. | 19:33.39 |
norbertj | Robin_Watts: if I click Stage Changes X-icon then it moves (with ? icon) to Unstaged Changes | 19:37.22 |
Robin_Watts | perfect. | 19:37.34 |
| it's telling you that you have changes that you have not "staged" for commit. | 19:37.58 |
norbertj | but then pushing commit :No changes to commit. You must state at least 1 file before you can commit. | 19:38.29 |
Robin_Watts | norbertj: I have to go, sorry! | 19:41.30 |
norbertj | Robin_Watts: ok. | 19:41.40 |
| henrys: I think the commit is now ok. | 20:04.58 |
henrys | norbertj: okay we usually like to reference the bug number in the commit, then you can check it in right on master. | 20:07.38 |
norbertj | henrys: how do I push to master? git push master ? | 20:10.05 |
henrys | norbertj: no you rebase to your local master check that out and push it. If you're not familiar with git this is all hell, don't let anyone tell you differently. | 20:13.07 |
| norberj:http://git-scm.com/book/ch3-6.html | 20:15.25 |
Jogux | assuming no one has committed to the master git since norbert rebased an hour or so back isn't it just "git push golden" ? | 20:16.04 |
henrys | Jogux: I thought I just saw a commit | 20:16.36 |
Jogux | henrys : d'oh. yeah. actually norbert didn't rebase earlier in the end anyway it seems :( | 20:17.43 |
| Robin talked him through it but it seemed to fail somewhere | 20:18.37 |
henrys | Jogux: I always get things loaded on my local master than rebase than push. | 20:18.37 |
Jogux | yeah, that is what Robin tried to talk him through | 20:18.49 |
henrys | norbertj: probably take you less time to check out the local master branch and reapply your patch since you have to redo the commit message anyway. | 20:20.19 |
| norbertj: bet you'll be submitting your patches for us to commit henceforth ;-) that's what I would do. | 20:22.14 |
norbertj | henrys: so local master branch is the same as trunk-master + changes to pximage.c/sdctd.c . That is commited and then pushed to my repos/ghostpdl.git on casper. | 20:22.47 |
| That is then what I see in http://git.ghostscript.com/?p=user/norbert/ghostpdl.git;a=summary | 20:23.11 |
Jogux | norbert> your respository is missing 8 or so commits compared to http://git.ghostscript.com/?p=ghostpdl.git;a=summary | 20:24.01 |
henrys | norbertj: we are sort of starting over. git checkout master; git rebase; apply patches, git add files, git commit files - correct log message git push | 20:25.03 |
norbertj | henrys: do I have to make a new local master? | 20:27.10 |
henrys | argh I meant git checkout master ; git pull --rebase | 20:27.12 |
| that gets you square with the remote master. | 20:27.30 |
henrys | assume you didn't put these changes on master but maybe you did. | 20:29.08 |
| norbertj: I have to head out for a bit. no big deal if you don't get it in, I can commit it when I get back, just send me email. | 20:35.25 |
norbertj | henrys: sent you a mail. | 20:53.07 |
Jogux | pedro_: I presume Robin/Paul won't want to be reviewing things now :-) there are two bug fixes in my private repo when you have time (ignore the 2 ios changes at the top, they're still a WIP) | 21:47.16 |
| Robin_Watts : paulgardiner : enjoy the holiday! | 21:47.34 |
| time for me to relax. | 21:47.55 |
pedro_ | jogux: cool - one in mine too (dev/sot) | 21:50.37 |
| Robin_Watts & paulgardiner: enjoy :) | 21:51.05 |
Jogux | henrys : Miles mentioned that he believes we're doing releases that replace 'Picsel' with 'Artifex'. I didn't particularly know about that (though I know a release is planned), is that replacement job assigned to someone? | 21:52.08 |
| pedro_ : okay, will look in morning. code blind now. :) | 21:52.30 |
pedro_ | nods. | 21:52.50 |
henrys | Jogux: it is assigned to paulgardiner he may ask for you to do it. | 21:55.16 |
Jogux | henrys : righto. I'll check with Paul when he's back. | 21:55.38 |
henrys | Jogux: but mace is supposed to be handling website changes. | 21:57.00 |
Jogux | henrys : ah, right. there'll need to be some code changes / graphics updates too. | 21:58.28 |
| we're completely stopping using picsel.com website I guess? | 21:58.37 |
henrys | Jogux: I don't know details there miles is going directly to mace for that. | 21:59.56 |
Robin_Watts | Jogu: "The mode efficient" | 22:11.25 |
| calling malloc is bad. | 22:12.13 |
Jogux | henrys : right. there's various in-app links to picsel.com pages that we should hence likely change. but I can ask mace about that. | 22:12.19 |
| robin_watts : will fix both. oops. | 22:12.33 |
Robin_Watts | calling malloc and then calling Pal_Mem_free is very bad :) | 22:12.43 |
Jogux | :-) | 22:12.55 |
| actually I suspect it isn't as fortify didn't blow up :) | 22:13.06 |
Robin_Watts | Windows debug malloc is absurdly forgiving. | 22:13.47 |
| Wasn't there a picsel coding style edict that we should avoid // ? | 22:18.19 |
| pedro: your commit looks good except for: | 22:19.04 |
| 1) the use of // - not sure that's an issue. | 22:19.12 |
pedro_ | there was, but seemed to be less strictly adhered to once every toolchain we used accepted it | 22:19.18 |
Robin_Watts | 2) really minor thing - in commit messages, leave a blank line between paragraphs or git log concats them all in --oneline mode. | 22:19.54 |
pedro_ | nods - want me to tweak before pushing? | 22:20.16 |
Robin_Watts | 3) Is there no way to defer an error during a translation? So that the document will open with a "some elements may be lost" warning ? | 22:20.48 |
| we don't want to fail opening, but if the file was entirely a wmf and we ignore it, we ought to give some sort of message? | 22:21.42 |
pedro_ | mm, was thinking that too - you get a warning when we have deferred OOM, but not for translation issues afaik | 22:23.18 |
Robin_Watts | If the answer to 3) is no, then perhaps we should open an enhancement bug for that. | 22:23.30 |
| And you should go ahead and push. | 22:23.37 |
pedro_ | obviously we could hook something up | 22:23.38 |
| yes | 22:23.41 |
| sounds like the right thing - I'll raise a bug anyway | 22:24.08 |
Robin_Watts | With MuPDF, we have a 'cookie' | 22:24.19 |
| When you start a render you pass a pointer to the cookie structure in. | 22:24.37 |
| Every time we hit an error, we increment the error counter, and optionally bale if the error counter hits the max. | 22:25.02 |
| So the callers have the choice over how forgiving to be. | 22:25.41 |
pedro_ | I think we'd really need to figure what types of errors too; missing gfx, unsupported items, parsing errors etc to make an informed choice | 22:26.50 |
Robin_Watts | pedro_: Yeah. If we had some sort of central structure that was passed around everywhere we could have a bitfield in it. | 22:27.33 |
| I don't think the context is ideal for this though, and I'm not sure what else we have. Maybe the Edr pointer itself? Something to ponder on. | 22:28.13 |
pedro_ | sure - probably pass translation status through to the DA interface layer and notify translation issues via events at that point. The unloved loadData structure could possibly be extended as a two-way 'translationData' structure | 22:34.35 |
| we did have many of these events pre-tgv but I think it was rationalised quite a bit | 22:37.46 |
| jogux: I'm happy with the changes, same comments as Robin. The '3denabled' switch for using fill or stroke colour does worry me a bit though, and we may possibly be better generating a valid colour for the rendering rather than deciding not to render. | 22:40.29 |
pedro_ | offs for the night | 23:21.41 |
| Forward 1 day (to 2014/04/03)>>> | |