| <<<Back 1 day (to 2014/04/14) | 2014/04/15 |
fargac | hi | 06:25.22 |
ghostbot | hi, fargac | 06:25.22 |
fargac | Hi again,first of all ı want to thank you for quick answer. Here my steps to compile mupdf 1.3 Install Android SDK (http://developer.android.com/sdk/index.html) Install Android NDK Download the setup.exe from http://cygwin.com/install.html and follow instructions. Download mupdf-1.3-source.tar.gz Extract mupdf source folder somewhere in my hard disk Instaall visual studio from this link http://www.microsoft.com/en-us/do | 06:40.30 |
mvrhel_laptop | blah spent the day (and night) working on text selection with mouse movements that go outside the window to get it to scroll through the pages in this case (selecting text along the way). I think I finally have it working. Need to clean up a few things tomorrow | 06:54.08 |
| night kens | 06:54.11 |
kens | night Michael | 06:54.16 |
mvrhel_laptop | or good moring | 06:54.17 |
kens | WSell its morning for me ;-) | 06:54.24 |
mvrhel_laptop | yes | 06:54.27 |
aditsu | hi, where can I download urwfonts? the links at http://www.ghostscript.com/GhostPCL.html are not working | 09:30.45 |
chrisl | aditsu: the PCL fonts are included in the GhostPDL source archive | 09:37.15 |
Robin_Watts | fargac: and then what? | 09:38.31 |
fargac | hi robin,sorry | 10:24.28 |
| for late answer | 10:24.30 |
| i was outside | 10:24.37 |
| i sent you mail | 10:24.48 |
| * Go to mupdf source folder, open win32 folder and double click generated.vcproj then it will open with visual studio. | 10:26.05 |
| In visual studio just run the c++ application. Then ,it create a folder inside mupdf source. the folder name is âgeneratedâ and it contains some header files (.h files) | 10:26.23 |
| Open Cygwin console, and type following command and enter. cd /cygdrive/c/mupdf-1.3-source/android /cygdrive/c/ndk/ndk-build | 10:27.05 |
| After ndk-build, com.artifex.mupdfdemo pakcage was created,libmudpf.so was created... | 10:27.27 |
| Final step, create an Android application using mupdf android source (inside android folder). Create this project as a library project and include it in mymain project. | 10:27.40 |
| This is how I call mu-pdf activity from my project. Uri uri = Uri.parse("path to pdf file"); Intent intent = new Intent(context, MuPDFActivity.class); intent.setAction(Intent.ACTION_VIEW); intent.setData(uri); context.startActivity(intent); | 10:28.07 |
Robin_Watts | fargac: Do you have a backtrace for the error? | 10:28.44 |
fargac | this works succesfully on nexus 5,but not work on samsung galaxy ace(android 2.3.3) | 10:28.47 |
| ı do not have backtrace for the error,because i am at work now :( | 10:29.24 |
Robin_Watts | I cannot immediately see why this doesn't work. | 10:29.25 |
fargac | but ı remember this error : | 10:29.43 |
| Caused by: java.lang.UnsatisfiedLinkError: Library mupdf not found | 10:29.58 |
| ı don't understand why this works on nexus 5 (android 4.4.2) but not android 2.3.3 | 10:30.52 |
| robin watts : samone suggests this at site http://stackoverflow.com/questions/5350653/compiled-jni-library-libmupdf-so-working-fine-for-android-2-2-android-emula | 10:31.42 |
| but ı can not try yet. | 10:32.01 |
Robin_Watts | yeah, I'd run adb logcat and see what it says. | 10:32.29 |
| Certainly I have mupdf running on 2.3 (and even 2.2 with a trivial hack) | 10:32.42 |
fargac | ı can also send you android project (ChoosePdfActivity) of mupdf | 10:34.35 |
pedro | fargac: it could potentially be down to poor handling of LD_LIBRARY_PATH on specific devices; I've had issues with library resolution on HTC devices with 1.6, 2.2 forexample | 10:41.20 |
| much less of an issue on more recent android releases | 10:41.40 |
fargac | so,what should i do ? | 10:48.14 |
pedro | fargac: have you listed the library dependencies for mupdf and checked that they exist on your 2.3.3 device? If they are present, you should be able to verify if the library can be loaded correctly by doing System.load(abs_path_to_mupdf_lib) rather than System.loadLibrary() which uses LD_LIBRARY_PATH | 11:01.47 |
| (that's not a solution, just to verify that the library can be loaded on the device) | 11:03.18 |
| fargac: iirc the issue was that some manufacturers used <package-dir>/lin rather than <package-dir>/armeabi/lib in LD_LIBRARY_PATH so those devices could not find the native library using LoadLibrary() | 11:07.03 |
fargac | ı was trying to find mupdf library dependencies | 11:08.05 |
pedro | meant <package-dir>/lib not <package-dir>/lin | 11:08.26 |
fargac | sorry,i am a bit confused. what should i do ?sould i change " static { System.loadLibrary("mupdf"); }" in MupdfCore.java ? | 11:18.24 |
pedro | fargac: I think the first thing you should do is check that mupdf's library dependencies actually exist on the 2.3.3 device . You should be able to list them with the arm-linux-androideabi-readelf exe in your NDK (use "arm-linux-androideabi-readelf -d mupdf.so" on the commandline) | 11:26.36 |
| then check if the libraries listed by readelf are present on your device | 11:26.53 |
| if any are missing then mupdf will not load | 11:27.04 |
fargac | thank you very much pedro | 11:27.46 |
| i will try to check,after turn back to home. | 11:28.20 |
pedro | if those libraries *are* all present then it may be a problem with the LD_LIBRARY_PATH setup n the device (basically not finding mupdf.so) | 11:28.23 |
| then you can change the System.loadLibary("mupdf") to System.load(<path to mupdf lib>) | 11:29.21 |
fargac | ok,what if one of those librarires are absent ? | 11:35.29 |
| does this mean "mupdf does not support all android devices" ? | 11:35.57 |
| sorry but ı want to ask another question ,should package name same with my project package name ? because my mupdf package com.artifex.mupdfdemo whereas my project com.fargac.mycompany | 11:42.56 |
pedro | I think the first step is to see wether the device has the dependant libs, and go from there. | 11:48.51 |
| For your second question, are just refactoring the muPDF demo? | 11:50.19 |
fargac | after ndk build, ı changed package name and all other files com.artifex.mupdfdemo to com.fargac.mycompany | 11:59.40 |
| but i got error | 11:59.44 |
| java.lang.UnsatisfiedLinkError: Native method not found: com.fargac.mycompany.MuPDFCore.openFile:(Ljava/lang/String;) | 12:00.13 |
| pedro, ı have choosepdfactivity project(mupdf android project) , ı can send you now,if you want to see if something is wrong ? | 12:02.54 |
| i am downloading android-ndk-r9d now,to check libraries | 12:04.08 |
pedro | fargac: probably best for Robin to look at that - I'm not developing muPDF (just had similar issues with other android native apps before) | 12:05.41 |
fargac | ok,thank you very much:) | 12:06.49 |
pedro | no problem | 12:07.19 |
fargac | ı know,ı asked some stupid questions but ı am novice about native libraries :) sorry again:) | 12:11.13 |
pedro | they're not stupid questions - the whole native apps support on Android has been quite error-prone/difficult especially for 1.x and 2.x devices | 12:12.10 |
| but much better now (thankfully!) | 12:12.27 |
fargac | ok,pedro.ı hope this problem was solved. see you... | 12:17.17 |
Robin_Watts | fargac: Presumably when you changed the name to com.fargac.mycompany you changed all the references in mupdf.c too, right? | 12:48.31 |
fargac | ı check now,mupdf.c the only place that ı did not change:) sorry | 12:51.47 |
| ı will change and try again:) | 12:51.57 |
henrys | paulgardiner: can you bring up read at the meeting and we'll all talk about it? | 13:12.52 |
| s/read/raed/ | 13:13.01 |
paulgardiner | henrys: will do. Forgot it was that day already | 13:32.13 |
ray_laptop | morning, all | 14:08.15 |
ray_laptop | reviews the logs | 14:08.22 |
kens | Hi ray_laptop | 14:11.30 |
ray_laptop | hi, kens | 14:13.52 |
| kens: I think that having someone (chrisl or anyone else) review your patches (as I think is our policy) would be good -- if nothing else to spot typos in the commit messages prior to pushing to the master | 14:17.54 |
| now, if I'm mistaken and it's no longer our policy I (for one) will be glad to stop having to ask someone to review my patches :-) | 14:18.57 |
kens | I was under the impression this was a MuPDF policy, not a Ghostscript policy,did it change ? | 14:19.11 |
Robin_Watts | ray_laptop: It's definitely our policy for mupdf. | 14:19.12 |
| I thought it was 'recommended' for gs, but not required. | 14:19.38 |
| (though personally, I'm all for it) | 14:19.48 |
kens | If we are going to discuss that I'd prefer we do it at a staff meeting | 14:20.15 |
| rather than IRC | 14:20.19 |
henrys | not required and I think we all know how everyone weighs in so let's not beat on it again. | 14:20.31 |
ray_laptop | kens: the staff meeting is IRC -- or do you mean the "in person" staff meeting | 14:20.49 |
kens | I mean in person yes, again I was under the mipression that 'staff meeting' was the in person meeting | 14:21.19 |
ray_laptop | henrys: OK, I thought that we _did_ discuss it at a previous staff meeting (not the most recent one) and that we were supposed to have reviews for gs | 14:22.18 |
kens | I don't recall agreeing to that | 14:22.32 |
henrys | is there no editor with a spell checker on windows that can pop up with git commits? | 14:22.42 |
ray_laptop | henrys: but if I misunderstood, sorry. And I'll be glad to not have to do it | 14:22.50 |
kens | I do have a spell checker for git commits | 14:22.55 |
| At elast, I do with Git gui | 14:23.14 |
ray_laptop | kens: "sunthesize" passes your spell checker ? | 14:23.17 |
kens | I don't know, I can always miss these things | 14:23.30 |
henrys | ray_laptop: artificial sunlight | 14:23.49 |
kens | The specll checker doesn't like things like PDF anyway, so I never get a commit without red in it | 14:24.02 |
ray_laptop | kens: It really isn't critical, and I won't bug you further. I just thought it might help our QC | 14:24.21 |
henrys | kens: yeah same here with emacs | 14:24.26 |
| kens:but I can add words but I usually don't | 14:26.21 |
Robin_Watts | henrys: We've got a skype group going for discussing GhostDocs stuff. | 14:27.05 |
| me/paul/jogu/pedro/you. | 14:27.22 |
ray_laptop | kens: and a spell checker won't spot things like "implement the BDC, BMC and BDC pdfmarks" (that I think you meant to say "implement BMC/EMC and BDC pdfmarks" | 14:27.29 |
kens | also quite true | 14:27.48 |
henrys | Robin_Watts: oh okay | 14:28.03 |
Robin_Watts | I can add anyone else who is interested, but it's likely to be boring :) | 14:28.21 |
ray_laptop | so no meeting for the rest of us? hurray :-) | 14:28.28 |
chrisl | Well, I have to head out at about ten minutes to the hour, so..... | 14:29.01 |
ray_laptop | I do notice that the SOT support from paul and robin seems to be pretty heavy | 14:29.02 |
Robin_Watts | ray_laptop: No, this isn't as a substitute for the meeting, just for a place to discuss stuff that we can't talk about in public. | 14:29.04 |
ray_laptop | Robin_Watts: just wishful thinking :-) | 14:29.42 |
Robin_Watts | ray_laptop: You mean the support mail from potential customers ? | 14:29.49 |
marcosw_ | morning | 14:29.56 |
mvrhel_laptop | good morning | 14:29.58 |
henrys | Robin_Watts: I don't see any activity on skype | 14:30.22 |
Robin_Watts | I *hope* that that'll calm down a bit as customers sign on properly. At the moment we are in the heavy "start up" phase. | 14:30.28 |
ray_laptop | Robin_Watts: is Wuu just a potential customer ? | 14:31.49 |
henrys | tor8:mupdf release? | 14:31.57 |
Robin_Watts | ray_laptop: He's working his way through with Miles at the moment. | 14:32.16 |
tor8 | henrys: done. | 14:32.20 |
Robin_Watts | The problem is that Miles has to contact people and explain to them that Picsel isn't an option any more and they need to move to us. They (understandably) are reluctant to blindly make a move, so it's a slow process, and we get all the questions that Picsel haven't answered for them. | 14:33.19 |
marcosw_ | Robin_Watts: Miles has asked me to put together that comparison between the old SOT and the new SOT, when is the new SOT going to be released? | 14:33.32 |
henrys | tor8:done for you or has chrisl put it up to. Nice to get an email for us to all download and test? | 14:33.48 |
| s/to/too | 14:33.56 |
Robin_Watts | marcosw_: Not a clue. I'm still fighting fires. I think pedro actually has done some improvements (shadings etc?) | 14:34.16 |
chrisl | henrys: I updated the web page and did the Artifex release yesterday | 14:34.20 |
paulgardiner | marcosw_: releasing without regression testing doesn't seem wise and ATS is proving very difficult to recreate | 14:35.08 |
jogux_mac | ATS is going to be late next week at best :( | 14:35.36 |
henrys | chrisl: if it isn't too much trouble can there be an email like ghostscript? | 14:35.39 |
chrisl | henrys: I thought tor8 would do that with the "main" release | 14:36.05 |
tor8 | henrys: what's wrong with IRC? :) | 14:36.09 |
chrisl | But I can certainly ping a mail out when the release is "complete" | 14:36.26 |
tor8 | we've never done email announcements for MuPDF but I guess we could start doing that | 14:36.37 |
henrys | chrisl: you send to the mail list also right? | 14:37.31 |
chrisl | henrys: to gs-devel? Yes, but I wouldn't do that for mupdf | 14:37.57 |
henrys | chrisl: yeah I guess that wouldn't make sense. | 14:38.39 |
tor8 | henrys: there is no mupdf mailing list. | 14:39.21 |
chrisl | henrys: tbh, as I said to tor8, the mupdf releases are really just PR anyway, almost all interested parties pull straight from git..... | 14:39.46 |
henrys | chrisl: customers don't pull from git | 14:40.40 |
chrisl | I thought many did..... | 14:41.27 |
Robin_Watts | We've tried to get some to use git, but they still like to be driven by releases I think. | 14:42.35 |
henrys | chrisl: I'll double check, nonetheless mupdf is now a real product : 15% of revenues and growing, we need a solid release procedure. | 14:43.24 |
mvrhel_laptop | wow. I did not realize it had reached 15% | 14:43.54 |
tor8 | henrys: there is a mupdf twitter account, where I post release announcements, but it has like 5 followers so there's not much point there | 14:43.54 |
chrisl | I'll certainly send a mail out to the staff when my bit is finished | 14:44.18 |
| ... in the future | 14:44.26 |
henrys | chrisl: okay | 14:44.28 |
| tor8: I think about how we can spread news a bit more. | 14:44.39 |
| s/I/I'll/ | 14:44.51 |
| marcosw: there was a cluster discussion in the logs yesterday, don't know if you saw it. | 14:45.12 |
| marcosw: and I had another thought: each client could fuse mount across the net - check a file for when it needs to do it's cluster business. As opposed to logging into the server every minute. | 14:47.30 |
tor8 | henrys: nfs, smb or sshfs? | 14:48.20 |
| sshfs is quite convenient, when it works | 14:48.31 |
henrys | mvrhel_laptop: yeah and it is the "growing". gs is fairly stable. | 14:48.45 |
chrisl | henrys: do have anything else specifically for me? As I said, I had to head out a little before the nominal end of the meeting..... | 14:49.00 |
henrys | tor8: I was thinking sshfs - user space fs is nice. | 14:49.22 |
| chrisl: sorry chrisl no I"m good. | 14:49.30 |
ray_laptop | henrys: is that 15% by number of customers or by revenue ? | 14:49.46 |
henrys | revenue | 14:49.53 |
| this quarte | 14:50.15 |
| r | 14:50.17 |
ray_laptop | henrys: how come Miles doesn't share that info ? | 14:50.37 |
| (I guess I should ask him). | 14:51.00 |
henrys | ray_laptop: I'll request that he does at the next meeting. | 14:51.04 |
marcosw_ | henrys: I'll look at the logs for yesterday's discussion re. the cluster. | 14:51.11 |
chrisl | henrys: BTW, I got the pcl and xps executables building in the gs build system on Windows - still needs work and tidying up, but progress, at last! | 14:51.24 |
henrys | chrisl: wow yeah! | 14:51.47 |
chrisl | Right now it's using recursive nmake calls, and I'm looking at removing that requirement | 14:52.14 |
henrys | ray_laptop: I don't know if it is a great idea that we pay attention to it. | 14:52.15 |
| ray_laptop: I mean that *all* of us pay attention to it. | 14:52.46 |
marcosw_ | I've been thinking about re-writing the cluster system to not require the nodes to login into casper every minute, but instead just check a port for a message. The same way that jobs are sent now. | 14:52.48 |
ray_laptop | henrys: I agree. | 14:52.58 |
Robin_Watts | chrisl: I had a random idea the other day. | 14:53.20 |
| chrisl: We could put an x86 gnu make binary in gs/tools/ | 14:53.42 |
chrisl | No | 14:53.57 |
Robin_Watts | and then have the solution call that. That would get us parallel builds. | 14:54.00 |
henrys | paulgardiner: read's bug should be P2 or P1 and assigned to a human maybe tor8 | 14:54.22 |
chrisl | I hate the idea of having binaries in a source distribution | 14:54.27 |
Robin_Watts | chrisl: fair enough. | 14:54.47 |
chrisl | Robin_Watts: plus, I'd have to rewrite the windows makefiles | 14:54.58 |
Robin_Watts | I did wonder how hard it would be to tweak gnu make to read nmake format makefiles. | 14:55.10 |
| The differences are fairly minimal I think. | 14:55.19 |
henrys | marcosw_: my beef is these cron job announcements messing up the logs, I like to leave a system console open | 14:55.24 |
chrisl | Oh, so not just a binary, but a specially patched binary - you've been back on SOT too long!! | 14:55.59 |
| Robin_Watts: ^^ | 14:55.59 |
henrys | also paulgardiner was just about to fix the temp file business when he was taken over by SOT should we hand that off to someone? | 14:56.17 |
ray_laptop | Robin_Watts: conditionals is pretty easy, but changing to nmake's way of replacing macro assignments might be harder | 14:56.26 |
Robin_Watts | chrisl: No, I was thinking that if we supplied a binary, it would be a vanilla one. | 14:56.29 |
marcosw_ | henrys: the cron logs on the cluster nodes shouldn't be an issue anymore, I rewrote the polling script to be peristent. | 14:56.36 |
paulgardiner | henrys: okay on Raed's bugs, although the first of the two requires sinking into the depths of openssl... not that I relish the thought of doing so, just that I have already had to do so before | 14:56.50 |
henrys | marcosw_: oh great thanks | 14:56.57 |
| marcosw_: also (I know you're busy) but there were some customer fixes that need notification . | 14:57.30 |
Robin_Watts | marcosw_: One of your machines is prodding casper a lot. see /var/log/auth.log | 14:57.42 |
paulgardiner | henrys: gs temp file business? If we aren't in too much of hurry, it's quite a nice thing to play with on flights | 14:58.28 |
henrys | 2 more minutes: anything else? feel free to joint the Skype discussion | 14:58.42 |
chrisl | OKay, I'm off - back in about two hours..... | 14:59.02 |
henrys | paulgardiner: okay, and raed's bug? | 14:59.20 |
mvrhel_laptop | henrys: I did not get much done last week since I was out. I am hoping to get all the features into gsview this week | 14:59.21 |
| that I want | 14:59.25 |
| then there will be a bit of clean up | 14:59.32 |
henrys | mvrhel_laptop: yeah I figured | 14:59.34 |
ray_laptop | henrys: did we ever figure out about that "POSSIBLE BREAK-IN" IP ? | 14:59.49 |
marcosw_ | Robin_Watts: I don't see anything suspicious in the auth.log (other than the POSSIBLE BREAK-IN ATTEMPT from Miles' office). | 15:00.02 |
mvrhel_laptop | bbiab | 15:00.11 |
marcosw_ | and: error: Bad prime description in line 90 | 15:00.19 |
ray_laptop | marcosw: Oh, that's Miles' office ? | 15:00.20 |
henrys | ray_laptop: I assumed that was because of the frequency of marcosw_ attempts but that was a guess | 15:00.20 |
Robin_Watts | marcosw_: If you tail -f the log, you'll see that stuff is appearing rather more often than we might like. | 15:00.27 |
henrys | marcosw_: looks like a DOS | 15:00.56 |
ray_laptop | yeah, like every second | 15:00.59 |
Robin_Watts | i.e. we seem to have a connection attempt every 1-2 seconds or something. | 15:01.03 |
marcosw_ | Robin_Watts: so you suggest we turn off logging? | 15:01.04 |
Robin_Watts | marcosw_: I thought you said that stuff was persistent now, so we shouldn't see that? | 15:01.25 |
ray_laptop | marcosw: or fix the problem causing so many attempts | 15:01.29 |
marcosw_ | Robin_Watts: that sounds about right, I have 14 cluster nodes in my house, they connect every 15 seconds, so that works out. | 15:01.46 |
henrys | let's call it adjourned thanks | 15:01.59 |
ray_laptop | turning off the logging is liking putting black tape over the oil pressure light on the dash | 15:02.01 |
marcosw_ | is the problem we are trying to solve that the logs messages are too big or am I missing something? | 15:02.26 |
Robin_Watts | marcosw_: If they have a persistent connection, why can't we send a heartbeat using that rather that connecting/disconnecting. | 15:02.34 |
marcosw_ | the connection is not persistent. | 15:02.48 |
Robin_Watts | marcosw_: The thing that got us into this is that the logs are basically useless for anything else now. | 15:02.51 |
henrys | marcosw_: why do you need a full public key authentication every 15 seconds is the question - logs aside? | 15:03.12 |
Robin_Watts | We looked at the logs to see why norbertj was having problems getting in, and couldn't see for the noise. | 15:03.14 |
henrys | marcosw_: per machine | 15:03.24 |
ray_laptop | henrys: right now it's just used to indicate that the machine is up (it touches a file) | 15:04.40 |
marcosw_ | when we started the cluster we had 4 or 5 cluster nodes and each connected once a minute, now we have 24 cluster nodes and the connect every 15 seconds. | 15:04.41 |
Robin_Watts | If it's just for a heartbeat (each machine saying "I'm alive"), we can implement that using a cgi script through apache? | 15:04.42 |
paulgardiner | henrys: Raed's openssl bug probably isn't going to get fixed over a single flight. How quickly do you feel we need to fix it? I'm very happy for someone else to look at it... just wondering if I might have an advantage given I've delved into openssl before. | 15:04.57 |
Robin_Watts | have each node fetch clusterheartbeat.pl?machine=name | 15:05.38 |
henrys | He can wait for openssl - we do need to fix his real bug where the pdf doesn't load right? | 15:05.44 |
paulgardiner | henrys: I also have some finishing off of the MuPDF iOS app that I've temporarily shelved because of the push to get a SOT release | 15:05.44 |
marcosw_ | Robin_Watts: then people wil complain that the apache logs are too full (or thathttp://www.ghostscript.com isn't resposive). | 15:06.03 |
paulgardiner | henrys: ah okay. Make sense for tor8 to look at that one, if he's happy to. | 15:06.34 |
marcosw_ | the right way to fix it is what I said earlier, use a port to communicate from the cluster nodes to casper, the same way that casper currently sends cluster jobs to the nodes. | 15:06.45 |
Robin_Watts | marcosw_: Well, if we were building this from the ground up again, I'd suggest a persistent connection from client to server, where we send heartbeats every 15 seconds. | 15:06.55 |
tor8 | paulgardiner: is this bug 695112? | 15:06.56 |
Robin_Watts | marcosw_: Right. | 15:07.10 |
paulgardiner | tor8: that's him | 15:07.29 |
Robin_Watts | mvrhel_laptop: If you want to be included in the GhostDocs meeting, please sign into Skype now :) | 15:08.03 |
marcosw_ | and if we were to build ghostscript from the ground up I'd suggest not writing the PDF interpreter in PostScript. It's easy to be an architect in hindsight :-) | 15:08.25 |
henrys | mvrhel_laptop: probably should if you plan to work on it. | 15:08.29 |
paulgardiner | Robin_Watts: would that be TLS heartbeats? :-) | 15:09.20 |
Robin_Watts | marcosw_: I am not in any way attempting to blame you. | 15:09.45 |
| marcosw_: 20:20 hindsight etc. | 15:10.00 |
marcosw_ | I realize that, it's just that the cluster system had a lot of limitations imposed on it at the beginning due to the original idea of staff being able to deploy the cluster node code on a system that was behind a firewall and one I didn't have access to (this was primarily Ralph's requirements). | 15:11.35 |
| so the initial system used only ssh (and scp) to connect from the nodes to the master. | 15:11.58 |
Robin_Watts | marcosw_: And indeed, it's very clever. | 15:12.40 |
mvrhel_laptop | Robin_Watts: ok | 15:13.23 |
marcosw_ | I'll modify the heartbeat and start a job code to not use ssh but communicate via a port instead (I actually do that for the statuses messages from the nodes to casper, those used to be sent via rsync over ssh as well). | 15:14.29 |
mvrhel_laptop | Robin_Watts: how do I join the group? | 15:15.39 |
| ok I am in now | 15:15.52 |
marcosw_ | I'll also try to figure out why Miles ip address is being flagged as a POSSIBLE BREAK-IN ATTEMPT and onces from my ip address are not, it can't just be frequency of connections, since I have 14 nodes and miles only has 10. | 15:16.23 |
jogux_mac | marcosw_ possible break in attempt usually means 'dns mismatch'. | 15:19.32 |
ray_laptop | marcosw: I think it says that | 15:20.10 |
jogux_mac | ie. reverse lookup on the ip then a forward lookup of the resulting hostname don't get you back to the same ip. | 15:20.12 |
ray_laptop | which is probably a comcast problem | 15:20.31 |
| reverse mapping checking getaddrinfo for 50-240-204-222-static.hfc.comcastbusiness.net [50.240.204.222] failed - POSSIBLE BREAK-IN ATTEMPT! | 15:21.28 |
marcosw_ | yes, but I use comcast as well. though I have a dynamic ip address and miles' is static, so presumably comcast handles dns on those differently. | 15:21.57 |
ray_laptop | now why that host name (that contains the IP address) doesn't result in that same machine -- that is comcast doing it | 15:22.09 |
marcosw_ | apparently adding it's easily fixed on our end by adding entries to /etc/hosts | 15:23.08 |
jogux_mac | ray_laptop: the hostname doesn't exist. :( | 15:23.17 |
marcosw_ | there, those messages are gone. now to get rid of the rest of the noise... | 15:25.07 |
| henrys: I was going to log in to your macpro to confirm the the once/minute cron messages were gone, but I can't connect. Did you ip address not get updated in the ddns or is something else wrong? | 15:28.25 |
henrys | marcosw_: I'll check shortly in the Skype meeting, you might want to pop in. | 15:34.42 |
marcosw_ | henrys: I'm on the skype but need to be added to the group. | 15:38.05 |
henrys | marcosw_: okay I've asked Robin_Watts to add you | 15:42.24 |
kens | Hmm what's the group ? | 15:42.27 |
henrys | kens:ghostdocs | 15:45.20 |
Robin_Watts | kens: If you want in, I can add you. It's just a multiway conversation. | 15:45.23 |
kens | Robin_Watts : please do, I'll only be reading though, not contributing | 15:45.41 |
ray_laptop | stepping through image_render_color_icc -- it's REALLY inefficient when scaling an image down. It converts the entire source line using the link profile (which is probably OK since that might result in some efficiencies in the CMS), but then it converts each to a color index even if it isn't going to use it. | 15:55.36 |
kens | ray_laptop : is this the interpolation bug report ? | 16:00.45 |
| mvrhel_laptop : Silly question if you have a minute | 16:01.28 |
mvrhel_laptop | kens: sure | 16:01.58 |
Robin_Watts | ray_laptop: I'm looking into the interpolation bug report. | 16:02.10 |
| 80%+ of the CPU time is spent in zoom_y | 16:02.23 |
kens | I have a annotation with a /CA (constant alpha) value defined. I want to draw all the graphics ops for that annotation wiht the CA, any ideaqs how I can do that ? (this is in the PDF interpreter) | 16:02.42 |
Robin_Watts | In this file, we zoom up the image massively, and only a small patch is actually used. | 16:02.56 |
| I have code to reduce the calculations done to just the patch that's required, but it's still larger than we'd like in that case. | 16:03.26 |
mvrhel_laptop | kens: is the pdf14 compositor getting installed? | 16:04.19 |
kens | mvrhel_laptop : I have not a clue | 16:04.29 |
| THis is new code I'm trying to write, so its at the PostScript level | 16:04.40 |
mvrhel_laptop | oh | 16:04.46 |
kens | let me try pdf_info.ps and see if it thinks the file is transparent | 16:06.03 |
| I guess if the interpreter thinks its transparent we will have the compositor | 16:06.18 |
mvrhel_laptop | kens: I guess I am a little confused. Do you want to know how to do all the math and compositing as to what to draw or do you want gs just to handle it in its normal transparency route | 16:06.22 |
kens | mvrhel_laptop : I just want to do whatever ops are required to make the following stuff render with all the compositing done | 16:06.47 |
| pushing/popping groups, etc. | 16:06.58 |
| The problem is I don't know anything about how the PDF interpreter sets up CA for *anything* | 16:07.29 |
mvrhel_laptop | I see. kens. so the pdf14 device has to be pushed and then you should be good as far as the interpreter. the pdf14 device will swallow up the ca changes | 16:07.52 |
| and handle them appropriately | 16:08.00 |
kens | mvrhel_laptop : yes, but how do I communicate the CA ? | 16:08.06 |
| Its no in an ExtGState or anything | 16:08.22 |
| Its part of the annotation dictionary | 16:08.32 |
mvrhel_laptop | kens; we have transparency files that have this | 16:09.08 |
| pdf files | 16:09.12 |
| I know they work | 16:09.15 |
| hold on | 16:09.22 |
kens | mvrhel_laptop : Not with a FreeText annotation | 16:09.26 |
| Because I only just added the code to deal with FreeText | 16:09.39 |
mvrhel_laptop | no but in pdf | 16:09.42 |
kens | HBmm, looks like it is not deteced as having transparency | 16:11.38 |
mvrhel_laptop | zsetopacityalpha is what you want I think | 16:12.09 |
| kens: but you will have to add logic to detect the presence of transparency in your stuff | 16:12.25 |
kens | Yeah I was just figuring that, thanks for the operator, I'll fix the detection tomorrow | 16:12.43 |
mvrhel_laptop | ok | 16:12.48 |
kens | Yeah definitely not being detected, that's going to be awkward, oh well. | 16:13.37 |
| An opportunity to learn another bit of the remarkable PDF interpreter code..... | 16:14.16 |
| THink I'll head out now, goodnight everyone | 16:14.38 |
mvrhel_laptop | Robin_Watts is Python 2.7.6 fine or does it need to be Pyton 2.7.0? | 16:26.48 |
| i.e. http://www.activestate.com/activepython/downloads | 16:27.03 |
Robin_Watts | mvrhel_laptop: Urm... | 16:27.13 |
| in my root dir on casper: python-2.4.4.msi | 16:27.38 |
mvrhel_laptop | this was the page that joseph has in the twiki | 16:27.39 |
Robin_Watts | I didn't think that 2.7 worked? | 16:28.18 |
| Certainly I'm using 2.4.4 here as it's the version that Picsel were using. | 16:28.29 |
mvrhel_laptop | http://twiki.ghostscript.com/do/view/GhostDocs/BuildWindows?twiki_redirect_cache=bbcde0bdd6b4775ceb3c18673944d826 | 16:28.35 |
Robin_Watts | yeah, I'm looking at that page now. | 16:29.01 |
| Try 2.7, see if it works. | 16:29.09 |
mvrhel_laptop | ok | 16:29.14 |
Robin_Watts | It would be nice not to be stuck with using old versions of things. | 16:29.27 |
mvrhel_laptop | brb | 16:37.45 |
jogux_mac | mvrhel_laptop : I believe I'm using the latest 2.7.x; no problems noted other than the annoyance with the file association losing parameters. | 16:38.03 |
mvrhel_laptop | jogux_mac: ok great. thanks | 16:38.25 |
| I think I have all the tools. cloning repos now | 16:38.50 |
jogux_mac | mvrhel_laptop : please do feel free to update the wiki page as you discover things :) | 16:39.27 |
| I'm not sure what problem Robin might have thought he had with 2.7 :( but it has been okay for me. | 16:40.16 |
Robin_Watts | jogux_mac: I can't remember offhand what the problem was, just that there was one. ISTR that even when I was still working at Picsel there was something that stopped us using the latest 2.x python. | 16:42.38 |
jogux_mac | there defiinitely was an issue at one point (but I forget what it was) | 16:45.18 |
| generally Picsel were pretty paranoid about making sure everyone had the same version, so there was a lot of "you must use version xxxx because that's what everyone else has". | 16:46.14 |
mvrhel_laptop | while the cloning chugs along, back to work on gsview... | 16:47.35 |
henrys | marcosw_: huh the ip is correct, so the port must be blocked | 17:03.02 |
marcosw_ | henrys: it's not urgent, whenever you have a chance. | 17:03.38 |
henrys | marcosw_: should be alright now | 17:06.04 |
| marcosw_: yup looks good | 17:47.52 |
mvrhel_laptop | ok. I will attempt a sot build this afternoon. I have all the parts now | 18:58.27 |
| I am going to attempt to use VS2013 though | 18:58.41 |
| Robin_Watts: are we using tabs or spaces in SOT? | 18:58.53 |
jogux | mvrhel : spaces. | 19:41.17 |
| mvrhel_laptop : I'm not sure anyone has tried Robin's solution thing with vs2013 - you may want to start with a command line build - eg. python epage\scripts\tgvbuild.py -debug --dispman-ue2fileviewer -vs2013 app/simple | 19:42.33 |
| though I won't guarantee that works either :( | 19:42.41 |
mvrhel_laptop | jogux: ok thanks | 19:47.34 |
Robin_Watts | My cooker just decided to fuse the whole house. | 19:52.18 |
| I think I've isolated the circuit that's wrong, but if I fall off the net again, that'll be why. | 19:52.44 |
spanners | what a lot of bolls | 19:58.30 |
| Forward 1 day (to 2014/04/16)>>> | |