| <<<Back 1 day (to 2012/09/13) | 2012/09/14 |
ray_laptop | mvrhel: (for the logs). Thanks for taking me off the hook. We also had the kid's 'back to school night' where we go through there class schedule, meet each teacher for 10 min, then move on to the next class -- welcome to middle school. | 07:18.56 |
kens | Lol apparently I've just been awarded a bronze badge for PostScript in Stack Overflow. | 07:19.53 |
| luser_drrog claims its the first one.... | 07:20.02 |
ray_laptop | on the subject of interpreting PDF's with > 2Gb offsets -- a bit ago, our PS on 64-bit builds was having 64-bit ints which had different results than an Adobe (CPSI) or QL reference, so alexcher made 64-bit machines FORCE 32-bit ints in PS. | 07:21.00 |
| I can't imagine it would be hard to have a build switch to force 64-bit PS ints (int64_t / uint64_t) even on 32-bit machines, and it should work fine. | 07:22.06 |
| and since the 64-bit linux builds worked OK, it should not have a big bug tail (famous last words ;-) ) | 07:22.53 |
| kens: CONGRATULATIONS on your bronze medal :-) | 07:23.28 |
| except for the fact that it keeps me from getting real work done, it is sort of fun to answer dumb and/or newbie questions (for a while) | 07:24.33 |
kens | True, but I only answer them when I have time..... | 07:24.59 |
ray_laptop | kens: enough time ? what kind of concept is that ??? | 07:25.25 |
kens | I was more amused by the fact that someone thinks its the first on awarded for PostScript | 07:25.33 |
ray_laptop | kens: well, probably so, since PS is rather old, and stackoverflow is rather new | 07:26.07 |
| I'm sure henrys could become a medalist for PCL :-) | 07:26.34 |
kens | Hmm, I didn't think to look and see if there was a PCL/PXL/GL2 tag | 07:27.06 |
ray_laptop | Artifex could do a PCL/PS/XPS triathalon tag-team :-) | 07:27.08 |
| kens: or probably more PDF than XPS (sort of like having rowing in a triathalon) | 07:29.02 |
| well, time to sleep. see you in a few hours... | 07:29.53 |
kens | goodnight Ray | 07:30.12 |
ray_laptop | g'nite | 07:30.18 |
chrisl | Jeeze, a bug report about building 8.63..... really?!?! | 09:04.43 |
kens | Quite. | 09:04.52 |
| But actually the problem is more likely to be the gcc version | 09:06.37 |
| Have you tried 4.2.0 and current code ? | 09:06.57 |
| On Solaris of course :-( | 09:07.04 |
chrisl | No, not on Solaris - there's no 4.x.x package available on Solaris, and it's a bit of a 'mare to build it, apparently | 09:07.33 |
kens | Well, it could be that current code won't build with it either.... | 09:07.52 |
chrisl | I'm hoping that, with the new versions using configure, we might get a working build | 09:08.22 |
kens | Yes I saw your reply | 09:08.34 |
| I thin its a reasonable thing to ask them to do, as you point out 8.63 is long unsupported | 09:08.56 |
| chrisl I see GG made another loss last quarter, despite increased sales | 09:47.13 |
chrisl | kens: that's not too good, I wonder how that happened | 09:51.52 |
kens | don't know to be honest. | 09:52.07 |
chrisl | I'd have thought costs would have stablised by now.... | 09:52.37 |
kens | There's not a lot in the report on their site. I do note that they seem to have held the results back, I was expecting them at the end of July, and they actually released them at the end of August. | 09:52.48 |
| chrisl, maybe they have :-) | 09:52.57 |
| chrisl that Sun compiler one is intersting. | 10:01.51 |
chrisl | It's weird, I'll give them that! | 10:02.18 |
kens | Defaults to maximum optimisation.... | 10:02.30 |
chrisl | Yeh, I seem to remember the Sun compiler has done that for some time. I think I ran across it with Jaws, which also didn't work with maximum optimisation enabled! | 10:03.19 |
kens | Not really a great endorsement of the optimiser... | 10:03.39 |
chrisl | No, but it's not uncommon, in my experience - it just seems silly to default to maximum | 10:04.47 |
kens | Yes, maximum seems like a really daft default | 10:05.02 |
chrisl | But that kind of thing is exactly why we explicitly set "-O2" | 10:05.41 |
kens | wyeah, "Don't mess with the makefiles" | 10:07.13 |
chrisl | Well, I've told them my results, we'll see what transpires. There is something odd about their Makefile, but I can't really be bothered trying to track it down | 10:08.29 |
viorel | Hello! I told you several days ago about my problem regarding several open pdf's making the android application to crash. As I am looking through the native part of the android project, I see that there's only one fz_document variable, and I'm thinking that this is the cause of my problem: when I am creating the first document everything goes fine, then, after adding the second document, thus replacing the old one with a new conte | 10:19.02 |
| ->and crashes. Please correct me if I'm wrong. | 10:19.15 |
Robin_Watts | Sounds plausible. | 10:20.05 |
| The android app can be thought of as being in 3 layers. | 10:20.27 |
| At the core we have the cross platform mupdf library itself. This is the same on all platforms, and is perfectly capable of doing what you want. | 10:20.59 |
| On top of this we have some very small native classes that are just complex enough to do what we want in the android app and gives a bridge to the java. | 10:21.59 |
| Then we have the java layers. | 10:22.05 |
| It's likely that to make any real changes to the app you'll need to change both the java and the thin bridge layers. | 10:22.30 |
| OOOH. Vistart. Puts the start menu back on windows 8. | 10:52.43 |
viorel | Thanks! So, I will replace the variables with a vector of structs, have an element for each open document, and take care of the alloc/dealloc through functions called from java. | 10:53.51 |
Robin_Watts | viorel: OK. best of luck. Let us know how you get on. | 10:55.30 |
| Did I talk to you about licensing? | 10:55.38 |
| (Someone appeared the other day, and disappeared before I got a chance to mention it). | 10:55.54 |
| MuPDF is released under 2 different licenses; The GNU GPL and the Artifex commercial license. | 10:56.21 |
| If you are complying with the GNU GPL (including, but not limited to giving away all the source code for your app) then you can clearly use MuPDF under the GNU GPL, and we need say no more. | 10:57.09 |
| If you are developing an app just for you, and are not giving it to anyone else, then distribution doesn't come into it, and you're fine again. | 10:57.37 |
viorel | thank you! I'll let you know. I'm not very strong with C, and don't have an IDE that knows autocomplete and all, but I'll try. I didn't talk about licensing, but my company is interested in the commercial use of the library. | 10:57.38 |
Robin_Watts | OK. So it sounds like you need a license. Let us know then you're ready to talk licensing and we'll put you in touch with Scott. | 10:58.23 |
viorel | yup! I'll need it as soon as I manage to integrate the engine into our app, and see how it does. | 10:59.16 |
sebras | Robin_Watts: eeeep! licensing, don't mention it! | 11:41.22 |
| ;) | 11:41.29 |
Robin_Watts | sebras: I think it's only fair to mention it (gently) up front. | 11:42.27 |
sebras | Robin_Watts: yeah, I agree with you. I'm just seeing a trend. | 11:42.54 |
Robin_Watts | is finding the same problems with viewJPEG under windows 8 that the (potential) customer is. | 11:43.00 |
| kens: ping | 12:32.02 |
kens | pong | 12:32.11 |
Robin_Watts | We've got a potential customer that wants to use gs as part of a metro app. | 12:32.33 |
| he's given us a test app that first makes a jpeg from an input file, then calls gs to pdfwrite that into foo.pdf using the viewjpeg.ps thing. | 12:33.15 |
| That works fine. | 12:33.22 |
kens | OK | 12:33.26 |
Robin_Watts | BUT... when he tries to do more than 1 page, it fails. | 12:33.35 |
| In that we only get the first page output. | 12:33.47 |
kens | How do you do more than on epage of a JPEG ? | 12:34.10 |
| As far as I know JPEGs only have one page | 12:34.20 |
Robin_Watts | It doesn't matter if I use viewjpeg.ps or just print some text and then showpage. | 12:34.24 |
kens | Robin_Watts : well that sounds like something weird and Metro related | 12:34.44 |
Robin_Watts | No, I mean, he's trying to "(a.jpg) viewJPEG showpage (b.jpg) viewJPEG showpage" etc. | 12:34.54 |
kens | And does it work on Windows 7/Linux/Mac etc ? | 12:35.16 |
Robin_Watts | Using the same command line, yes. | 12:35.30 |
kens | Then its a Metro thing of some kind | 12:35.40 |
| I have no idea what that would be | 12:35.50 |
Robin_Watts | There is a problem in that on metro I'm using his gsapi based calls, and on windows 7 I'm using the command line. | 12:36.08 |
| Let me try the command line on metro. | 12:36.13 |
kens | OK that seems like a good idea | 12:36.20 |
Robin_Watts | OK. The command line from metro works fine. | 12:39.32 |
kens | So that suggests something screwy in his app | 12:39.44 |
Robin_Watts | Sorry. The command line on windows 8 works fine. | 12:39.45 |
| Though interestingly the inbuilt pdf viewer on windows 8 dislikes the file. | 12:40.01 |
kens | That's always possible | 12:40.15 |
| What does anythign else think of it ? | 12:40.28 |
Robin_Watts | acrobat doesn't object. | 12:40.38 |
kens | I'm assuming that running under the Metro UI doesn't do anything outlandishly different from the desktop command line | 12:40.53 |
Robin_Watts | Nor does gs. | 12:41.00 |
| In theory, no. | 12:41.12 |
kens | Robin_Watts : Acrobat is a tart, doesn't object to anything ;-) | 12:41.12 |
| then I would say its something with his app | 12:41.26 |
| THere's not a lot else it can be.... | 12:41.36 |
Robin_Watts | yes, but I can't obviously see anything. but it is crazy C++... | 12:41.49 |
| I will swear at it some more after lunch. Thanks. | 12:42.02 |
kens | Hmm is that crazier than ordinary C++ ? | 12:42.06 |
| I'd offer to look but I don't have a Windows 8 setup | 12:42.22 |
Robin_Watts | sorry, C#. | 12:42.38 |
kens | Oh yes, that's crazier than C++ | 12:42.49 |
chrisl | Robin_Watts: How about adding code in the gs_api layer that dumps the command line args, to check whether the C# binding is messing up the arguments? | 12:44.44 |
Robin_Watts | chrisl: A plan! | 13:14.29 |
| oh, because I can't see the sodding output :( | 13:14.54 |
chrisl | Write it to a file? | 13:15.16 |
Robin_Watts | I can try that. Metro is a pain about where it lets you write, but... | 13:15.41 |
kens | Metor, or Windows 8 ? | 13:21.30 |
| Vista is quite bad enough about permissions | 13:21.43 |
Robin_Watts | Metro is worse, AIUI. | 13:24.40 |
kens | Hmm, could that be related to the problem ? | 13:24.55 |
| Its hard to see how, since pdfwrite will open temp files and such no matter whether its one page or more, but it sounds like a difference between the Metro app and command line | 13:25.43 |
| You could also try debugging the app I suppose ? | 13:26.09 |
Robin_Watts | Can't step from C# into the dll. | 13:27.26 |
| Gotta love it! | 13:27.29 |
kens | OMG really ? | 13:27.33 |
| THat's awful | 13:27.38 |
Robin_Watts | at least, not with my version of VS that I can see. | 13:27.52 |
kens | C# is your friend, trust C# you have no need to use C.... | 13:28.21 |
Robin_Watts | OK, args look fine to me. | 13:30.56 |
| Time to output stdout and stderr too. | 13:31.14 |
chrisl | Robin_Watts: can you see stdout and stderr, or will you have to do some fiddling? | 13:35.07 |
Robin_Watts | I can't see them. | 13:35.22 |
| I'm going to make the gs stuff write them to files. | 13:35.31 |
kens | That goes back through the api sstuff though, doesn't it ? So it should be easy to do | 13:35.53 |
| For small values of easy | 13:36.00 |
Robin_Watts | yeah. | 13:36.25 |
chrisl | The app can't be run from the command line, then? | 13:36.25 |
Robin_Watts | his app? no. | 13:36.36 |
kens | Not a Mtro app | 13:36.40 |
chrisl | What a pile o' cr*p :-( | 13:36.58 |
| I was thinking you could add some debug prints to the PS to see if we're actually interpreting the entire line | 13:37.56 |
Robin_Watts | kens: Do we perchance close the file, then open it again after each page ? | 13:41.58 |
kens | Robin_Watts : I don't think so. | 13:42.09 |
chrisl | Robin_Watts: why does the string argument have a trailing ' ? | 13:42.14 |
kens | Let me have a look | 13:42.24 |
Robin_Watts | chrisl: Cos he's cutting and pasting badly. | 13:42.39 |
kens | Robin_Watts : are you getting a valid 1-page PDF file produced from teh 2 page job ? | 13:43.07 |
chrisl | I'm looking at the code he attached to one of the earlier mails: | 13:43.13 |
| "\"("+inputPath+") viewJPEG showpage\'" | 13:43.23 |
Robin_Watts | kens: yes. | 13:43.32 |
kens | Robin_Watts : then you must be executing pdf_close | 13:43.43 |
| Which only happens when the device is closed | 13:43.50 |
Robin_Watts | chrisl: Yeah, he tried lots of different quotes, and is incapable of posting the correct one into the damn email. | 13:44.01 |
kens | Either because you are using %d in teh format string, or because the device is explicitly closed | 13:44.11 |
| Either way that *will* cause the files to be closed. | 13:44.25 |
chrisl | Robin_Watts: so what does your version of that line look like? | 13:45.05 |
kens | So qeustion 1, are you using %d in the OutputFile string ? | 13:45.11 |
Robin_Watts | No quotes at all. | 13:45.14 |
| kens: No. | 13:45.16 |
kens | Robin_Watts : then I guess the device is being closed after the first page and presumably not reopened | 13:45.43 |
Robin_Watts | "-dNOPAUSE", | 13:45.46 |
| "-dBATCH", | 13:45.48 |
| "-sDEVICE=pdfwrite", | 13:45.50 |
| // Set the input and output files | 13:45.52 |
| String.Format("-sOutputFile={0}", outputPath), | 13:45.54 |
| psFile, | 13:45.55 |
| "-c", | 13:45.57 |
| "10 10 moveto /Helvetica findfont 40 scalefont setfont (ass) show showpage 10 10 moveto /Helvetica findfont 40 scalefont setfont (biscuits) show showpage", | 13:45.59 |
| "-c", | 13:46.01 |
| "("+inputPath+") viewJPEG showpage", | 13:46.03 |
| "-c", | 13:46.04 |
| "10 10 moveto /Helvetica findfont 40 scalefont setfont (hole) show showpage", | 13:46.06 |
| "-c", | 13:46.07 |
| "("+inputPath2+") viewJPEG showpage", | 13:46.09 |
| "-f" | 13:46.10 |
kens | You seem to have 2 '-c' in there | 13:46.33 |
Robin_Watts | I have 4. | 13:46.43 |
kens | err, 4 | 13:46.45 |
| Is that valid ? | 13:46.59 |
Robin_Watts | Yes. | 13:47.03 |
| Having just 1 makes no odds. | 13:47.11 |
kens | Hmm, never knew tht | 13:47.12 |
chrisl | And you only get the first page? | 13:47.26 |
Robin_Watts | yes. | 13:47.35 |
| you'll note that the first -c should give me 2 pages. | 13:47.45 |
kens | Well something is causing the device to close after the first page | 13:48.34 |
chrisl | What about adding whitespace between the showpage and the trailing " ? | 13:48.39 |
Robin_Watts | ok, the stdout is showing all 5 pages. | 13:51.33 |
| Hmm. press return to continue ? | 13:51.49 |
kens | You need to specify -dNOPAUSE to avoid that | 13:52.06 |
| which I see you do | 13:52.14 |
Robin_Watts | oh! It's losing argv[0]! | 13:52.32 |
chrisl | argv[0] is the app name, it should be "gs" or something | 13:52.58 |
Robin_Watts | chrisl indeed. So it's not seeing -dNOPAUSE, and it's never getting a return... | 13:53.17 |
kens | chrisl maybe not in C# | 13:53.18 |
chrisl | kens: but that's how the command line processing in gs expects it to be | 13:53.39 |
Robin_Watts | Trying with a "BOGUS" in as argv[0] now. | 13:53.41 |
| And that's it! | 13:54.30 |
kens | Well there you go :-) | 13:54.39 |
chrisl | Doh! | 13:54.41 |
Robin_Watts | What a fool. Should have spotted that. Thanks guys | 13:54.47 |
chrisl | Perhaps it would be good to suggest hooking up the stdout and stderr handlers | 13:55.24 |
Robin_Watts | yeah. | 13:55.43 |
chrisl | Interestingly the jpeg option would have "worked" because argv[0] == "q" | 13:56.22 |
Robin_Watts | yes, hence the confusion. | 13:59.56 |
| and now I can run screaming from windows 8 again. | 14:01.44 |
kens | Until the next time.... | 14:02.22 |
henrys | marcosw:we'll be getting the pdf ats shortly it should probably go to you and become part of the overnight test. | 15:42.59 |
marcosw | henrys: should we add it to the cluster test as well :-) | 15:43.38 |
henrys | yeah let's do 128 nodes at amazon - miles has money | 15:45.07 |
Robin_Watts | http://www.theregister.co.uk/2012/09/12/raspberry_pi_supercomputer/ | 15:46.36 |
| Let's build our own... runs off one mains socket. | 15:46.50 |
| (plus lots more for the network switches) | 15:46.59 |
henrys | we'd melt the leggos | 15:49.30 |
| the way my office feels some days | 15:49.59 |
Robin_Watts | henrys: That's cos the thermal output of a PC is huge. | 15:50.21 |
| Thermal output of a raspberry pi is bugger all. | 15:50.35 |
henrys | yes that's true. ain't going to work for us anyway we need local storage, we'd have to put little ssd on each rasberry. | 15:53.02 |
fm__ | can someone help with https://bugs.launchpad.net/ubuntu/+source/ghostscript/+bug/1049635 ? | 15:53.31 |
| chrisl: i guess you are the one again ... ;) | 15:53.58 |
Robin_Watts | henrys: We can mount one (or more NAS) on the ethernet and pull the jobs over to the local SD card before running. | 15:54.13 |
chrisl | fm__: it'll have to wait until Monday. In meantime you could narrow it down to a single page of the PDF? | 15:54.40 |
fm__ | it prints none ;) chrisl | 15:55.17 |
| if i print all, i get an empty page and the ERROR page | 15:55.30 |
| should i try printing just the second, etc? | 15:55.47 |
chrisl | OKay, it would be easier to cut down the PS if we start with a single page. | 15:55.51 |
fm__ | pdftk burst as a start? | 15:56.16 |
| or does that modify the pdf too much? | 15:56.25 |
chrisl | fm__: that would probably work - or even just specify a page range in the print dialog | 15:57.03 |
| fm__: the "Captured print job data" you posted on the bug is the cups queue PDF, not the Postscript that goes to the printer | 15:59.11 |
fm__ | yeah thats the procedure linked in the wiki. where can i find instructions how to get your data? | 15:59.57 |
chrisl | You'll have to ask tkamppeter, I have no idea | 16:00.20 |
| fm__: can you open a bug report @ http://bugs.ghostscript.com and attach the PDF file? Select PS Writer for the "Component", and you can include a link to the Ubuntu bug | 16:21.05 |
fm__ | chrisl: http://bugs.ghostscript.com/show_bug.cgi?id=693334 | 16:28.18 |
chrisl | fm__: thanks. It's going to need the PS Writer maintainer to look into it | 16:31.57 |
| fm__: did this used to work with the old poppler based conversion? | 16:42.01 |
Robin_Watts | Why do we have #ifdef DEBUG around if_debug1(); statements ? | 17:26.42 |
henrys | probably just folks treating it like gs_debug_c which is unnecessary but not harmful. | 17:33.29 |
| hope the shoulder is better today Robin_Watts | 18:28.16 |
Robin_Watts | henrys: yeah, pretty much better thanks | 18:28.30 |
| henrys, alexcher: The jbig2 interface code for luratech is based on malloc/free, not the gsalloc functions. | 19:20.56 |
| state->image = malloc(state->height*state->stride); | 19:21.18 |
| And not just for small amounts of memory. | 19:21.32 |
henrys | Robin_Watts:that does need to be fixed. I'll make a note for the tuesday meeting | 19:44.02 |
ianneub | Hey all, with muPDF is it possible to read the bookmarks in a PDF file? | 20:47.40 |
Robin_Watts | Stupid windows. Need to reboot. | 23:36.45 |
| Forward 1 day (to 2012/09/15)>>> | |