| <<<Back 1 day (to 2012/11/15) | 2012/11/16 |
tidemann | Hi, anybody here that can help me installing mupdf on Solaris? When I try to issue 'make prefix=PATH install' I get the following error: make: Fatal error in reader: Makefile, line 3: Badly formed macro assignment | 09:30.36 |
| Any help would be greatly appreciated :) | 09:31.26 |
kens | We don't have a huge amount of experience with Solaris | 09:31.45 |
| Are you using Gnu make ? | 09:32.13 |
tidemann | I see. It was a breeze compiling on OS X, though it would be so on Solaris as well. | 09:32.18 |
| which make | 09:32.37 |
kens | GNU | 09:32.43 |
tidemann | hehe yields /usr/ccs/bin/make | 09:32.44 |
| I was trying to be clever and post the output of which make | 09:32.56 |
kens | I think that's the Sun compiler version of make | 09:32.57 |
| Which may well not like a gnu makefile | 09:33.19 |
| chrisl ping | 09:33.26 |
chrisl | Sun make almos certainly won't work | 09:33.28 |
kens | :) | 09:33.33 |
tidemann | Yeah, but using gmake prefix=$HOME install | 09:33.43 |
| yields "Assertion failed: ap <= end, file job.c, line 2412 | 09:33.48 |
| Abort | 09:33.48 |
| " | 09:33.50 |
| Maybe I'm not using gmake correctly? | 09:34.16 |
| OK, I'll try to look more online to see for help on GNU make then. | 09:36.23 |
chrisl | I don't think the mupdf makefiles have got a Solaris target, it might need some makefile hacking to get it working | 09:36.50 |
tidemann | OK, I see. This will require some effort. | 09:37.59 |
| If I get it to work, I will get back to you, either here or on mupdf.com. | 09:38.26 |
| But don't get your hopes up :) | 09:38.32 |
chrisl | Sun make *definitely* won't work, the mupdf makefile(s) have a lot of gnu make specific stuff in there | 09:38.36 |
| tidemann: if I have time, I'll look at it, but it may not be soon | 09:39.14 |
tidemann | Thanks chrisl, that is really appreciated. I totally understand this might not be on top of your priority list, take your time :) | 09:39.52 |
chrisl | tidemann: are you on x86 or Sparc? | 09:40.44 |
tidemann | > uname -a | 09:41.01 |
| SunOS selje.idi.ntnu.no 5.10 Generic_118833-36 sun4u sparc SUNW,Sun-Fire-V440 | 09:41.02 |
chrisl | Right-oh, that means I'll have to wind up the rubber band on my clockwork sparc box ;-) | 09:41.58 |
tidemann | Hehehe :) | 09:43.00 |
| OK, once again, thanks for your time. See you later! | 09:43.30 |
Robin_Watts | Gee. Approaching 24hours downloading the PDF_1.7_ATS stuff. | 10:56.30 |
kens | good grief, how much is there ? | 10:58.14 |
Robin_Watts | 2.63Gig so far, which doesn't sound much for 24 hours. | 11:24.39 |
kens | No, that seems really slow | 11:24.50 |
Robin_Watts | Maybe SVN is horribly inefficient. | 11:27.03 |
kens | Ah, I've always found it to be horribly slow transferring test files. | 11:27.29 |
| You're probably better getting them directly from the server | 11:27.45 |
| Unless you think you really need them under source control | 11:27.55 |
Robin_Watts | Hmm. It's used 3Gig of my data allowance since midnight. | 11:28.45 |
| kens: I'd prefer them under source control because then I can update them. | 11:29.27 |
kens | Robin_Watts : but do you think the files will be updated ever ? :-) | 11:29.44 |
Robin_Watts | No, but when new ones are added, I'd quite like to be able to get them brainlessly :) | 11:30.06 |
kens | Umm, I thought we ahd all the files in there, I sort of doubt they will add new ones. But its up to you. | 11:30.42 |
Robin_Watts | kens: No, I mean, we'll add new directories with new sets etc. | 11:31.58 |
| And historically, we *have* added new files to such dirs in the past (fixing brokenness in the tests etc) | 11:32.32 |
kens | Well, like I said, its up to you, but I would just grab them when needed. I stopped keeping the test files under svn control some time ago because it was too slow for words | 11:32.42 |
Robin_Watts | kens: eek. 22 Gig in that directory in the checkout on peeves. | 11:45.51 |
| I think I will be making use of the unlimited overnight bandwidth my ISP provides by downloading a bit each night. | 11:46.11 |
kens | Sounds like the best solution | 11:46.25 |
Robin_Watts | Ah, that's better. I've downloaded 10gig, but only cost me 3 of my allowance. | 11:52.26 |
| That's much more reasonable. | 11:52.33 |
kens | yeah that sounds better | 11:55.51 |
chrisl | Hmm, the cluster seems to be having a bit of a tantrum...... | 11:59.02 |
kens | its Robin using up all the bandwidth :-) | 12:02.19 |
chrisl | It's all those "started producing errors" which have always produced errors happening again | 12:03.04 |
| tor8: there's a couple of small changes for mupdf building on Solaris in my repository - if you're interested....... | 12:04.20 |
kens | Maybe we shouldn't be running files whic produce errors... | 12:04.26 |
chrisl | Most of these are PCL/PXL files which are supposed to produce errors, *and* output, which we need to md5 :-( | 12:05.09 |
kens | that's a bit of a problem then I guess | 12:05.29 |
chrisl | There is "singular_ctm_3_tr_mode.pdf" which, when run through ps2write results in an error reading the ps2write output | 12:06.30 |
kens | Hmm, don't know about that one | 12:06.50 |
| Except that I haven't done anything awful in that area recently | 12:07.06 |
| I think that's a degenerate matrix test though | 12:07.25 |
chrisl | I've seen it before, but since the input is, strictly speaking, broken, I've never worried about the ps2write output being broken, too | 12:07.41 |
kens | seems reasonable to me | 12:08.17 |
chrisl | I've also got a bunch of differences despite my push being a "clean" master tree | 12:08.40 |
kens | PCL ? | 12:08.56 |
chrisl | PCL, PDF, PS, PXL..... | 12:09.18 |
kens | 111I saw it mostly with PCL/PXL files recently | 12:10.03 |
| But since my change was in pdfwrite, and most of the diffs were in rednering I decided it wasn't me... | 12:10.30 |
chrisl | It's mostly PCL/PXL, but three or four PDF/PS ones, too | 12:10.49 |
kens | some files have long been non-deterministic | 12:12.02 |
| there are som ePCL patterns which always seem to vary with pdfwrite, I'm not at all sure why | 12:12.18 |
| THey do seem to be coming up more frequently than they used to though | 12:12.32 |
chrisl | With 18 files showing differences, and 280+ allegedly newly producing errors, it can be difficult to sort the wheat from the chaff | 12:13.13 |
kens | Maybe I should look at the PCL patterns... | 12:14.44 |
chrisl | Are they problematic? | 12:15.48 |
kens | there are sopme (few) cases which show non-deterministic behaviour | 12:17.00 |
| THey are clearly 'constructed' files no real-worl dfiles so I haven't bothered much with them, the real files using patterns seem to be OK. | 12:17.29 |
| It sort of looks like an /Indexed colour problem again | 12:17.53 |
| or palette to put it in PCL terms | 12:18.06 |
chrisl | Well, I'm more bothered by the files that error out.... | 12:18.13 |
kens | PDF/PS ones ? | 12:18.26 |
| I could look at the singular_ctm one but liek you say, the input is broken, so.... | 12:18.54 |
chrisl | No, PCL/PXL - the couple of PS/PDF involved ones I know about them. But 200+ PCL/PXL files that may or may not be inteded to error out is more of a pain | 12:19.59 |
kens | Hmm, yes that's going to be difficult :( | 12:21.18 |
chrisl | I wonder if we should have an explicit list of files which are intended to error out, so the cluster knows not to report them | 12:21.50 |
kens | I htougth of that, then decided it was an awful lot of trouble | 12:22.39 |
chrisl | It may work out as less trouble then checking them all every couple of weeks! | 12:23.29 |
kens | Well, I wouldn't have to write the script, so I'm fine with that :-) | 12:31.59 |
chrisl | I'll maybe ask Marcos about it - if I remember..... | 12:35.17 |
tor8 | chrisl: thanks. | 13:04.50 |
chrisl | tor8: np | 13:13.16 |
tor8 | chrisl: the problems he mentioned sounded more ominous than just missing -I directives and not using gnu make | 13:15.21 |
chrisl | tor8: I'm not promising it will work on his sytem, but it works on mine..... ;-) | 13:16.03 |
Robin_Watts | tor8: I see you've pulled chrisl's changes into your mupdf repo. | 13:54.19 |
| Have you pushed all your changes too ? | 13:54.24 |
tor8 | Robin_Watts: I rebased and pushed the patches you looked at yesterday | 13:54.40 |
Robin_Watts | So you have. | 13:54.46 |
| Fab. | 13:54.52 |
| tor8, paulgardiner: Some changes on my repo for review... | 13:58.13 |
| Fax decompression speedup (work in bytes rather than bits) | 13:58.35 |
| and the ndk-profiler stuff. | 13:58.43 |
| The profiler stuff involves adding a 'can write files to storage card' permission to the manifest. | 13:59.22 |
sebras_ | Robin_Watts: isn't it possible to make the latter part conditional on profile-builds? | 14:01.40 |
Robin_Watts | sebras: If you can see how, I'd love to know. | 14:36.37 |
| The new ndk appears to have dependencies broken, meaning it always rebuilds *everything* in an ndk-build. | 14:47.00 |
| That gets old very quickly. | 14:47.05 |
sebras_ | Robin_Watts: argh. | 14:54.49 |
| Robin_Watts: I have had a number of patches sitting in AOSP for over a year without them being included. :-( | 14:55.09 |
Robin_Watts | AOSP ? | 14:56.36 |
sebras_ | Robin_Watts: Android Open Source Project | 14:56.51 |
Robin_Watts | Ah. | 14:56.55 |
sebras_ | Robin_Watts: like this one: https://android-review.googlesource.com/#/c/36363/ | 14:57.38 |
| Robin_Watts: one of the things that attracts me to mupdf is the turn-around time for patches. not so much for AOSP. | 14:59.09 |
Robin_Watts | a 1 character change. A 10 minute wait. Android. | 15:00.14 |
| I'm sure there is a haiku in there fighting to get out. | 15:00.25 |
sebras_ | Robin_Watts: I wonder how they managed to break the ndk. I mean -- its based on make so they ought to get dependency checking for free... | 15:02.00 |
Robin_Watts | sebras_: I occasionally get warnings of 'clock skew' (but not every time). | 15:02.36 |
mssm | Interesting problem: I found this to fix my corrupted PDFs: http://superuser.com/questions/278562/repair-corrupt-pdf however, ghostscript does not seem to produce any output files anymore. There simply is not output file. | 15:24.46 |
| I have also tried -sOutputFile= instead of the -o flag. | 15:25.10 |
| It definitely does something, showing me each page very quickly (using -dNOPAUSE), but that's it. | 15:25.48 |
| Apart from some stuff printed to StdOut, no resulting file. The weird thing is - when I tried this half a year ago, it worked. | 15:26.10 |
| Version is 9.06-1 | 15:26.32 |
kens | Rather than pointing to a thread somewhere else, why don't you jst say what it is you are doing | 15:26.48 |
mssm | Trying to run my broken PDF files through ghostscript to fix them. | 15:27.25 |
| So that my ebook reader displays them. | 15:27.30 |
kens | Give me the command line | 15:27.35 |
| I think I can already see your problem | 15:27.47 |
mssm | kens: thats why I pointed to that thread | 15:27.48 |
| kens: the first answer contains the line | 15:27.54 |
kens | SO post it here, I'm not off to some random place to look | 15:28.05 |
mssm | gs -o repaired.pdf -sDEVICE=pdfwrite -dPDFSETTINGS=/prepress corrupted.pdf | 15:28.16 |
| ignore the white space | 15:28.23 |
kens | I think you have a typo | 15:28.26 |
mssm | Yay! :D where? | 15:28.46 |
kens | If Ghostscript is displaying the outptu, then its not running pdfwrite, its running the default (display) device | 15:28.56 |
mssm | Ok, that's good to know. This displaying I found unusual. I don't remember it being there. | 15:29.30 |
kens | So I don't hink you are putting -sDEVICE=pdfwrite | 15:29.35 |
| Check tghat most carefully, also put it in front of the -o | 15:29.53 |
| order is important in parameter evluation | 15:30.04 |
kens | had someone on gs-devel tyuping -sDEVCIE=pdfwrite and havingt the same problem, only on Linux | 15:30.57 |
mssm | kens: alright, this was indeed the problem | 15:31.43 |
| I managed to write it in camel case | 15:31.50 |
kens | Yep, that won't work, its case sensitive | 15:32.00 |
mssm | :-/ | 15:32.02 |
| kens: thank you | 15:32.06 |
| Linux here btw :) | 15:32.08 |
kens | NP | 15:32.09 |
mssm | If that's what you were referring to. | 15:32.35 |
kens | Nopt especially, the difference with the gs-devel one was that he had no X windows, so was getting an X error | 15:33.00 |
mssm | oh; so he was running it in the frambuffer? | 15:33.23 |
| Interesting to fix documents like that outside of X. | 15:33.52 |
| I mean... how would you check it quickly? | 15:33.59 |
kens | Err, he intended it to be pdfwrite, but because of the typo he was geting the display device, which uses X< but couldn't connect to X windows, because it wasn'rt installed, hence an error | 15:34.00 |
mssm | ah, ok | 15:34.28 |
| ok, i am out; thanks again for the help with that silly typo :/ | 15:34.43 |
Robin_Watts | Crumbs. | 16:02.25 |
| This file has 160 fax images on a page. Each 15571x15571 | 16:02.49 |
| And I was worried that it took 8 seconds to display after each zoom. | 16:03.22 |
henrys | wow external makefile projects now work in xcode - everything seems okay, intellisense, debugger gui cool. | 16:16.39 |
Robin_Watts | henrys: Excellent! | 16:18.02 |
| By "external makefile projects" does that mean "ghostscript" ? | 16:18.49 |
henrys | yeah I'm debugging away in pcl and gs. I find it pokey and prefer emacs, I guess if I got sucked into intellisense I would switch - but I'm afraid I'd squander whatever memory cells I have left if I didn't use them to remember what code I intended to write. | 16:47.31 |
| It's good and bad when you go back to a bug and it suddenly starts working. | 16:49.56 |
kens | OK I'm off for a pizza, goodnight everyone have a good weekend | 16:59.37 |
henrys | bye kens | 16:59.44 |
Robin_Watts | night kens | 16:59.47 |
mvrhel_laptop | bbiab | 17:41.39 |
henrys | I think it is easier just to use git rev-list --bisect and do it manually, the git bisect thing just annoys me when you are looking for a fix. | 17:48.40 |
Robin_Watts | I'm getting incredible deja vu with this code. Where have I written subsampling code before? | 18:00.31 |
| (in the Artifex universe) | 18:00.36 |
henrys | hmm I don't remember anything, but that doesn't mean a lot ;-). Have an appointment be back in an hour or so. | 18:01.54 |
Robin_Watts | Oh, the downsampler of course. | 18:15.20 |
| tiffscaled and all that. | 18:15.29 |
| Forward 1 day (to 2012/11/17)>>> | |