IRC Logs

Log of #ghostscript at irc.freenode.net.

Search:
 <<<Back 1 day (to 2013/08/12)2013/08/13 
sebras toothrot: no problem. /me sleeps.00:11.15 
ray_laptop this SMask issue for cust 532 is driving me bonkers. I mostly fixed it, with minimal differrences, EXCEPT for Bug690777.pdf -- it's a total failure on that one :-(02:50.12 
  in the meantime, I've been working on an image "alignment" issue with us vs. Adobe (sound famililar?)02:51.05 
  mvrhel: I found that on the customer file, if I "fudge" the image coordniates by 0.5 I get results the same as Adobe (with my previous patch for "pix_alpha")04:21.13 
  so I am (trying to) run a regression with the 0.5 fudge04:22.00 
mvrhel_laptop ray_laptop: I will get going on the file shortly04:29.48 
ray_laptop mvrhel: OK. As I mentioned, I am looking into the "image offset" issue that also makes the customer's file lighter than Adobe (trying to put together a test file to help me)04:32.22 
mvrhel_laptop oh that is interesting. Adobe is not paying any attention to the CA term only the ca term06:02.30 
  I remember this bug before06:02.34 
  anyway. will beat on this more tomorrow06:08.30 
  ray_laptop: for the logs, I may have something that works for us07:01.40 
  off to bed now though07:01.45 
  kens: I am so sorry, I have not had a chance to look at the color issue you gave me07:01.59 
kens mvrhel : no problem, I've seen you and Ray are busy07:02.25 
mvrhel_laptop between doing timing tests for a potential japanese customer, helping ray and the windoze phone I have been swamped07:02.28 
  I am going to try to get to it this week07:02.50 
kens OK thanks, I'll need to talk about pdfwrite and colour again this next dtaff meeting07:03.10 
  staff*07:03.14 
mvrhel_laptop ok good night talk to you all in the morning07:09.41 
kens goodnight michael07:09.49 
Robin_Watts Morning all.09:33.50 
kens Hi Robin_Watts feeling OK now ?09:33.53 
Robin_Watts yeah, much better thanks.09:34.03 
kens did you manage to enjoy any of the 2nd week ?09:34.30 
Robin_Watts oh yeah.09:35.28 
  I lost an afternoon, but it was an afternoon where we weren't doing anything.09:36.04 
kens oh well that's not bad at all09:36.18 
Robin_Watts and I felt a bit ill for the next couple of days, but didn't miss anything.09:36.21 
  The only thing I missed out on was dignity.09:36.42 
kens :-)09:36.47 
pipitas Hi all09:38.51 
kens Hi09:38.55 
pipitas Is it possible to use Ghostscript (or a PostScript program) to extract JBIG2 encoded images from a PDF file?09:38.55 
kens Extract in what sense ?09:39.12 
pipitas Ultimately, I want to arrive at a script or commandline which just checks if the JBIG2 encoded image did use LOSSY or lossless compression. Lossy is the one giving problems it looks (lossless may be "safe") -- even if all JBIG2 is lossless, of course.09:39.14 
  Background to my question is flaw in the Xerox WorkCenter devices: when scanning to PDF they may mangle numbers (see http://www.dkriesel.com/en/blog/2013/0811_all_three_compression_modes_mangling_numbers_on_xerox_workcentre_7545 and http://www.dkriesel.com/en/blog/2013/0802_xerox-workcentres_are_switching_written_numbers_when_scanning ).09:39.28 
kens ROFL09:39.33 
pipitas There are people out there who want to check the probability of their years old archived business documents being corrupted or not.09:39.42 
  I used "gs -Zw ..." to get some debug output about JBIG2, but I cannot discover anything about lossless or lossy.09:39.51 
  So to extract the JBIG2 image completely may be the only resolve, and then look at it with some other tools.09:40.04 
kens IIRC its not as simple as'lossy' or 'not lossy'09:40.22 
pipitas The Xerox flaw happens when you do "Scan-to-PDF"09:40.36 
kens Yes but that's not the same thing09:40.43 
  JBIG2 has a number of featuires.09:40.56 
  One of them is to deifne text areas09:41.08 
Robin_Watts pipitas: The problem is that one of the JBIG2 compression methods is "recognise that this area is actually text, and replace the bitmap region with a char".09:41.10 
kens and yhou only store the 'text' one, then call it out as required09:41.21 
Robin_Watts So you could theoretically scan a JBIG2 file for such a compression method being used.09:41.36 
  but all you could say is whether the method is used, not whether there is a flaw or not.09:41.56 
kens wIndeed, but that measn using a JBIG2 decoder09:41.58 
pipitas JBIG2, what I read about it, has 2 internal modes: lossy and lossless — but it is not at all clear how these may be implemented. The "standard" is only how to decode a JBIG2 file, not about how you encode it. That's a lot up to the implementation..09:42.16 
Robin_Watts You'd probably want to use "mutool extract" or something built from it.09:42.25 
kens So you would need two steps, first extract the image, tehn use jbig2dec (or somethign imsilar) to determine if the compression uses text areas09:42.37 
  pipitas, I think what you've been reading is a bit of a simplistic treatment09:43.31 
pipitas Robin_Watts: "mutool extract" doesn't extract the JBIG2 image AS JBIG2, or? It does extract/convert to PBM, PPM or JPEG, isn't it?09:43.53 
Robin_Watts pipitas: I beleive extract just extracts. No conversion.09:44.19 
pipitas kens: The simplistic treatment is my own telling of something a bit more complicated I read :)09:44.39 
kens pipitas, JBIG2 in its simplest form is basically CCITT G4 fax. The more clever stuff you use, teh more potential there is for error.09:45.25 
pipitas I already extracted a stream of JBIG2 encoded image manually, but jbig2dec complained about "missing header".09:45.46 
kens And it relies on being able to define things like image and text areas and extracting teh text to hold separately09:45.49 
pipitas Then I looked in the PDF spec to find that this header is skipped in PDFs. I tried to insert a header manually, but jbig2dec still didn't read the file...09:46.30 
kens jbig2dec is used by Ghostscritp (GPL) so I woulod venture to suggest you're not doing it right09:47.00 
  Probably you need to manufactuire a header using the image parameters in the PDF file09:47.16 
  You might be able to get some ideas by reading the Ghopstscritp PDF interpreter....09:47.56 
pipitas kens: yes — but my goal is to check PDFs after the fact: I can easily check if they contain JBIG2 images — but how can I extract them or how can I check if they used the "lossless" compression or the "lossy" one?09:47.57 
kens pipitas, you would need to decode the data09:48.14 
  Robin_Watts : had a suggestion for extracting teh data, after you've xetracted it, you need to decode the compressed data09:48.44 
pipitas kens: decode the data? How?09:48.53 
  kens: you meant to decode the complete JBIG2 encoded stream?09:49.17 
kens Using a JBIG2 dsecoder....09:49.23 
Robin_Watts pipitas: You need to feed the data through a modified jbig2 decoder that flags up if the particular compression method is used.09:49.41 
kens As I recall you will need to decode the entire stream yes09:49.41 
  You can throw the decompressed bits away of course09:50.05 
pipitas kens: as I said, I manually copied the stream to a binary file, and used jbig2dec on it but it complained09:50.11 
kens pipitas, yes and as I sadi,. you must have created the file header incorrectly09:50.31 
  Most likely you need some information form the PDF image parameters to create a usable header09:50.58 
pipitas It should be possible because "pdfimages" from Poppler can extract the image as PPM (as can mutools extract)09:51.01 
Robin_Watts pipitas: A quick hack to do this would be just to take mudraw.09:51.13 
pipitas OK, kens, thanks: I'll try it with the header parameters some more...09:51.30 
kens pipitas, it *must* be possible, because GS uses jbig2dec to deompress JBIG2 images in PDF files09:51.51 
Robin_Watts Then in the thirdparty/jbig2dec/ code, find the code that handles the text compression method, and put an abort(); in there :)09:51.53 
kens chrisl, did you see the reply from Luratech just now ?09:53.06 
pipitas Robin_Watts: your "quick hack" involves modifying the mupdf source code? Then it's beyond me… (I cannot even compile the current un-modified version — it bails out with an error.)09:53.14 
kens pipitas, you could do the same hack on GS09:53.37 
  have the JBIG2decode throw an error if that refinement is used09:53.54 
Robin_Watts pipitas: In that case, this whole project sounds beyond you (no offense!)09:54.19 
kens It will mean modifying JBIG2dec though09:54.20 
pipitas kens: "refinement", yes, that's the magic word some papers used to describe "lossless" modes...09:54.35 
kens Its a JBIG2 term09:54.49 
pipitas Robin_Watts: I don#t feel offended — I'm not a programmer, only a Bash scripter :)09:55.04 
kens I'm afradi any attempt to detect the use of the text refinement (I forget its proper name) will require coding09:55.54 
pipitas Robin: I'm trying to compile current Git version of mupdf on a Macbook. I get this error: 10:05.07 
  LINK build/debug/mupdf-curl10:05.17 
  Undefined symbols for architecture x86_64:10:05.17 
  "_curl_easy_getinfo", referenced from:10:05.17 
  _data_arrived in curl_stream.o10:05.17 
  […]10:05.17 
  ld: symbol(s) not found for architecture x86_6410:05.31 
  clang: error: linker command failed with exit code 1 (use -v to see invocation)10:05.31 
  make: *** [build/debug/mupdf-curl] Error 110:05.31 
  Robin_Watts: Any idea?10:05.39 
Robin_Watts pipitas: Simplest thing is to remove thirdparty/curl and rebuild :)10:05.56 
kens You have latest third-party code ?10:05.58 
pipitas I git-updated the sub-modules10:06.02 
Robin_Watts then you won't build mupdf-curl.10:06.09 
pipitas Ok, will try, Robin_Watts10:07.12 
tor7 Robin_Watts: actually, that's not true.10:08.35 
  the makefile requires curl to be present at all times10:08.46 
Robin_Watts tor7: Ah. Then we should check the macbook build.10:08.58 
  I haven't unpacked the macbook yet.10:09.10 
tor7 Robin_Watts: my current linux build fails without thirdparty/curl because I don't have curl-dev installed10:09.21 
  Robin_Watts: I'll fix the makefile to make it optional10:09.30 
Robin_Watts I can't think why pipitas is getting that error though.10:09.51 
pipitas Robin_Watts: deleting the subdir for curl didn't help.10:11.51 
  I did again a "git submodule update —init" and now get a different error:10:12.27 
  thirdparty/curl/lib/if2ip.c:47:12: fatal error: 'stropts.h' file not found10:12.30 
  # include <stropts.h>10:12.31 
  ^10:12.31 
  1 error generated.10:12.31 
  make: *** [build/debug/curl/if2ip.o] Error 110:12.31 
tor7 Robin_Watts: pipitas: first patch on tor/master should make the curl build optional10:13.33 
pipitas tor7: but since the mutool build system doesnt sport a "configure", you just delete the curl subdir to skip curl?10:14.51 
tor7 yes. all "configure" like actions are done in the Makefile (usually a combination of Makethird and Makerules)10:15.30 
pipitas has to go for lunch now — will be back later10:15.51 
tor7 Robin_Watts: and welcome back! hope you're feeling better now.10:17.59 
  pipitas: to extract jbig2 data you can do bash scripting of "mutool show file.pdf grep | grep JBIG2Decode" to find the JBIG2 streams and their object numbers, and then "mutool show -be X" to extract the raw JBIG2 stream data10:21.53 
  where X is the object numbers from the previous step10:22.22 
chrisl kens: I see the comments from Luratech - should be easy to resolve, *if* I can access an AIX machine. I've always had trouble with the IBM VLP setup11:26.12 
pipitas is back11:33.21 
tor7 pipitas: see my comments above11:33.56 
pipitas tor7: I extracted the JBIG2 stream using mutool. But "jbig2dec myfile.jbig2" says "jbig2dec FATAL ERROR Not a JBIG2 file header". — Just the same as with my manually extracted stream. (These two do differ in 2 bytes, but I don't know why, yet)11:35.42 
  tor7: I also used pdf-parser.py to extract the stream, using the command "pdf-parser.py -D -o 9 -d myfile.jbig2 my.pdf". It also gives almost identical results as the other two, but differing in yet two other bytes...11:37.33 
tor7 are they the first two or last two bytes?11:38.44 
  could be line-ending issues there11:38.50 
  where the stream starts and/or ends11:39.00 
pipitas First two bytes… actually 3 Bytes. Mutool gives three more bytes at the beginning of the file, and 1 less at the end. (Now checking if that is only newlines or what...)11:40.11 
  tor7: If you want to look at first and last bytes of my extractions, you can look here: http://pastebin.ca/243144411:47.52 
tor7 pdfparser looks like it has truncated the first leading bytes that are "0" and added a newline at the end11:50.12 
pipitas things it is unfortunate that in the PDF spec available from Adobe there isn't much info about the JBIG2 header. Now googling for a free-as-in-beer version of the JBIG2 spec…11:52.08 
kens pipitas, I have a copy of the draft JBIG2 spec11:53.24 
  As far as I know I can give you a copy of that,its nto the 'real' spec, butits probably close enough for your purposes11:54.05 
pipitas kens: is it called "fcd1492.pdf" dated 1999, july 16th? Then I found it just now...11:55.55 
kens I changed the name so I could findit more easily, but yes that's it11:56.22 
  section 0.2.4 describes exactly the Xerox problem11:56.56 
  actually 0.2.1 sorry11:57.37 
  "The hazard of lossy symbol coding is to have substitution errors, that is, to have the encoder replace a bitmap11:57.43 
  corresponding to one character by a bitmap depicting a different character, so that a human reader misreads the11:57.43 
  character."11:57.43 
pipitas kens: I also change names in these case, but I keep the original name as a a part: "draft-jbig2-spec—fcd1492.pdf"11:59.09 
kens I have 2 copies, one with the original name11:59.23 
pipitas The Xerox problem is basically described in *all* white papers and articles about JBIG2 since the beginning of time…11:59.56 
kens Its a fairly obviopus potential flaw12:00.18 
pipitas kens: Your initial reaction was 'ROFL' (as was mine) — but when you think of it: the guy who discovered it found it on a copy of a plan for a high rise building, where he recognised some square meter markup to be obviously incorrect.12:01.58 
kens Yep12:02.11 
  THe reason for ROFL was 'why did they use this ?' more than anything else, I was aware of ther potential for problems12:02.38 
pipitas If you think of it… you do not want to drive across a bridge where they give out plans to the building workers which passed through a Scan-to-PDF stage on a Xerox.12:02.43 
kens Indeed, I cannot understand why they bothered to use JBIG2 (which requires a licence), its not like the saving is that large by modern standards12:03.37 
pipitas Another problem is that users cannot even discover what kind of compression they use: on their user interface panel, there is a slider saying "Image Quality/File Size:" and you can choose "normal/small  |  higher/larger  |  highest/largest"12:06.35 
  Who wouldn't use "normal/small" when it is even the factory default setting?12:07.10 
kens I expect there will be a patch 'real soon now' which turns off JBIG2 entirely :-)12:07.39 
Robin_Watts kens: I suspect they'll just disable text refinement.12:09.33 
kens Robin_Watts : in which case its pretty much equivalent to CCITT G412:09.49 
  If it were me I'd kill JBIG2 just to say 'this can no longer happen'12:10.06 
Robin_Watts kens: right. but with the advantage that they can say "we support both jbig2 and ccitt g4". 2 standards always sounds better than 1 :)12:10.40 
kens Untiel one of them starts changing your output :-)12:10.55 
  Hmm, my PostScript code to add /OutputDevice is causing a seg fault12:12.20 
selim hello12:46.16 
ghostbot que tal12:46.16 
paulgardiner tor7, Robin_Watts: 4 commits on paul/master. Fairly big changes, perhaps not ideal for the jetlagged. :-)13:03.34 
Robin_Watts paulgardiner: you have a UTF8 fixme in the first one.13:06.20 
  fz_runetochar13:06.57 
paulgardiner Yeah. It just adds to several UTF8 fixmes already in the fle13:07.02 
Robin_Watts ok.13:07.14 
  I am suspicious of the error handling in layout_text13:07.56 
  You fz_free_text in the catch.13:08.07 
  but never in the non exception case.13:08.24 
  and the non exception case goes through a loop.13:08.43 
  Should it be an fz_always(ctx) { fz_free_text() } ?13:09.33 
paulgardiner I'm struggling to find that. Which commit?13:09.54 
Robin_Watts http://git.ghostscript.com/?p=user/paulg/mupdf.git;a=commitdiff;h=890c1c9da2f6ecc2300c28b0e2135079f3075d2a13:10.02 
  pdf-appearance.c13:10.16 
  second hunk.13:10.20 
paulgardiner Ah, but I return it in the non-exceptional case13:11.16 
Robin_Watts oh! sorry!13:11.28 
paulgardiner np. Better to check everything that's at all suspicious13:12.00 
selim I have a problem I use redmond and PostScript and it works fine but when I print several copies only one PDF is generated (i use google translate, i'am french) thank13:12.05 
kens selim what did you expect to happen ?13:13.21 
haggl hey there13:14.01 
selim I would like more PDF as the number of copies13:14.35 
haggl i've got a small patch for the x11 mupdf application, which enables horizontal scrolling for touchpads and mice with 7 buttons13:14.57 
kens Selim, just copy the PDF file as many times as you want13:15.03 
haggl what do i do with it?13:15.17 
Robin_Watts haggl: OK. Go to bugs.ghostscript.com, and open an enhancement bug.13:15.28 
  add the patch to that, and follow the instructions with the "Contributor License Agreement" and we'll have a look, thanks.13:15.56 
selim it does not match the use I want to do13:16.41 
kens selim, that's life13:17.00 
  THe alternative would be to create one PDF file with each page copied N times13:17.20 
  Create the file then copy it afterwards. Or I don't understand what you meanpossibly13:17.42 
Robin_Watts kens: It sounds like selim has a valid concern.13:17.55 
kens Then I don't understand it Robin_Watts13:18.12 
Robin_Watts selim: So you print a file from an app, and choose "5 copies" in the printer dialogue box ?13:18.21 
  If I do that to a PDF printer, then I'd expect to see the PDF contain 5 copies of the thing I printed.13:19.27 
kens Robin_Watts : we don't do that no13:19.45 
selim yes it is legitimate when I run "5 copies" from an application I would like to have 5 PDF13:19.56 
kens Selim then copy the PDF file 5 times after creation13:20.19 
Robin_Watts I have to go for food, bbs.13:20.22 
  kens: he shouldn't have to, IMHO. If the print path is losing the "5 copies", then that's a bug.13:20.51 
kens Robin_Watts : I disagree, its by design and I'm not changing it, it would be a nightmare13:21.17 
selim when you copy and paste 5 file you expect a copy of the file 5 and not a single, no ?13:23.52 
kens We ignore /NumCopies, as I said it is by design and I will not change it13:24.31 
  It owuld be very hard to do13:24.36 
  If you want 5 copies of the final PDF file, copy it 5 times after creation13:24.50 
selim impossible is part of a wider application13:25.53 
kens Well, its open source, feel free to modify it to match your reuqirements13:26.14 
selim I do not have the necessary knowledge13:27.15 
  it must be a variable or a patch applier to this problem ?13:29.54 
  it must be a variable or a patch to solve this problem?13:30.19 
kens No. the PDF file is created from PostScript, the only knowledge of the number of copies is the /NumCopies key in the page device dicitonary which we deliberatly ignore.13:30.53 
  It is not possible to simultaneously createmultiple PDF file.It would be significantly difficult to copy the PDF file after creation. Its easier to do that from outside Ghostscript (ie modify your own application)13:31.47 
selim I do not have hands on application that prints13:34.49 
kens You are getting a PDF file produced, copy it as many times as you require after it is produced13:35.15 
selim I do not know the number of copies13:36.27 
  My program is just generate the pdf has a PDF and sent to a printer on a remote server, i do not have hands on application that prints13:37.52 
kens There is no facility ot do this in Ghostscript and pdfwrite, and it is deliberate, I am not about to change it, you must find a solution you can use. Somewhere someone is setting the number of copies, find some way to get that informaiton yourself13:39.14 
selim and what do you think about " -dDoNumCopies " I use it but it does not work13:41.16 
kens Because we ignore NumCOpies, I said that13:41.26 
selim but I want to take into account how13:43.08 
kens There is no facility to do so.13:43.27 
selim left to sound arrogant, I'm French and Napoleon to say "Impossible is not French" (" Impossible n'est pas francais ")13:45.50 
kens You just told me that it was impossible to change your way of working13:46.14 
  I said it would be difficult and I wasn't prepared to do it, because the current bahviour is by design13:47.25 
selim it just requires that you include my problem, I do not have all the hands on application of the person who prints13:48.15 
kens You wnat me to change the behaviour of the software to suit your purpose, when a deliberate decision ahs been taken not to do it that way13:49.30 
  Which I'm not prepared to do, if you waqnt it to work that way, then you may change it yourself13:49.56 
selim I Thank you for your help but why -dDoNumCopies does not work13:52.14 
kens Because we ignore /NumCOpies13:52.25 
  For pdfwrite13:52.30 
  In fact we don't ignore it, but what that does is produce copies * number of pages in the single output file13:54.03 
  Which is not what I understand you to mean13:54.17 
  So if you have a 2-page file and set NumCOpies to 2, you will get 1 file with 4 pages13:54.38 
selim he will gather, it does not interest me13:57.08 
henrys I stumbled over an interesting pdf problem - insert a blank page after every page in a pdf with gs. I tried redefining EndPage to be 2 show pages but that gives me an infinite loop at least I think it does (execstackoverlow) didn't look more carefully.14:06.28 
  kens:interesting to hear from Gemma - so basically they are just going unsupported and didn't switch to something else14:09.37 
kens yep14:09.49 
  seems that wya14:09.51 
chrisl hardly a surprise14:10.08 
kens henrys doing a showpage inside EndPage will definitely give an infinte loop14:14.55 
  because teh showpage will call EndPage...14:15.05 
henrys yeah that's what I figured.14:15.26 
kens Hmm 86 minnutes as aginst 2 minuntes, that's something really slow about interpolation with clist14:16.44 
  I winder if its interpolating the whole image on each band or something14:17.01 
Robin_Watts kens: Sorry, just got back.14:18.00 
kens Robin_Watts : np14:18.09 
Robin_Watts Can you explain what you mean by "So if you have a 2-page file and set NumCOpies to 2, you will get 1 file with 4 pages"14:18.15 
  Suppose I have a file with page A then page B.14:18.27 
  and I print that file with numcopies = 2.14:18.35 
kens THen you get AABB14:18.43 
Robin_Watts OK, so you collate.14:18.52 
kens Withg -dDoNumCopies14:18.53 
henrys Robin_Watts:welcome back Robin_Watts! 14:19.18 
Robin_Watts No, sorry, you don't collate. collation would be ABAB, right?14:19.18 
kens Yes14:19.23 
  We do just what a refgular printer would14:19.33 
Robin_Watts kens: Right, and that sounds perfect to me. I thought he was saying he just got "AB".14:20.02 
selim I do not want to assemble pdf but separate pdf14:20.11 
kens Robin_Watts : no, he wants *2* PDF files each containing AB14:20.26 
Robin_Watts oh, well, that's an unreasonable expectation, IMHO.14:20.40 
kens :-)14:20.52 
Robin_Watts We COULD do "ABAB" by just doing "AB" then duplicating the page trees.14:21.28 
  but I accept that that's more work, and not required.14:21.50 
  henrys: thanks.14:22.02 
kens Yep and its not what he's asking for either :-)14:22.02 
Robin_Watts Helen has just gone down with a horrible fever/stomach bug.14:22.25 
kens Oh crumbs, hope she feels better soon14:22.37 
  At least hers wasn't on holiday14:22.45 
Robin_Watts Me too. We've got loads of laundry to do :)14:22.50 
kens I just tested Acrobat Distiller and it also ignores /NumCOpies completely14:23.05 
paulgardiner :-) Send it out the mother in law14:23.26 
  s/out/out to/14:23.43 
selim I think it's a shame this operation14:25.05 
chrisl henrys: do you want to reply to Luratech that their suggestion seems to be correct, and we'll (I'll) deal with it?14:27.29 
Robin_Watts selim: I'm afraid your opinion is in a minority here. It's open source though, so you are free to change it.14:27.31 
henrys chrisl: just got up and going through mail now.14:28.15 
Robin_Watts selim: If you had a file "AB" and you print it to a normal printer with num copies =5 then the printer will wake up, print AAAAABBBBB and then stop.14:28.31 
chrisl henrys: okay, just figured as you'd made initial contact, the confirmation would be best from you14:28.58 
Robin_Watts With the PDF printer, the device wakes up (starts a file), prints AAAAABBBBBB (as pages within the file) then stop (close the file).14:29.08 
  i.e. the PDF printer is doing exactly what the normal printer would do.14:29.24 
selim I know but the problem is that I have no knowledge required to modify the program14:30.14 
Robin_Watts selim: Well, you can employ someone to do it. Or you can become a supported customer of Artifex and we can do it as NRE work. But the latter option will not be cheap :(14:31.03 
  paulgardiner: Did your stuff get reviewed?14:32.51 
selim I'm 18 and not a round :)14:32.55 
paulgardiner Robin_Watts: Don't think so. I thin only you have dipped into it so far.14:33.29 
Robin_Watts damn. so I need to try to understand this splitter stuff :)14:33.43 
paulgardiner Robin_Watts: Maybe not. The splitter stuff is actually a copy of some existing code, changed so as to produce an fz_text object, rather than directly create an appearance stream14:34.50 
  Robin_Watts: eventually we'll be able to remove the old version when all is swapped over to using the pdf device14:35.19 
henrys I don't know how marcosw does this he must have different brain chemistry, same theory I have for elementary school teachers.14:44.55 
kens I think its just practice14:45.10 
Robin_Watts henrys: Someone once explained how mothers can be patient with their small children by telling me that "the brain comes out with the placenta".14:47.21 
henrys oh maybe marcosw suffers from the delusion that he sired all these customers - that would explain it.14:53.06 
  meeting in 5 minutes14:56.24 
Robin_Watts tor7, sebras, chrisl: Tickets arrived, BTW.14:58.24 
chrisl Robin_Watts: cool! Must remember to give you money at the staff meeting.......14:58.56 
haggl_ Robin_Watts: Hmmm.. still did not receive the registration e-mail from bugs.ghostscript.com. How long does that normally take?14:59.57 
Robin_Watts haggl_: minutes.15:00.09 
  check your spam filter?15:00.28 
haggl_ Robin_Watts: I don't have any.15:00.50 
kens there's no review process is there ?15:01.05 
Robin_Watts nafaik.15:01.14 
haggl_ Robin_Watts: I'll try a different e-mail address.15:01.26 
kens Thoguth not, but if there was marcos being on holiday wouldn't work15:01.36 
haggl_ Robin_Watts: Ok.. that worked within seconds. oO15:02.25 
kens strange....15:02.39 
Robin_Watts paulgardiner: First one looks OK.15:02.51 
henrys michael isn't here yet but we can start I really didn't have much other than reviewing the agenda items.15:02.54 
paulgardiner Robin_Watts: Great thanks. Next two not so bad, but the last is again fairly big15:03.30 
kens speak of the devil15:05.02 
mvrhel_laptop sorry I am a little late15:05.04 
kens Short night mvrhel_laptop15:05.15 
mvrhel_laptop yes. seems like I was just here15:05.30 
henrys no problem mvrhel_laptop 15:05.38 
  mvrhel_laptop: anything for the meeting?15:06.32 
  we all seem to be out of words today.15:06.46 
  meeting words that is.15:07.23 
mvrhel_laptop no. I did not get a whole lot done on the windows phone this past week. I am still hoping to have it wrapped up though before the meeting15:07.33 
henrys great news15:08.26 
Robin_Watts I didn't get anything done :)15:08.42 
  and I have to look at this gs/interpolation/landscape/4bpp device bug next, so I may be lost for several weeks :(15:09.23 
henrys Robin_Watts: running with the rhinos ....15:09.24 
Robin_Watts henrys: In more than one sense.15:09.33 
paulgardiner Progress on signature support has been better than expected. MuPDF signed a document this morning... and that's not just my oppinion: Adobe Reader agrees.15:09.56 
mvrhel_laptop Oh nice15:10.05 
Robin_Watts paulgardiner: Congrats!15:10.08 
mvrhel_laptop paulgardiner. Is this something that we can have in the Android app to demo in Japan?15:10.33 
henrys I was wondering about that: food poisoning and being charged by a rhino have similar symptoms - are you sure of what happened?15:10.35 
mvrhel_laptop ha15:10.48 
Robin_Watts henrys: The rhino was unaware of our presence. Fortunately I was downwind.15:11.05 
henrys paulgardiner: oh that's fantastic and unexpected from what you've been saying.15:11.36 
paulgardiner mvrhel_laptop: getting it running on the android app is the next job. It's back again to looking possible for Chicago15:11.41 
mvrhel_laptop ok15:11.50 
paulgardiner henrys: yeah. Several points at which I was expecting problems didn't happen.15:12.14 
mvrhel_laptop but we can definitely say that we are able to do signatures now and not be fibbing about it15:12.21 
  though15:12.26 
paulgardiner I was particular dreading finding that our own checking code accepted the signatures we generated, but Reader rejected them. But Reader seems happy.15:13.08 
Robin_Watts paulgardiner: pdf_signature_set_value, v needs to be NULL initially.15:14.35 
paulgardiner My surprise was palpable.15:14.35 
tor7 paulgardiner: that's awesome news!15:14.43 
paulgardiner Robin_Watts: missed an initialisation needed for error handling?15:15.22 
Robin_Watts yeah.15:15.28 
paulgardiner Right ta15:15.32 
henrys Robin_Watts: best to put mupdf stuff on hold and do the P1 (dreaded) sorry about that. I thought any of us working on it would have taken substantially longer than you so we saved it.15:15.38 
Robin_Watts henrys: yeah, no problem.15:16.05 
henrys tor7:how's OpenGL?15:16.57 
tor7 henrys: there is the release to do; we haven't gotten any feedback on the release candidate so I think we're pretty much good to go. there is one makefile fix on tor/master that should probably go in though (to make curl support optional)15:16.58 
henrys tor7:oh gosh I forgot about that.15:17.15 
Robin_Watts tor7: Does it build for you on macosx with curl in place?15:17.34 
  I don't understand how pipitas can be getting the error he was seeing.15:17.56 
  but I haven't tested it yet.15:18.14 
tor7 Robin_Watts: ah, no. haven't been using my mac since I got a proper linux workstation set up.15:19.15 
Robin_Watts we should check it in case it's a silly macos ism.15:21.32 
tor7 yeah. I'm getting dismal speeds from casper today...15:21.57 
henrys tor7:the release is not contingent on the mac os x issue is it?15:24.12 
tor7 henrys: I think it is.15:24.24 
  it doesn't build out of the box on macosx as is :(15:24.31 
Robin_Watts henrys: having the default build fail on macosx would be bad.15:24.32 
henrys right15:25.30 
  does this mean restarting marcosw's prerelease testing again.15:25.49 
Robin_Watts henrys: Probably not.15:26.13 
  It's a config thing in curl.15:26.23 
  he doesn't test the curl build.15:26.28 
mvrhel_laptop henrys: I am going to have to leave the 9:00 am meeting a little early today (like around 9:15)15:27.24 
tor7 Robin_Watts: the curl_config.h stuff doesn't fly at all on macosx. damn curl has a lot of autoconf voodoo code #ifdef hell all over the place.15:27.25 
henrys mvrhel_laptop: oh sure15:28.01 
Robin_Watts tor7: Autogen a macosx one, then compare them, and hand hack with #ifdef MACOSX ?15:28.18 
tor7 Robin_Watts: the easy way out for now would be to remove mupdf-curl from the "default" target in the make file15:28.28 
  Robin_Watts: yeah, that's probably what we'll need to do15:28.41 
Robin_Watts tor7: Give me a couple of hours to try for a working solution.15:28.56 
tor7 I'd like to prune down the number of #define HAVE_xxx to the minimum required to reduce the risk of future breakage though15:29.01 
Robin_Watts tor7: I thought I had.15:29.10 
tor7 the USE_xxx stuff is trimmed, not the HAVE_xxx feature tests15:29.22 
Robin_Watts or at least, I'd shrunk the build options.15:29.25 
  right.15:29.28 
tor7 it's HAVE_STROPTS_H that tripped pipitas up15:29.40 
  fixed that one, and another one falls down15:29.46 
sebras Robin_Watts: oh, tickets! and welcome home!15:29.58 
Robin_Watts paulgardiner: Rest of the last review looks OK to my jetlag addled brain.15:30.33 
paulgardiner Great. I'll remove the logo and commit15:31.03 
tor7 Robin_Watts: I have a small diff that allows curl to compile on macosx15:31.21 
Robin_Watts tor7: excellent.15:31.34 
henrys okay meeting adjourned but I guess most of us will be around for a while15:34.20 
  so if we do have a release problem that would require macosw testing to happen again we'd have quite a delay because of his vacation. Unless somebody else knows how to start the tests.15:35.17 
chrisl henrys: that thought had crossed my mind, too.....15:35.58 
henrys I'll add a note to the agenda that there should be instructions for that somewhere.15:36.40 
chrisl henrys: running the tests is only part of it: there's checking the results, too15:38.27 
Robin_Watts henrys: As I say, the only issue we have so far is the curl one, and that isn't covered by marcosw's tests anyway, so I think the release is good to go with that added in.15:39.03 
henrys Robin_Watts: that's good.15:39.27 
  chrisl: yup it should be documented nonetheless15:39.48 
chrisl Does Marcos do any mupdf release testing?15:40.18 
henrys chrisl: I thought he did15:41.15 
henrys makes coffee before the next meeting15:41.40 
chrisl I didn't think mupdf had "weekly regression testing" like GS does15:42.16 
tor7 Robin_Watts: I had to tweak /home/git/thirdparty/curl.git a bit so I can push15:49.10 
  chgrp -R gs-pub, chmod +s on the directories, and "git config core.sharedrepository 1"15:49.41 
ray_laptop kens: I saw above that you had a file with interpolation that runs slow with clist. 15:50.37 
kens ray its a but now15:50.50 
ray_laptop kens: ???15:51.12 
kens 69451415:51.15 
  bug*15:51.21 
ray_laptop oh, s/but/bug/15:51.26 
kens sorry on the phoine15:51.27 
tor7 maybe the +s mode isn't required with the sharedRepository flag, I think I set all the old repos "sticky" before I discovered that config option15:51.44 
ray_laptop kens: np. I'll post my comments on the bug then15:51.58 
tor7 Robin_Watts: tor/curl has the fix15:52.01 
kens ok15:52.02 
henrys ray_laptop:It is likely the customer will disable banding see the email discussion.15:53.58 
kens It is still an incredible performance hit though15:54.45 
kens is off phone now15:54.51 
henrys kens:yes15:55.32 
kens presumably if we were unable to use page mode (higher resolution) tghen it would come back again15:56.44 
mvrhel_laptop ray_laptop: I am going to be working on this transparency issue today15:58.07 
henrys mvrhel_laptop: you need to leave early anything else for the meeting?15:59.30 
Robin_Watts tor7: Urm... We shouldn't use //16:00.07 
  We should use /* #undef ... */16:00.34 
  but otherwise looks good to me.16:00.42 
ray_laptop mvrhel: OK thanks16:00.46 
mvrhel_laptop henrys: no nothing else. I am doing timing tests on the J files with the raspberry for the group we had the skype call with last week and working with ray on this trans issue for the customer16:00.49 
henrys the "J file"?16:01.21 
Robin_Watts mvrhel_laptop: Is that gs or mupdf on the rasp pi?16:01.25 
ray_laptop henrys: the JEITA test files16:01.41 
mvrhel_laptop J**.pdf files that ray gave me16:01.44 
henrys mvrhel_laptop: so no trip?16:01.47 
mvrhel_laptop yes those16:01.49 
ray_laptop Japanese printer test files16:01.58 
mvrhel_laptop oh no there is a trip. They are building a laser printer with a device about the same as the raspberry pi16:02.24 
  so timing test s on that are comparable16:02.37 
ray_laptop I put them on ghostscript.com:~ray/public16:02.46 
henrys oh okay well good luck16:02.50 
mvrhel_laptop Robin_Watts: gs on rasp pi16:02.50 
Robin_Watts mvrhel_laptop: Right.16:02.59 
mvrhel_laptop Robin_Watts: I may go ahead and put mupdf on it too16:03.13 
  I am bringing it on trip for demo. 16:03.25 
ray_laptop mvrhel: Did you discuss having a printer to demo ?16:03.48 
  small cheap inkjet for Takane-san to have16:04.20 
mvrhel_laptop ray_laptop: no I did not. That has slipped through the cracks. I will ping Takane-san16:04.30 
ray_laptop but if the customers are laser, it may not be that relevant16:05.49 
henrys chrisl: release news?16:06.07 
mvrhel_laptop ray_laptop: well that customer is laser, but others may not be16:06.16 
chrisl henrys: none. Barring anything new, I'll do the release tomorrow or Thursday16:06.33 
mvrhel_laptop I may see what I can get here and just bring with me16:06.34 
ray_laptop mvrhel: I thought the people you were seeing were the board manufacturers -- so there is a printer company targeted as well ?16:06.45 
mvrhel_laptop ray_laptop: we have a very full schedule with a pile of places. but you are talking about for the demo on the friday right.16:08.12 
kens OMG Hin-Tak suggesting we change all teh Luratech source....16:08.38 
ray_laptop mvrhel: The only printer company that Scott-san mentioned you were going to see was cust 86116:09.12 
mvrhel_laptop we are hitting at least 2 a day. Takane-san has put together a full schedule16:09.37 
ray_laptop mvrhel: I hope the traffic allows that. The trains are on time, but getting to the train station in the traffic can be a problem16:10.23 
mvrhel_laptop well luckily takane-san should know what he is doing16:10.40 
chrisl kens: I know, although, I do rather take his point - if the code relies on a particular signed-ness, why not use the facilities in the language to guarantee it's what you need?16:10.44 
mvrhel_laptop better than us16:10.45 
atulagrwl Is there a known bug with opening password protected file with mupdf?16:10.47 
kens chrisl I agree, but really, he should think a little about what he's suggesting!16:11.15 
ray_laptop on one trip, we ran for the train and missed it and had to cancel one visit. That's when scott-san decided to only have luggage with wheels16:11.37 
chrisl kens: bugzilla often seems to encourage brain disengagement...... :-(16:11.51 
kens :-)16:11.57 
atulagrwl I am getting "error: aes padding out of range" with every password protected file created using Acrobat.16:12.21 
henrys mvrhel_laptop: having takane-san will be huge, I remember last trip how many times I wished we had someone to lead us around.16:12.47 
kens atulagrwl : best bet is to open a bug report and attach a smample file to look at16:12.51 
atulagrwl kens, Do you happen to know a way to create a password protected pdf with works well with mupdf.16:14.17 
kens atulagrwl : no16:14.27 
atulagrwl kens: Thanks :)16:14.36 
henrys anybody have anything else for the meeting, read the agenda - the in person meeting is near...16:14.38 
mvrhel_laptop ok. have to head out. bbiab16:15.01 
kens bye Micahel16:15.12 
henrys bye16:15.19 
chrisl Oh, henrys, I heard back from Norbert, and what I sent them get them up and running.16:15.57 
henrys chrisl: can you forward to support so we are all up to date?16:17.06 
  oh god another support email. marcosw is not allowed to go on vacation anymore ;-^16:17.38 
chrisl henrys: yeh, sorry..... I forgot16:17.41 
selim search desperately :) a way to generate a PDF by any one application with multiple copies16:20.16 
henrys meeting adjourned early, yeah!16:21.39 
kens THen I will run off, bye all16:21.49 
selim sorry I go through a 3g modem in the middle of Morocco 16:26.52 
chrisl henrys: I has Shelly been in touch about the OpenJPEG2 GS integration?16:31.30 
selim I am the only one who has this problem ?16:31.57 
chrisl selim: several copies of the same file isn't a requirement I've ever seen before..... seems like a waste of space, to me16:32.35 
henrys chrisl:I just follow the bug discussion - last I looked zeniko and shelly were collaborating.16:33.16 
chrisl henrys: okay, there's a bit more been happening behind the scenes: there was some stuff that Alex added in the 1.5 code that doesn't appear to have ever been fed back to the OpenJPEG devs, and wasn't really documented - to do with identifying channels in the decoded output.16:34.52 
selim you are certainly correct in the background but if you ask me 2 pdf I have to generate 2 pdf16:35.15 
ray_laptop selim: you can have gs print out the NumCopies for you to use afterwards16:35.40 
chrisl henrys: so, Shelly is trying to get some feedback from the OpenJPEG devs if there is an "official" way to differentiate between (specifically) color sample, opacity and "unknown" channels. If there isn't he's going to see if there is a way to add that to the OpenJPEG API without breaking the ABI, and had them a patch.16:37.21 
selim sorry I did not understand can you rephrase your answer (I'm french I master limited English sorry)16:37.37 
  sorry I did not understand can you rephrase your answer (I'm French I can speak little English sorry)16:37.59 
henrys chrisl:tell shelly behind the scenes is bad for his wallet. Put the information in the bug report, he'll get a higher return with more documented work.16:38.43 
chrisl henrys: fair enough. TBH, I don't think he expected it to take so long to get an answer from OpenJPEG - they really can be glacial.....16:39.34 
henrys chrisl: I mentioned before that patches look like they are being done by non owners I wonder if it is soon to be abandoned and if we should just fork it.16:40.51 
chrisl henrys: I'm not sure, I'll try to keep an eye on what's happening with it. I *strongly* suspect that the main developers use it for their application (whatever that might be) and are mainly interested in what they need out of it16:42.18 
ray_laptop for example: add the command line option: -c "<< /NumCopies 1 /EndPage { exch pop 2 eq { (numcopies.txt) (w) file currentpagedevice /NumCopies get 10 string cvs writestring false } { true } ifelse } setpagedevice" -f16:45.00 
  selim: the above was for you16:45.09 
henrys ray_laptop:ah you many know an evil trick to do this: I want to insert a blank page after each page in a pdf (or something similar) first attempt was to redefine EndPage to do a show page - infinite loop - execstack overflow. Do you know how to get around that?16:46.45 
ray_laptop selim: you can put that into a file "storecopies.cmd" and just put @storecopies.cmd on the command line if you want16:46.48 
atulagrwl Anybody having some experience in MuPDFActivity.java file? I might have found an issue and wants to validate the change I am thinking about it.16:48.19 
selim @C:\pdf.rsp -sOutputFile="C:\Documents and Settings\user\Bureau\%d.pdf" -c "<< /NumCopies 1 /EndPage { exch pop 2 eq { (numcopies.txt) (w) file currentpagedevice /NumCopies get 10 string cvs writestring false } { true } ifelse } setpagedevice" save pop -f - this is good?16:48.46 
ray_laptop henrys: redefine /showpage with: /showpage { systemdict /showpage load dup exec exec } bind def ?16:48.51 
  henrys: I haven't tried it. might work16:49.55 
chrisl ray_laptop, henrys: that will probably only work for Postscript input, it will probably require more trickery to work with PDFs16:49.59 
henrys yeah PDF is the problem.16:50.17 
ray_laptop but the pdf interpreter may have already bound in the showpage16:50.36 
chrisl henrys: Is it urgent?16:50.44 
henrys chrisl: no not at all - top of the head question.16:51.00 
chrisl henrys: I'll have a play tomorrow, I'm sure there's a way16:51.20 
selim ray_laptop my redmon parameters are good ?16:51.45 
henrys 3g in the middle of morocco, that's interesting, what city?16:53.20 
selim Agadir16:53.53 
ray_laptop selim: try it and see ?16:54.23 
selim it generates a blank PDF16:55.21 
ray_laptop henrys: you can modify pdf_ops.ps to put the extra showpage in the definition of /endpage16:55.57 
chrisl henrys: something like this *seems* to work: http://pastebin.com/HWL3wgkU16:57.05 
ray_laptop henrys: if you want to make it optional, put the modified pdf_ops.ps in a known directory e.g. junk/ then add -Ijunk/ on the command line16:57.27 
chrisl ray_laptop: see above...16:58.24 
henrys chrisl: ah yes… makes sense16:58.47 
chrisl henrys: *many* disclaimers about not tested etc - and also adds an extra blank page at the end, but close :-)16:59.36 
ray_laptop chrisl: I thought henrys wanted a blank page after every page. Why is an extra blank page at the end a problem ?17:00.22 
  i.e. print a single page file and get output which has page 1 and one blank page17:01.24 
chrisl ray_laptop: I wasn't sure, I thought he might just want a blank page *between* all the marked pages17:02.06 
ray_laptop selim: which file is blank (your pdf file or numcopies.txt) ?17:02.50 
  selim: note that I didn't bother writing an eol in numcopies.txt17:03.52 
selim the PDF and I do not have the file numcopies.txt17:04.37 
ray_laptop selim: why is there a 'save pop' before the '-f -' in your example ?17:07.53 
selim sorry but I removed it and it did not change anything17:09.19 
ray_laptop and if you remove the -c and the string in " " the PDF comes out ?17:10.03 
selim still the same17:12.13 
ray_laptop selim: so you must not be running gswin32c correctly. Are you getting the output file, and it is a blank page PDF? If so, then ghostscript probably didn't get the input PS file17:14.33 
  if the input comes from stdin, then just use '-' not '-f -'17:15.05 
  selim: but -f - works OK for me the same as just - alone17:16.27 
selim @C:\pdf.rsp -sOutputFile="C:\Documents and Settings\user\Bureau\%d.pdf" << /NumCopies 1 /EndPage { exch pop 2 eq { (numcopies.txt) (w) file currentpagedevice /NumCopies get 10 string cvs writestring false } { true } ifelse } setpagedevice -17:16.35 
ray_laptop selim: that is missing the -c and the " characters around the string17:17.45 
  selim: and you are missing the >> before setpagedevice17:18.27 
  selim: it should be:17:19.49 
  @C:\pdf.rsp -sOutputFile="C:\Documents and Settings\user\Bureau\%d.pdf" -c "<< /NumCopies 1 /EndPage { exch pop 2 eq { (numcopies.txt) (w) file currentpagedevice /NumCopies get 10 string cvs writestring false } { true } ifelse } >> setpagedevice" -17:19.51 
selim @C:\pdf.rsp -sOutputFile="C:\Documents and Settings\user\Bureau\%d.pdf" -c "<< /NumCopies 1 /EndPage { exch pop 2 eq { (numcopies.txt) (w) file currentpagedevice /NumCopies get 10 string cvs writestring false } { true } ifelse } >> setpagedevice" -17:19.55 
ray_laptop selim: and if you want the EOL in the .txt file then add: (\n) concatstrings before writestring17:21.29 
selim thank pdf is filled but there a single PDF file and no numcopies.txt17:21.31 
ray_laptop selim: the numcopies.txt may need an absolute path for you to be able to find it easily. change: (numcopies.txt) to (C:/Documents and Settings/user/Bureau/numcopies.txt)17:24.07 
mvrhel_laptop ok back and working on trans now17:24.27 
ray_laptop cheers mvrhel on !17:24.46 
selim @C:\pdf.rsp -sOutputFile="C:\Documents and Settings\user\Bureau\%d.pdf" -c "<< /NumCopies 1 /EndPage { exch pop 2 eq { ("C:\Documents and Settings\user\Bureau\numcopies.txt") (w) file currentpagedevice /NumCopies get 10 string cvs writestring false } { true } ifelse } >> setpagedevice" -17:28.31 
  is it good ?17:31.11 
  I filled out the pdf but not numcopies.txt17:41.35 
ray_laptop selim: do NOT use \ in the numcopies.txt filename string '\' is an escape character in PS strings. Use ("C:/Documents and Settings/user/Bureau/numcopies.txt")17:53.44 
  oops. and NO "'s. Should be: (C:/Documents and Settings/user/Bureau/numcopies.txt)17:54.17 
  the stuff following the -c is PostScript code, and has to follow PS rules!17:54.46 
selim @C:\pdf.rsp -sOutputFile="C:\Documents and Settings\user\Bureau\%d.pdf" -c "<< /NumCopies 1 /EndPage { exch pop 2 eq { (C:/Documents and Settings/user/Bureau/numcopies.txt) (w) file currentpagedevice /NumCopies get 10 string cvs writestring false } { true } ifelse } >> setpagedevice" -17:56.45 
  and no numcopies.txt17:57.07 
  ?18:13.33 
ray_laptop selim: don't know what to tell you. I tried that (cut and paste your line) and it WORKSFORME18:22.55 
selim what is worksforme18:28.43 
ray_laptop mvrhel: how's it going. Do you have any hair left ?20:32.22 
  it's frustrating that it is so close, except for that one type of file.20:32.54 
  ahh. For bug 694514 there is a really easy change to not use HL images when interpolated and the device is the pattern-clist.20:39.13 
  Trying that on the cluster. Fixes the performance hit -- with -dBufferSpace=64m we are 35 seconds for all 6 pages vs. 31 seconds for page mode20:54.04 
  henrys: BTW, we can use 1000m or 1g instead of 100000000020:54.36 
mvrhel_laptop hi ray_laptop: with my most recent fix, I have your original file working, the file that you broke working another file that I broke working and trying now to figure out this last broken file21:18.19 
ray_laptop mvrhel: OK. I think I follow that. You have fixed some cases, but still have a (new) broken file ?21:24.08 
henrys ray_laptop:right about the abbreviations, I forgot.21:41.45 
  ray_laptop:I do believe I saw worse times than you on henrysx6 which is reasonably modern. - I know I left and came back in a while and it was still going.21:42.35 
ray_laptop henrys: well, 1109 seconds _is_ a long time. Getting 35 seconds now with -dBufferSpace=64m and my patch. Even the default is decent at 83 seconds21:50.15 
henrys ray_laptop:right.21:51.04 
ray_laptop darn. cust 532's gxblend1.c is so much different to ours that my first patch broken something (I tried just limiting how much I changed)21:51.28 
  I used their flaky comparison tool and it didn't show me the correct differences :-(21:52.22 
  I guess selim gave up. Seems like a funky request anyway.21:53.35 
  I need to run home. bbiab.21:54.04 
mvrhel_laptop ray_laptop: yes22:04.07 
Robin_Watts tor7: The patch from bug 694515 looks good to me.22:35.42 
tor7 Robin_Watts: yes. feel free to take it on.22:36.45 
Robin_Watts and bug 694516 looks (at first sight) like it should be fixed for the release.22:37.05 
  it's android specific, so won't affect marcosw's testing.22:37.19 
tor7 does 694516 only happen on android?22:37.37 
Robin_Watts AIUI, yes.22:37.52 
tor7 right. should probably be fixed by release.22:38.03 
  so now we have two issues that ought to be resolved.22:38.10 
  I'll make a proper fix for the curl stuff tomorrow22:38.24 
Robin_Watts tor7: Did you sort tickets etc for coming over for the wall?22:38.27 
tor7 Robin_Watts: we have flights sorted out, yes. but not hotels.22:38.46 
Robin_Watts I spoke to sebras before I went on holiday and volunteered a spare room and a sofa bed here if you want them.22:39.16 
tor7 were you able to put us up for the friday night, then we can hopefully find a place in london for saturday night.22:39.19 
  (we don't want to impose too much)22:39.39 
Robin_Watts We are 90 miles from london, so might not be massively convenient, but...22:39.39 
tor7 I certainly plan on visiting you this time! Convenience be damned :)22:40.00 
Robin_Watts Helen has a concert on saturday, I think, so we could meet up with her in London after our respective concerts.22:40.28 
  When do you fly out?22:40.51 
tor7 arrive sep 13 at 15:00, depart sep 15 at 19:0022:42.00 
Robin_Watts might be better to come back with us after the concert on the saturday and stay here then. We could then go visit oxford on the sunday, from where it's easier to get back to london or something? All depends on timings.22:42.02 
tor7 that could work22:42.27 
Robin_Watts That might make more sense - you could stay in London on friday night, look around on saturday, we could go to the gig, travel back here, stay here, then go to oxford on sunday and we can get you on a bus to the airport from there.22:43.52 
sebras tor7: pling!22:44.15 
tor7 sebras: see robin's suggestion above. sound good to you?22:44.27 
Robin_Watts Will discuss this all with Helen tomorrow, as she has a good head for this stuff. We just need to make sure we can find somewhere to leave your stuff securely as you won't want to carry it during the concert :)22:45.11 
  but I must go to bed now... night!22:45.24 
tor7 Robin_Watts: I plan on bringing as little as possible!22:45.53 
  Robin_Watts: alright. good night!22:46.00 
sebras tor7: Robin_Watts: yup, seems reasonable.22:46.22 
haggl_ I added a second patch to the bug tracking system, which implements continuous scrolling for the 'j' and 'k' keys. I think many people would like to have that.. :-)23:25.44 
ray_laptop mvrhel: for the logs. Looks like some issues. I don't think Altona_Technical_v20_x4.pdf is right23:59.31 
 Forward 1 day (to 2013/08/14)>>> 
ghostscript.com
Search: