| <<<Back 1 day (to 2014/09/21) | 2014/09/22 |
hdon | hi all :) i'm using "exit" to preempt "loop" and "for" loops. how can i "exit" from more than one nested loop? | 00:54.08 |
Amyn | Hey I am using GS 9.14 for PDF to Image Conversions. I get the following error when my application calls gswin32c.exe multiple times | 04:42.19 |
| "C++ Runtime error R6016 -not enough space for thread data" | 04:42.30 |
| Any help? | 04:43.02 |
kens | Amyn we don't use C++, so the error likely comes from your own app. However, we can use threads, so we may be exhausting some WIndows resource (I note you are using the Windows executable, though you don't say this is on WIndows). That said, nobody is likely going to be able to help if you don;t give us more details, in particular how are you executing Ghostscript ? | 06:55.29 |
| hdon, use another exit. Each exit only terminates the innermost execution context. See the PLRM p585 the exit operator. | 06:57.26 |
Amyn | i have a perl script and i convert that script to exe using perl2exe. what it does is that it takes in the pdf file name and the output directory where the pdf will be converted to images using gs 9.14 | 06:58.27 |
| the perl exe is run from visual studio c# | 06:58.49 |
kens | Amyn, and teh command line you are using ? THat's more important than anythign else. | 06:58.50 |
Amyn | meaning command line i am using? | 06:59.05 |
kens | Yes | 06:59.09 |
| The *Ghostscript* command line | 06:59.18 |
Amyn | yes the perl script calls gswin32c.exe to convert pdf to image | 06:59.39 |
kens | And what is the Ghostscript command line ? | 06:59.49 |
Amyn | u mean what is the command that my perl script runs? | 07:00.09 |
kens | Yes! | 07:00.17 |
Amyn | $cmd = "$version1\\bin\\gswin32c -I" . "$version1\\lib;fonts -dSAFER -dNumRenderingThreads=2 -dBATCH -dUseCropBox -dNOPAUSE -sPDFPassword=\"$password\" -sDEVICE=tiffg4 -r" . $resolution . " -sOutputFile=\"$whereto\\$outf\" \"$toOpen\" >> \"$whereto\\conv.tmp\" 2>> \"$whereto\\conv.err\" " ; | 07:01.22 |
| $version1 = "gs9.14" | 07:01.34 |
kens | And now without all the missing variables please, I don't speak Perl | 07:01.43 |
Amyn | "gs9.14\\bin\\gswin32c -Igs9.14\\lib;fonts -dSAFER -dNumRenderingThreads=2 -dBATCH -dUseCropBox -dNOPAUSE -sPDFPassword=\"\" -sDEVICE=tiffg4 -r300 -sOutputFile=\"output\" >> \"output\\conv.tmp\" 2>> \"output\\conv.err\" " ; | 07:03.07 |
kens | Well at 300 dpi I would not expect NumRenderingThreads to be making any difference, so you could simply drop that | 07:03.33 |
| It seems likely to me that your script is ending up with multiple Ghostscript processes running, have you check that with Task Manager or Process Monitor ? | 07:04.22 |
| How many executions are required in order to exhibit the problem ? | 07:04.46 |
Amyn | well that i am not aware of. people using my application have reported that some users are getting this error | 07:05.32 |
| i am not getting this error. i have executed this many many times (100 or more times) | 07:06.10 |
kens | THen you need to go back to your users and get them to tell you how to reproduce the problem, then you can investigate it. When you can reproduce the problem, we can discuss it reasonably, without that you can't answer our questions. | 07:06.18 |
Amyn | yes i agree completely. till then do u have any clue where might be the problem? | 07:06.58 |
kens | As I said, I believe the most likely problem is that there are too many Ghostscript processes running and that is exhausting 'something' in WIndows. | 07:07.15 |
| I seem to remember that (in old versions of Windows at least) there was a limit on the thread pool. | 07:07.43 |
Amyn | but i am not using multi threading. at any given time, my application only runs 1 instance of that perl script and waits for it to end | 07:09.19 |
kens | You are running Ghostscript with NumRenderingThreads=2 so you *are* running Ghostscript with threading. | 07:09.50 |
Amyn | yes but my application does not run multiple perl script at one time. yes gs uses multi threading but my application waits for those threads to end | 07:12.34 |
kens | According to this site: | 07:13.18 |
| http://support.microsoft.com/kb/126709 | 07:13.19 |
| THat doesn't matter because the Thread Local Storafe isn't freed | 07:13.31 |
| "The thread local storage (TLS) is not freed by the system. | 07:13.52 |
| The failure occurs only if there is another Win32-based application active while you are doing the spawns. The message itself is not generated by Win32s. It is generated by the Microsoft C Run-time (CRT) libraries LIBC.LIB and LIBCMT.LIB. " | 07:13.52 |
Amyn | i have already gone through this but couldn't understand it | 07:16.58 |
| how do i make sure that TLS is freed? | 07:17.13 |
kens | You can't | 07:17.32 |
| The TLS is only freed by the Win32 DLL during cleanup | 07:17.52 |
Amyn | so meaning i can't do anything to avoid this error? this is a windows error and not my application error? | 07:18.51 |
kens | There's code in there which claims to instantiate TLS only once, I'd suggest you try that. | 07:19.46 |
| However, the article also says that version 1.25 of Win32s doesn't have the problem either. SO perhaps you need to make sure your users have at least that version installed. | 07:20.22 |
| In any event, I can't see that this is a Ghostscript problem. | 07:20.45 |
sebras | good morning. | 07:21.14 |
kens | Hi sebras | 07:21.20 |
Amyn | ok. thanks for the help :) | 07:23.33 |
kens | Hmm, win32s was apparently part of WIndows 3.1 and 3.11, it seems unlikely this can reallyt be the problem, surely your users are not using such old versions of Windows..... | 07:24.24 |
| I wonder if the problem really exists in Perl or something. | 07:25.33 |
| Aha, or possibly perl2exe, I wouldn't be surprised if that's the source of the problem. | 07:25.59 |
Amyn | windows 3.1 is which version of windows? i ,mean is it xp or 200 or 98? | 07:28.17 |
kens2 | Amyn, its WIndows 3.1...... | 07:32.17 |
| But since it actually affects the MSVC C++ Run time DLL (msvcrt20.dll) it doesn't matter which version of Windows you are running. | 07:33.11 |
| It only matters that a process is using that DLL. My money is on perl2exe using it, possibly in order to maintain ancient compatibility, possibly because they don't realise there's a problem. | 07:34.07 |
| In any event, this really isn't anything to do with us. | 07:34.32 |
Amyn | ok i will look into the causes of the error | 07:35.10 |
kens2 | Well bug #695354 is *still* crashing after Ray's fix, but not before. I assume this is due to all the white space changes in pdf_draw.ps moving the memory around enough to trigger some other fault :-( | 09:43.40 |
| And it *doesn't* crash on Linux....... | 09:57.45 |
chrisl | <broken record mode> what about -Z@ ?? | 10:01.56 |
kens2 | Makes no difference on Windows, crashes either way (trying Memento now) forgot to check on Linux | 10:02.20 |
| I did try under ddd and that didn't crash either, give me a mintue and I'll try -Z@ | 10:02.34 |
| You're correct,. -Z@ *does* cause the Linux executable to seg fualt | 10:04.25 |
chrisl | I really need to get the 9.15 release done today, but I look at 695354 more tomorrow | 10:05.24 |
kens | Stupid network glitch. | 10:06.47 |
| As I was trying to say.... It faults in do_validate_object because 'ptr' is 0xa1a1a1 which I assume means its a reerence to somethign which has already been freed. | 10:07.32 |
chrisl | Er, no, I think "a1" is just allocated.... | 10:09.04 |
kens | Oh, that's surprising then I guess. | 10:09.20 |
| <sigh> Its outline_levels in the PDF device structure again. I'll plod on with it. | 10:09.58 |
chrisl | Well, as I said, I really need to get the 9.15 release done today, but I can look at 695354 more tomorrow | 10:10.07 |
kens | NP, I could do with learning to debug these anyway. so I'll do this till I get bored, then go and try to fix Ray's commit properly for pdfwrite. | 10:10.46 |
| In this case I suspect bored = too annoyed with ddd | 10:11.06 |
chrisl | Use your remote debugger thingy | 10:11.26 |
kens | Yes, that's next on my list, just need to pay for it (I keep forgetting) | 10:11.50 |
| I'm currently running the test 3 ways on 2 OS'es, as soon as I get that reduced I'll sort this out | 10:12.37 |
Robin_Watts | Guys, a quick strawpoll.... | 10:36.02 |
| http://ghostscript.com/~robin/mainiconhist2.png | 10:36.10 |
| or | 10:36.11 |
| http://ghostscript.com/~robin/mainiconhist3.png | 10:36.14 |
kens | his2 | 10:36.29 |
| hist2 | 10:36.33 |
| brighter and shinier is better, right ? :-) | 10:36.42 |
Robin_Watts | There is a difference to the gradient, and a difference to the color. | 10:36.55 |
| I prefer the gradient in 3, but the color in 2. | 10:37.09 |
kens | Yeah, but I think those are too subtle to be important | 10:37.13 |
| Let me download it so I can zoom | 10:37.37 |
tor8 | Robin_Watts: mainiconhist3.png | 10:37.43 |
| the red in 2 is illegible, too saturated it just blows out | 10:38.02 |
Robin_Watts | Consensus seems to be on the darker red (on skype) | 10:38.25 |
tor8 | might look okay on non-IPS and non-[PM]VA screens though | 10:38.26 |
| Robin_Watts: yeah, that bright red is hard on the eyes | 10:38.36 |
kens | shrugs | 10:39.48 |
Robin_Watts | http://ghostscript.com/~robin/mainiconhist4.png | 10:41.22 |
| compromise. | 10:41.24 |
| Now the red in the text matches the red in the cube. | 10:41.35 |
kens | No need to compromise if the consensus is there | 10:41.44 |
| 4 is better than 2 | 10:42.32 |
Robin_Watts | I still think 3 is more readable than 4. | 10:43.02 |
kens | Probably true, but its legible to me easily enough and shiny is better | 10:43.22 |
tor8 | 4 is acceptable | 10:44.02 |
kens | Black would be even more legible of course | 10:44.02 |
tor8 | looser letterspacing would probably help as well | 10:44.40 |
kens | Yes, though a large part of the problem is the angled text, there just aren't enough pixels | 10:45.04 |
tor8 | it might help not using faked small caps? | 10:45.34 |
Robin_Watts | I could change to a thinner font. Arial rather than Arial Black. | 10:45.38 |
| I felt Small Caps was more legible. | 10:45.52 |
kens | I'm not sure its worth it, if people prefer the darker red, then use that | 10:45.56 |
tor8 | the strokes in the T are so much fatter than the rest | 10:45.57 |
| it just looks a bit odd, but now I'm being picky :) | 10:46.10 |
| Robin_Watts: Arial Bold maybe | 10:46.21 |
kens | Its about 9x9 pizels, how good can it be ? | 10:46.26 |
tor8 | should still be thinner than arial black | 10:46.30 |
| Robin_Watts: try all caps instead of small caps? | 10:46.52 |
| Robin_Watts: while you're here, your tiff patch is good to include? | 10:47.12 |
kens | caps will be better than lower case, at least its all straight lines this way | 10:47.16 |
tor8 | kens: true. | 10:47.24 |
Robin_Watts | tor8: yes. | 10:49.43 |
| http://ghostscript.com/~robin/mainiconhist5.png | 10:55.42 |
kens | Looks clearer to me, will just grab a local copy | 10:56.06 |
| Definitely clearer | 10:56.33 |
| The spacing looks a bit odd, but with so few pixels, I'm not surprised | 10:57.40 |
Robin_Watts | kens: which bit of spacing? | 10:57.56 |
kens | And its no worse than before | 10:57.59 |
Robin_Watts | I kerned the TR to move them together a bit. | 10:58.05 |
kens | The Tr seesm to have more space than the RI | 10:58.20 |
| So it almost looks like "T rial" | 10:58.44 |
| But that was true before as well | 10:58.55 |
Robin_Watts | http://ghostscript.com/~robin/mainiconhist6.png | 11:01.13 |
kens | Well, I think that's the best yet, what does tor8 think ? | 11:02.11 |
| I also like that the 'A' doesn't appear to have horns in that one | 11:03.11 |
Robin_Watts | kens: reload it. I moved the text very slightly. | 11:03.52 |
tor8 | Robin_Watts: kens: I like the font best in 5, but the color in 4 has more pop to it while retaining legibility | 11:04.18 |
kens | I've been storing them locally for a better look outside the browser, I'll get a new copy | 11:04.19 |
Robin_Watts | tor8: Let me put the color in 4 back. | 11:04.54 |
kens | Well, I preferred it without the horns. | 11:05.05 |
| But htat's very picky really | 11:05.29 |
Robin_Watts | http://ghostscript.com/~robin/mainiconhist7.png | 11:05.57 |
kens | The font in 5 and 6 is the same I think ? | 11:06.06 |
Robin_Watts | No horns, color of 4, spacing of 6. | 11:06.12 |
| same font, different spacing. | 11:06.31 |
kens | THat's what I thought, well I like any of them. If I had to choose I'd probably go with 7 | 11:07.00 |
| Of course, I've no idea how this looks on a mopbile device | 11:07.53 |
| Or even a mobile device | 11:08.02 |
tor8 | Robin_Watts: a handful of new commits on tor/master, fixing one issue and preparing for the release | 13:24.39 |
| we could probably strip the dtoa() function (and keep only strtod and ftoa) | 13:26.16 |
| but I like having it there for completeness sake | 13:26.24 |
| bah, no, let's not keep dead code... | 13:29.36 |
henrys | mupdfers: fred's made good progress on the qt port https://docs.google.com/a/artifex.com/file/d/0B0-CKuptzfNcY0RNV3NEYng3cWM/edit | 13:29.46 |
tor8 | henrys: nice progress, but needs a bit of UI polish (like retaining the current position when zooming) | 13:32.05 |
| still, it's good progress | 13:32.13 |
henrys | did we ever get his branch squared away? | 13:36.21 |
tor8 | henrys: yes. but he's messed up file name capitalization, which is not noticeable on MacOS X and windows with their case insensitivity. | 13:37.35 |
| so half the files are in 'platform/qt' and half in 'platform/Qt' | 13:37.48 |
| should be an easy fix with git filter-branch | 13:38.42 |
| or just fix at the end if we squashed his commits before pushing onto master | 13:39.20 |
rayjj | morning, all | 13:44.20 |
kens | morning rayjj, I have a question for you when you are up to speed | 13:44.38 |
rayjj | kens: in your commit you mention "the parent CIDFont 'glyphs' space is shared by all the descendants, so we *don't* want to free that when we free the descendant." When do the glyphs get freed (i.e., do we leak them)? | 13:45.35 |
| kens: now, your turn for a question :-) | 13:45.48 |
kens | THe glyphs are freed when we free teh CIDFont in free_copied_font | 13:46.05 |
| rayjj when you do the begintransparencygroup for the /SMask /None you don't reset pis->soft_mask_id is there a reason for that ? | 13:46.41 |
rayjj | kens: no reason -- just an oversight ? | 13:47.22 |
| kens: it makes sense, and wouldn't hurt to set it to "no_id" | 13:47.51 |
kens | COuld be :-) The reason I ask is that if I *don't* reset it, then the GState code for pdfwrite doesn't know that its changed and so does not emit a /None. | 13:47.52 |
| I was concerned that there might be a good reason not to reset it to no_id | 13:48.26 |
rayjj | kens: AFAIK, the soft_mask_id is there exclusively for high level devices (i.e., pdfwrite) | 13:48.56 |
| the pdf14 code and the clist code don't need it, relying on the sequence of compositor actions | 13:49.27 |
kens | For the CIDFonts, the CIDFont shares its glyp[h space with all the descendants, so if we freed it in any decendant we would multiply free it. Since we are going to free the parents glyph space anyway (in free_copied_font) we don't need to (and in fact must not) free it in the descendants | 13:49.30 |
| rayjj OK then I will either reset it in pdfwrite or back in the interpreter if I can find a good place. Right now I'm save/restore'ing the id manually in order to preserve it after I reset it to 0 | 13:50.14 |
rayjj | kens: I see, so this was only skipped for the descendant fonts not the parent. thanks | 13:50.22 |
kens | rayjj, correct, its only the sdescendants that don't free the glyph space (there's a new routine to avoid doing so). chrisl did poitn out that setting the pointer to NULL owuld have the same effect. | 13:51.01 |
| That's the glyphs pointer in the copied descendant font | 13:51.39 |
rayjj | kens: re soft_mask_id, this is ONLY in the handling of the pdfwrite transparency compositor code in gdevpdfg.c -- the rest of the graphics lib doesn't use it | 13:53.10 |
kens | In that case, I may just clear it in pdfwrite. | 13:53.27 |
| It'll be the easiest change | 13:53.40 |
rayjj | kens: I grepped for it in base/*.c and it doesn't exist | 13:53.41 |
kens | THat's fair enough, I hadn't got that far yet. | 13:53.52 |
rayjj | but it _was_ found in gdevpdfg.c and gdevpdft.c | 13:54.06 |
kens | I have a preliminary fix doign a cluster push now | 13:54.15 |
| gdevpdfg uses it to determine if the SMask has changed, that's where the fact that it hasn';t changed was causing me a problem | 13:54.41 |
| I was going to grep for it about the time you came online, so I figured I would just ask instead ;-) | 13:55.31 |
rayjj | kens: probably more typing to ask me than to just type the grep ;-) | 13:56.08 |
kens | Well I was doing something else at the same time, and feeling lazy | 13:56.25 |
| I had expected it to be used elsewhere, seems wasteful to have it just for pdfwrite, but..... | 13:56.43 |
| I'm certainly not going to try and change it now | 13:56.53 |
| So a few diffs, and a few files producing errors | 13:59.55 |
rayjj | paulgardiner: I looked for that smartoffice apk you mentioned and don't see it. I only see one file: sol-docs-ios.zip | 14:10.25 |
| paulgardiner: I was going to load it on my Asus tablet | 14:10.50 |
paulgardiner | rayll, sorry should have specified, it's in my home directory rather than being available as a web download | 14:12.12 |
rayjj | paulgardiner: thanks. That means I have to put it somewhere else to be able to access it, since I don't have SSH on the Asus | 14:13.38 |
| paulgardiner: can you recommend a good SSH client for android :-) | 14:14.19 |
| SanDisk has a 512Gb SD card ! | 14:15.03 |
pedro_mac | yeah, its a bit pricey though - would you trust that amount of data to a card? ;) | 14:15.36 |
| rayjj: I use JuiceSSH - seems pretty good imho | 14:15.53 |
paulgardiner | I could stick it up for download protected by a password. | 14:16.00 |
pedro_mac | even though the icon is a lemon | 14:16.03 |
paulgardiner | rayjj: hang on. I'll just sort out web access to it | 14:17.17 |
| rayjj: Done. Details just emailed. | 14:18.39 |
rayjj | paulgardiner: no rush | 14:18.49 |
| paulgardiner: thanks for the SSH suggestion | 14:19.18 |
| that way I'll be able to log in to my raspberry from my tablet | 14:19.58 |
| ;- | 14:19.59 |
| oops | 14:20.12 |
paulgardiner | pedro_mac: yeah I use JuiceSSH. Not used it for scp though. Does it do that? | 14:20.13 |
pedro_mac | no, just ssh as far as I know | 14:31.02 |
fredross-perry | mawninâ | 14:32.27 |
rayjj | paulgardiner: pedro_mac: looks like "ServerAuditor" (https://play.google.com/store/apps/details?id=com.server.auditor.ssh.client&hl=en) has SFTP on one of their sample screens. | 14:33.43 |
henrys | it looks like folks are still open water swimming in Chicago, Miles and I will probably go. Bring your suits if you want to give it a go. | 14:33.47 |
rayjj | I'd have to check a suitcase to fit my spring suit | 14:34.37 |
Robin_Watts | henrys: What's the surf there like? :) | 14:34.40 |
kens | sigh, network glitch again | 14:35.05 |
| As I was saying, what's teh water temperature likely to be ? | 14:35.22 |
rayjj | Robin_Watts: depends on how big a splash gets made when someone throws in their beer bottle | 14:35.42 |
henrys | Robin_Watts: it really is quite surprising how much surf these lakes can have, but I've never seen surfing there. | 14:35.42 |
Robin_Watts | I think I will probably be avoiding swimming in open air waterways for a while. | 14:36.25 |
| At least until I get the blood tests back to confirm that I didn't catch anything from swimming in the Russian River :) | 14:36.54 |
kens | Dumb question, why would you think you did ? | 14:37.24 |
Robin_Watts | kens: A while ago, I had a nasty persistent blurred patch in my right eye. | 14:37.51 |
kens | Oh I see | 14:37.56 |
Robin_Watts | I went to see my optician who referred me on to a retina specialist. | 14:38.11 |
henrys | and there you have it: http://sbsurfandkayak.com/surf-lake-michigan-freshwater-surf/ | 14:38.26 |
Robin_Watts | He confirmed that i had a cottonwool spot which was causing it. | 14:38.37 |
kens | Well, they are awfully big as lakes go, plenty of time for the wind to make some surf | 14:38.55 |
rayjj | Robin_Watts: what's that ? something on the cornea ? | 14:39.02 |
kens | What's a cottonwool spot ? | 14:39.06 |
Robin_Watts | Possible causes are 1) tick borne diseases (including from texas), 2) waterborne things (including from northern cali) 3) other stuff, 4) shit just happens. | 14:40.15 |
| a cottonwool spot is a tiny bleed on the retina. | 14:40.29 |
rayjj | I found http://www.aao.org/theeyeshaveit/optic-fundus/cotton-wool.cfm | 14:40.32 |
kens | Oh ok | 14:40.34 |
Robin_Watts | They generally clear up, and indeed, mine has. | 14:40.50 |
jogux | henrys: when you say suits... do you mean wet suits? | 14:41.00 |
kens | I think this time of year bathing suits are OK | 14:41.15 |
henrys | we didn't use wet suits last year | 14:41.19 |
Robin_Watts | but they can do investigations into the cause of them, and I'm waiting for an appointment to check if I've got any of the underlying causes. | 14:41.24 |
kens | In winter you'd want a wet suit (maybe even a dry suit, it gets cold there) | 14:41.55 |
henrys | jogux: I'm thinking of bringing mine though since I have to check a bag anyway. I'll be there for a week. | 14:42.06 |
jogux | ah right | 14:42.20 |
Robin_Watts | Also, our hotel is a 40 minute L ride from the water. | 14:42.32 |
jogux | went swimming in the river out the back of ours at the weekend. though... not on purpose. | 14:42.33 |
Robin_Watts | wheras Henry/Miles will be staying right near the water for the tradeshow. | 14:43.06 |
henrys | Robin_Watts: ah yea I forgot about that. | 14:43.26 |
Robin_Watts | Do we know what hotel yet? | 14:43.35 |
kens | I thought it was more like an hour, maybe it just feels that way | 14:43.36 |
henrys | last year we jogged to and from the water. | 14:43.39 |
jogux | errr. the meeting is at the hotel miles booked me into... right? | 14:44.25 |
Robin_Watts | There is a park near the hotel where you can see deer etc. It's slightly surreal being in the middle of this huge metropolitan area, and suddenly jogging past a family of deer. | 14:44.27 |
| jogux: I'm sure it will be, yes :) | 14:44.40 |
kens | Sionce we'll be there too, we hope so :-) | 14:44.54 |
Robin_Watts | We will all be staying (and having the meeting) at an airport hotel. | 14:44.59 |
| but the tradeshow is downtown chicago, and is for the 5 days before the meeting. | 14:45.13 |
| So Miles/Henry/Scott will stay at a downtown hotel, then transfer out for the last 2 nights. | 14:45.33 |
jogux | ahh right. | 14:45.39 |
| is there a plan for Saturday? | 14:45.57 |
kens | Boredom. | 14:46.06 |
| Or more probably a trip into town | 14:46.13 |
| THe public art is worth a look if you haven't seen it | 14:46.25 |
Robin_Watts | jogux: There is a science museum in Chicago that is supposed to be good. | 14:46.36 |
henrys | I hear good things about the planetarium strange I've never gone, will try to get there this time. | 14:47.12 |
Robin_Watts | ooh. competing science museums. | 14:48.06 |
| And an aquarium. | 14:48.24 |
kens | Time for both | 14:48.26 |
| we have some time SUnday also I think | 14:48.35 |
| Yeah departure is 18:45 | 14:48.56 |
Robin_Watts | If we go to the "Field Museum" then the planetarium and Aquarium are nearby, as is the bean. | 14:49.02 |
kens | Well we all like the bean | 14:49.11 |
Robin_Watts | Cos Joseph will want to see the bean. | 14:49.16 |
kens | and it doesn't take long :-) | 14:49.21 |
jogux | bean? | 14:49.37 |
Robin_Watts | Supposedly there is a submarine you can visit somewhere too. | 14:49.45 |
kens | jogux, the coffee bean, you'll have seen it in films/TV | 14:50.00 |
jogux | 's flight out is at 16:10 on Sunday. | 14:50.05 |
kens | its some of the public art | 14:50.05 |
jogux | kens: ah, right :) | 14:50.09 |
Robin_Watts | jogux: basically, someone decided to make something that looks like a late 1980s ray traced scene. | 14:50.27 |
paulgardiner | My memoryof Chicago was meeting a self lookalike although even more deformed than the origial> Don't think I've ever seen the bean though. | 14:50.33 |
henrys | jogux: get a steak at a good restaurant also. | 14:50.51 |
jogux | henrys: you really want to 'stake' your reputation on an American steak being better than a good piece of air dried Aberdeen Angus? ;) | 14:51.53 |
| paul: :) | 14:52.02 |
Robin_Watts | jogux: Yes. | 14:52.37 |
henrys | jogux: I do ... although I was impressed with UK food when I was there, you hear such bad things, I've heard there has been something of a turnaround in UK cuisine. | 14:52.52 |
Robin_Watts | Best steaks I've had were in Argentina. Worst were in Chile. The ones in the states are consistently brilliant though. | 14:53.34 |
jogux | :-) | 14:53.47 |
henrys | and one of them will feed all of you ;-) | 14:54.05 |
paulgardiner | There's something about the UK that prevents bacon being cooked properly. Not a problem in the US thankfully | 14:54.13 |
jogux | hehe | 14:54.14 |
Robin_Watts | and there is a really good place in london, Goodmans, that is basically an american steakhouse over here. | 14:54.19 |
rayjj | I've had good steak in Japan, but the portions are a bit small | 14:57.22 |
| I have to take my son to school. bbiab | 14:58.26 |
kens | paulgardiner : ping | 15:20.30 |
paulgardiner | 64bytes from paul (80.229.137.79) icmp_seq=1 ttl=58 time=1 | 15:22.20 |
kens | paulgardiner : http://stackoverflow.com/questions/25970382/mupdf-highlight-text-to-same-area/25975834?noredirect=1#comment40675273_25975834 | 15:22.30 |
| someone complaining that "When I select a part of the text to highlight, then I highlight the same place again. The highlight text will be blurred" which sounds kind of familiar ? | 15:22.55 |
paulgardiner | I don't think I know of that one. I do know an effect that does something like that. If you use alpha for antialiasing and accidently render multiple times then text goes pixelated. | 15:25.54 |
| not blured though. | 15:26.01 |
Robin_Watts | I think that's what he means. | 15:26.15 |
kens | Well, its hard to tell but it sounds like he's using the standard MuPDF app | 15:26.19 |
paulgardiner | I wasn't aware that happened in any of our MuPDF apps | 15:26.36 |
kens | I could easily be mistaken, there is a language barrier..... | 15:26.55 |
paulgardiner | stackoverflow and its silly rules: are you supposed to ask for more info in a comment or an answer? | 15:27.48 |
kens | Want me to ask some more questions ? I did ask for an example file and didn't get one | 15:27.48 |
| Either works | 15:27.56 |
| I normally use comments | 15:28.01 |
paulgardiner | kens: knowing the OS would be handy: i.e., which of our apps. | 15:28.32 |
kens | OK I cna ask that. | 15:28.40 |
Robin_Watts | The only app we have that does highlighting is Android. | 15:28.52 |
| given that we haven't released gsview yet. | 15:29.00 |
| If you repeatedly highlight a section of text, it gets harder and harder to read as it's more and more washed out by the highlight, but I wouldn't call that blurred. | 15:29.42 |
paulgardiner | Oh. highlight annotation. Silly me | 15:29.57 |
kens | Well I suppose if he hadn't stopped highlighting then it could happen | 15:30.31 |
paulgardiner | We have a sort of bug in the way we combine the highlight color with the text | 15:30.56 |
| We should have no effect on black | 15:31.23 |
kens | It sounds like he is repeaedly highlighting the text | 15:31.26 |
paulgardiner | Yeah, and I can't think of a good reason to do so | 15:31.47 |
| I think both iOS and Android support highlighting | 15:32.02 |
kens | OK so I asked him for more info and told him that repeatedly highlighting (which is indeed a dumb thing to do) will cause the text to get fainter with each repetition | 15:32.37 |
| Hmm, anyone know where 'the FTP site' is on picas ? | 15:50.16 |
henrys | kens:did you intend to ignore code in that last commit or should it be returned? | 15:56.34 |
kens | I'm ignoring it. | 15:56.53 |
| I always want to return 0 from there | 15:56.59 |
| I suppose I should not bother with teh assignment to code | 15:57.53 |
Robin_Watts | kens: If you assign to code and then don't read it, clang will give a warning. | 15:58.13 |
| and clang is standard on MacOSX these days. | 15:58.20 |
kens | Yes I know, I had forgotten | 15:58.30 |
| Its a trivial fix | 15:58.37 |
Robin_Watts | Probably you should put a comment in there to avoid some busy body looking at that code and saying "I'll fix that!" | 15:58.45 |
kens | That would be me :-) | 15:58.59 |
Robin_Watts | (i.e. protect me from myself) | 15:59.01 |
kens | TBH pdfwrite is already full of these, so I'm not hugely bothered | 15:59.21 |
| So nobody knows where the ftp site is located on picas ? | 16:01.21 |
henrys | then you have ignoring return value if you remove code, but I guess that's widespread in our code. | 16:01.30 |
| kens: we probably have a login - why would we have an ftp server? | 16:01.53 |
kens | Yes, and some instances in pdfwrite are not fixable (adding assignment causes an error, removing it causes an error) | 16:01.58 |
| henrys, because in bug #695396 Marcos says "This file in on picas.ghostscript.com on the FTP site." | 16:02.26 |
| I was planning to download it overnight | 16:02.45 |
chrisl | I don't have access to picas..... | 16:02.51 |
kens | BTW I fix the scan build and clang warnings after the fact if I forget them, when I get the mails from Marcos | 16:03.06 |
| chrisl, odd, I do. | 16:03.14 |
henrys | kens: oh didn't know that. Let me check if I have access | 16:03.18 |
kens | I just use my provate key | 16:03.19 |
chrisl | It seems it doesn't know my private key | 16:04.00 |
kens | Strange.... | 16:04.07 |
henrys | kens: no access for me | 16:04.08 |
kens | Ah wait, I'm on casper, that's not going to work | 16:04.18 |
henrys | marcosw1: doesn't replicate our accounts on all machines | 16:04.25 |
kens | I'm dure someone mentioed how to get to the ftp site there, maybe it was ray | 16:04.38 |
kens | digs back through support mail | 16:04.53 |
henrys | kens: for the customer we just "unsupported" he said it was in a separate email | 16:06.39 |
kens | Yeah, but I'm having trouble finding that email, I'm sure Ray said he had found it | 16:07.02 |
henrys | kens: oh I assumed he sent it just to them. | 16:07.27 |
| I don't seem to have it | 16:07.37 |
kens | Well, I could easily be mistaken | 16:07.37 |
| OK found the refrence from Ray in the irc logs, now to try and find the mail | 16:08.32 |
| Of course its buried in that stupidly long email thread form them | 16:08.55 |
| And indeeed there it is, easy when you have the date | 16:09.38 |
| congrats on getting the release out chrisl | 16:15.24 |
chrisl | kens: thanks, nice to get it done..... | 16:15.46 |
kens | OK I'm off, I'll test out the monster file from the customer's contractor tomorrow, assuming the download doesn't barf halfway | 16:16.03 |
| night all | 16:16.07 |
henrys | kens: don't contact the customer. | 16:16.49 |
Robin_Watts | Windows users... Control Panel -> Sound -> Sounds. Scroll down the "Program Events" and you'll find a "Microsoft Visual Studio" option. | 16:45.06 |
| You can set sounds for 'build success' and 'build failure'. | 16:45.17 |
| Very useful. | 16:45.22 |
kens | henrys I wasn't planning to contact the customer, but with masses of help from chris I've got a load of good memory fixes into pdfwrite, so I want to test their file and see if its OK now (Transparency_DesignGuide.pdf seems to be fine now) | 17:31.27 |
henrys | kens: good news! | 17:41.37 |
| hi fredross-perry_ | 17:42.48 |
fredross-perry_ | hey henry | 17:43.11 |
henrys | fredross-perry: about printing the xps stuff we have now certainly won't work on linux or mac. With the cups api I think you just throw the pdf at it. Of course we need page numbers and printer dialog stuff... | 17:44.33 |
fredross-perry | I presume I use Ghostscript or mupdf to write a PDF version of the file, and then print that? | 17:48.42 |
henrys | fredross-perry: yeah what do the QT docs say about printing? | 17:50.35 |
fredross-perry | not sure yet. (quick scan). The IF Iâm using now has me copying bitmaps. I get the printer resolution, render pages using mupdf at that resolution, and copy them. | 17:59.02 |
| this process could be sped up to some degree with multiple threads and banding, I suppose. | 17:59.39 |
kens | I think if you want to go from XPS to PDF you'll probably need GhostXPS and the pdfwrite device, or use MuPDF to render to an image and convert that to PDF> THat's slower but it'll be more reliable (the PDF outptu from XPS input is not exactly well used on GS) | 18:02.18 |
| You could do like the Windows PS driver and offer alternatives 'quicker' or 'more reliable' | 18:03.11 |
rayjj | kens: We _do_ have actual customers using xps->pdf -- that was the primary use of gxps initially | 18:03.25 |
| kens: but since we haven't gotten any bug reports for years, I'm not sure if any are still using it :-/ | 18:04.04 |
kens | I know we have some, but I'm kind of assuming they have a fairly fixed and predictable workflow. Over the years I've fixed some really bad problems in XPS->PDF but I'm fairly sure there are still some there | 18:04.11 |
| I'm reasonably sure I still have some bugs open against it | 18:04.35 |
| I can't find any now, maybe I fixed them all (or just closed tehm from lack of interest) | 18:05.41 |
fredross-perry | iâll do some research and see what Qt printing is actually capable of. | 18:06.19 |
Robin_Watts | fredross-perry: Urm... writing PDF and printing it relies on the print system having a PDF renderer in it. | 18:17.19 |
| bitmaps (with banding) seems like the best way to go if you can manage it. | 18:17.37 |
| That way you know you get 100% identical rendering on paper. | 18:17.47 |
fredross-perry | yes, but it may be that Qt will allow for detecting whether a given printer can accept PDF directly. But, that would then depend on our ability to render PDF from non-PDF, which from what Iâm hearing might be dodgy? Iâll look at banding first. | 18:19.49 |
Robin_Watts | We can't perfectly write PDF out from non PDF in. | 18:23.41 |
| So bitmaps is definitely the preferred option. | 18:23.58 |
| If later we can send PDF to PDF capable printers as an optimisation, then great. | 18:24.09 |
rayjj | fredross-perry: actually, converting PS->PDF using GS is quite stable and not "dodgy". We aren't that worried about getting XPS in | 18:28.30 |
| Robin_Watts: XPS is the only other input for mupdf, right ? | 18:28.54 |
Robin_Watts | cbz, jpg, tif, png... | 18:29.14 |
fredross-perry | itâs my understanding that I need GS to convert XPS to PDF, and then I can use mupdf to render. | 18:29.36 |
rayjj | which is better (gs or mudraw) at converting XPS->PDF is a question | 18:29.41 |
Robin_Watts | rayjj: gs. no question at all :) | 18:29.54 |
rayjj | Robin_WattsOK. | 18:30.04 |
| and as far as the other formats (cbz, jpg, tif, png..) we don't really care that much, do we ? | 18:30.40 |
| and converting raster images to PDF is rather trivial, right ? | 18:30.58 |
Robin_Watts | rayjj: Yes, we do that. (I think) | 18:31.16 |
| rayjj: The point is that MuPDF is designed to be extensible. | 18:31.28 |
| Anyone can add a new format handler. | 18:31.35 |
rayjj | it's not that likely that folks would be using 'gsview' to be their primary viewer for TIFF, JPEG, PNG, is it ? | 18:31.48 |
Robin_Watts | So someone could add a djvu decoder or something if they were feeling nuts. | 18:31.55 |
| People do use it for cbz. | 18:32.44 |
rayjj | Robin_Watts: still, that (being a collection of raster pages) should be fairly easy to make into a PDF, right ? Not like there's text or stuff, is there ? | 18:34.39 |
Robin_Watts | no, that would be fine, I think. | 18:34.58 |
| Nonetheless, my point about preferring bitmap stands I think. | 18:35.13 |
rayjj | Robin_Watts: but if the "printer" wants PDF, sending it raster from PS (or even XPS) seems rather rude | 18:36.49 |
Robin_Watts | rayjj: Do you get printers that only accept PDF and not bitmaps? | 18:37.34 |
| In the event that a printer supported both, I'd still be tempted to send a bitmap in case their rendition of a PDF was different to ours. | 18:38.27 |
rayjj | Robin_Watts: I'm thinking about printing to cups, but there are PDF printers, yes | 18:38.33 |
Robin_Watts | I am aware of their being PDF printers. But there are no PDF *only* printers, right? | 18:39.07 |
rayjj | Robin_Watts: as long as ours is "correct" :-) | 18:39.14 |
Robin_Watts | rayjj: Ours will be what is shown on the screen, so there will be no 'surprise' when the printer prints the same thing. | 18:39.36 |
rayjj | Robin_Watts: printers typically don't take PNG, TIFF, ... If they take bitmaps, it's "wrapped" in PS, PCL, or PDF | 18:40.02 |
| Robin_Watts: and if this is on a mobile device, doing "cloud print", sending bitmaps is rather painful compared to PDF's | 18:40.56 |
rayjj | isn't sure what platforms we are discussing here | 18:41.26 |
Robin_Watts | rayjj: Sending a PDF with a bitmap in, shouldn't be much more painful than sending a bitmap in PCL. As long as they compress the data. Surely they must compress the data. | 18:42.02 |
| rayjj: I've entirely lost track. | 18:42.18 |
chrisl | Surely start with the lowest common denominator (bitmaps) and add bells and whistles (like vector PDF/PS/PCL/XPS) later..... | 18:42.49 |
rayjj | Robin_Watts: a bitmap wrapped in PDF is probably preferred to PCL. PCL doesn't use much compression (at least for lossless) | 18:42.57 |
| chrisl: makes sense | 18:43.12 |
Robin_Watts | The sole points I was trying to make is that 1) MuPDF cannot reliably make PDF output from any input, and 2) if we have to pick either printing a rendered bitmap or printing PDF, then the former is a better initial bet cos it at least works everywhere. | 18:43.28 |
rayjj | I thought the Qt gsview would "print" through the underlying OS | 18:43.49 |
Robin_Watts | chrisl: aye. | 18:44.32 |
rayjj | Robin_Watts: If the input isn't PDF, then I agree. If the input is PDF and the "printer" accepts PDF, I'm not sure that dropping down to a bitmap is really "preferred" | 18:45.27 |
chrisl | rayjj: since we're going to need the bitmap print path (as a final fall back) it makes sense to implement that first | 18:46.33 |
rayjj | and if the input is PS, then gs's PS->PDF is being used anyway (for the viewer) so it's the same as starting with a PDF as far as the app cares | 18:46.43 |
| chrisl: for initial implementation, I agree. | 18:47.18 |
| time for lunch... | 18:47.54 |
Robin_Watts | If fredross-perry gets everything else working and then wants to add PDF -> PDF printer printing, then I won't object. | 18:48.01 |
fredross-perry | thatâs my plan | 18:48.22 |
henrys | fredross-perry: mac and linux support printing pdf files - cups does the rasterization: here's some code: http://www.glyphandcog.com/mac-linux-printing.html | 20:53.22 |
fredross-perry | henry, thanks. | 20:54.15 |
henrys | fredross-perry: and everything has already been converted to pdf when it is viewed right? | 20:54.26 |
| oh yeah you aren't using gs for xps and the comic book stuff. | 20:55.11 |
fredross-perry | Almost. Right now I am using mupdf to render to bitmap whatever it supports (pdf, xps, cbz). Anything thatâs NOT one of those would presumably be converted to PDF first, so yes. | 20:57.26 |
| you gave my answer. | 20:57.39 |
| but if we know itâs PDF, or we could get it there, thatâd be (perhaps) a better way. But weâre not absolved from needing to print bitmaps. | 20:58.17 |
henrys | fredross-perry: well you can use gs to go to pdf. Much better to have cups handle the file - it can figure out if the printer needs bitmap, postscript or takes pdf. | 20:59.12 |
fredross-perry | sure. | 20:59.27 |
henrys | fredross-perry: so the only thing we can't print well is cbz xps -> pdfwrite works fine. | 21:04.28 |
| I think it's even fine that we don't support cbz printing until somebody yammers about it. | 21:05.47 |
fredross-perry | Well, we can print it, but just slowly. Do the fz_* support banding, and where should I look? | 21:05.49 |
henrys | fredross-perry: I think you would just wrap a bitmap in pdf and use the same file mechanism. Robin_Watts : doesn't mupdf already support low level pdf output? | 21:07.54 |
| fredross-perry: but like I said I'm okay figuring out cbz printing later. I just don't know if there are enough users to worry about it. We can ask Tor tomorrow. | 21:09.54 |
fredross-perry | ok then. | 21:10.28 |
| henry - thanks for the link from glyphandcog. The CUPS method seems to work on OS X as well. Can we assume that Mac users have CUPS? | 21:34.59 |
henrys | fredross-perry: I would think, apple has owned cups and mike sweet for quite some time now. | 21:36.10 |
fredross-perry | ok | 21:36.19 |
| now Iâm looking for some examples that use options etc. | 21:36.45 |
Robin_Watts | mupdf supports *some* pdf output. | 22:41.52 |
| bitmaps are fine. | 22:41.55 |
| Forward 1 day (to 2014/09/23)>>> | |