| | <<<Back 1 day (to 2014/03/31) | 2014/04/01 |
| ray_laptop | looks like several of the regression servers are "down". IIRC those are in marcosw's house. | 04:47.20 |
| sebras | kens: morning! | 06:43.41 |
| kens | morning sebras | 06:43.48 |
| marcosw | I'm going to reboot fermis/plancks in a few minutes. | 07:15.13 |
| kens | ok | 07:15.20 |
| | You're up late marcosw | 07:15.28 |
| marcosw | I'm early for today's meeting ;-) | 07:15.46 |
| jogux_mac | paulgardiner_mac : I guess I'd be inclined to refund the guy you emailed about, but no idea who makes that decision :-) That multitasking issue comes up a lot in the reviews, and not sure about that cursor key issue. Did you open any of those in bugzilla? | 10:26.43 |
| | surely it can't be /that/ hard to make the cursor keys work... | 10:29.29 |
| ccxvii | I have an army of clones! | 10:29.35 |
| kens | HOW many tors ? | 10:29.41 |
| jogux_mac | what do you plan to do with your clone army? | 10:30.08 |
| kens | crush the rebels of course | 10:30.33 |
| | tehn build a giant battle station | 10:30.49 |
| ccxvii | mwahahaaa! | 10:30.53 |
| jogux_mac | that sounds like a lot of work tbh | 10:31.22 |
| kens | Nobody ever said Empire wa easy | 10:31.45 |
| ccxvii | that's what the clones are for! | 10:32.28 |
| | (that, and sebras being bored at work today, I imagine...) | 10:32.45 |
| tor0 | ccxvii: yes, indeed. | 10:32.55 |
| tor1 | ccxvii: I really should have changed the password for ccxvii too. | 10:33.17 |
| tor2 | ccxvii: that would have been a bit more fun. | 10:33.25 |
| tor3 | anyway, enough of this april 1st thing now. back to normal. :) | 10:34.00 |
| paulgardiner | jogux_mac: You reckon? I was imagining it would mean piping info up and down the UE2 hierarchy to move the cursor. If that stuff is already there and it's just that the key code isn't getting through then yeah easy, but I'd imagine not. | 10:42.47 |
| tor8 | paulgardiner: is it necessary to have iTunes to install an adhoc ipad app? | 10:43.08 |
| paulgardiner | jogux_mac: how long has inline editing been in. I'm sure that was an unsolved problem back in the day. | 10:43.39 |
| | tor8: don't think so. You can do it directly on the device. | 10:44.25 |
| | tor8: jogux_mac spotted some problems with that version so I'm working on an update | 10:44.57 |
| jogux_mac | tor8: you can do it from a website without (testflight or whatever) but that needs a plist file too. | 10:46.53 |
| | paulgardiner : inline editting is about 2 years old I think. | 10:47.13 |
| paulgardiner | Oh okay that figures | 10:47.34 |
| jogux_mac | It's certainly not trivial, but surely can't be worse than some of the layout issues we want to solve :-) | 10:47.56 |
| paulgardiner | Well it probably involves additions to UE2 although i guess fairly small ones. At least with the layout stuff it's all in the core | 10:49.53 |
| sebras | tor8: welcome back! | 10:50.08 |
| paulgardiner | I'll try another reply to him anyway | 10:50.31 |
| kens | hopes that's an April Fool | 11:09.10 |
| paulgardiner | tor8 Robin_Watts: should I expect the MuPDF android build to be broken? | 11:12.06 |
| tor8 | paulgardiner: yes, I haven't got around to testing it yet (it's on today's schedule) | 11:18.20 |
| paulgardiner | okay ta | 11:18.48 |
| tor8 | paulgardiner: how does the airprint stuff work on ios? dump a bitmap to some API? | 11:20.17 |
| paulgardiner | In several ways, one of which is pass a PDF file. Guess which method I choose. :-) | 11:21.28 |
| tor8 | paulgardiner: please choose on that works with XPS and the other input formats we also support... | 11:21.55 |
| | considering the #1 use of mupdf on ios is to view xps files... it'd be a shame if printing didn't work for that :) | 11:22.21 |
| paulgardiner | I think for now it'll have to be PDF only. This was supposed to be a quick job before getting back to SO. | 11:23.20 |
| tor8 | paulgardiner: maybe we should add a simple wrap-a-bitmap-in-pdf pdfwrite version? | 11:24.34 |
| | (the two top comments on itunes about mupdf complain about not being able to print xps files... hence my nagging) | 11:25.45 |
| paulgardiner | oh i see. I didn't realise that xps was explicitly mentioned | 11:26.41 |
| tor8 | I don't think anybody uses MuPDF on iOS for the PDF capabilities (given that the built-in PDF is good enough) | 11:27.08 |
| paulgardiner | I'll have a look at the other methods | 11:27.48 |
| jogux_mac | the annotation features in mupdf are quite nice, so I think people might use it for that - but there's a couple of issues that stop it working well (like afaics there's no way to get the document back out of mupdf to email it) | 11:28.02 |
| tor8 | jogux_mac: yeah. the ios way of handling files is really crippled and annoying. | 11:28.39 |
| jogux_mac | well that problem is more because we don't have a share button in mupdf :-) | 11:28.55 |
| | you also can't save annotation to documents that arrived via email, but I think Paul is fixing that (?) | 11:29.19 |
| paulgardiner | I dont know if I am. We'll need to check with henrys on priorities. | 11:32.14 |
| tor8 | paulgardiner: it *should* be a simple matter of just drawing a UIImage in the methods of a UIPrintPageRenderer | 11:32.58 |
| | but as with all things Apple, there's probably something magic going on that makes it devilishly tricky to get right | 11:33.33 |
| paulgardiner | yeah that might not be too difficult, assuming I can set up a way to test other than sending versions to jogux_mac | 11:34.37 |
| | tor8: yeah like the motions of subviews being affected by the opacity of tool bars? Don't you just love iOS. | 11:35.38 |
| | I would like to do the work to finish this off, otherwise all the iOS effort has been a bit of a waste | 11:41.05 |
| tor8 | paulgardiner: maybe http://www.netputing.com/handyprint/ could work? | 11:45.25 |
| jogux_mac | yeah, I suggested that yesterday :-) | 11:46.07 |
| tor8 | jogux_mac: oh, right. | 11:46.49 |
| paulgardiner_mac | Hmm. Would I need to plug my printer into the Mac? | 11:48.02 |
| tor8 | there's also supposed to be some sort of printer simulator in the iphone simulator | 11:48.13 |
| | paulgardiner: http://stackoverflow.com/questions/9028847/how-to-test-printing-from-iphone-app-if-we-do-not-have-airprint-enabled-printers | 11:48.23 |
| | apparently apple have removed the ability to print to shared printers from airprint, which sounds bizarre enough to be true... | 11:49.27 |
| paulgardiner_mac | Yeah, I tried simulator. It offered me loads of options that I didn't understand. | 11:49.53 |
| | That probably is the easiest option | 11:50.47 |
| | Just a PITA because I thought I was nearly done. | 11:51.02 |
| tor8 | paulgardiner_mac: sorry! | 11:51.23 |
| paulgardiner_mac | tor8: :-) not your fault. Stuff does need doing. | 11:52.49 |
| jogux_mac | paulgardiner : the print simulator "just works" for me. what opitions did it give you? :) | 11:54.58 |
| paulgardiner_lap | loads of printers | 11:56.45 |
| jogux_mac | seems you can even 'print' to it from devices. | 11:57.03 |
| paulgardiner_lap | Oh really?! | 11:57.20 |
| jogux_mac | paulgardiner : just pick one - mostly the difference is the print borders I think. eg. "simulated colour laser". | 11:57.21 |
| paulgardiner_lap | so do they all generate PDF? | 11:57.37 |
| jogux_mac | I don't know about internally, but with the simulator the end result is a page popping up in 'preview' showing the "printed" content, with yellow borders showing the non-printable areas. | 11:58.19 |
| paulgardiner_lap | Oh okay. | 11:58.47 |
| kens | Oh good grief. It looks like the problem with pdfmarks is because the pdfwrite code is using sscanf on a non-null terminated string (!!) | 12:00.14 |
| kens | smacks head against wall repeatedly | 12:00.34 |
| sebras | kens: '\0' here you go! :) | 12:13.30 |
| kens | sebras, if only it were that simple :-( | 12:13.43 |
| kubilayrd | hello, anyone? | 12:18.25 |
| kens | say 'hello' ghostbot, be nice | 12:18.46 |
| tor8 | hello | 12:19.05 |
| ghostbot | what's up, tor8 | 12:19.05 |
| kens | ROFL :-) | 12:19.16 |
| kubilayrd | I have troubles about installing fonts to ghostscript, hope that someone could give me a lift | 12:19.22 |
| kens | Not unless you say what your problems are. | 12:19.46 |
| | kubilayrd : you need to describe your problem | 12:22.19 |
| kubilayrd | I am using ImageMagick to convert a PDF file to JPG | 12:33.05 |
| | however, I get the following error: | 12:33.21 |
| | Error: /syntaxerror in pdfopen | 12:33.44 |
| | convert: Postscript delegate failed `ss.pdf': There is no such file or path @ pdf.c/ReadPDFImage/611 | 12:33.46 |
| | I've googled and found out that it might be about missing fonts that weren't embedded to pdf | 12:34.13 |
| | and that ghostscript comes without fonts by default | 12:34.26 |
| | so I want to install fonts | 12:34.31 |
| | am I on the right route? | 12:34.39 |
| kens | Well it *might* be fonts, but it equally well might not | 12:35.15 |
| | Its actually fairly unlikely, because new versions of GS will fallback even for missing CIDFonts, so you won't get an error | 12:35.39 |
| | What is you version of GS ? | 12:35.47 |
| kubilayrd | 8.70 | 12:35.54 |
| chrisl | Ghostscript very definitely comes with a core set of fonts by default - *if* a given distribution doesn't include those fonts, then the *really* should | 12:36.15 |
| kens | That's ancient, first thing to do is upgrade to the current version (9.14) or at hte very least 9.10 | 12:36.20 |
| | You should also try opening the PDF file directly with GS and see what it says (if anything) | 12:36.46 |
| kubilayrd | chrisl, how can I make sure that I have the core set of fonts, I mean, what's the path? | 12:37.03 |
| | kens, I'm upgrading now | 12:37.08 |
| | how can I try to open with GS? | 12:37.14 |
| kens | gs filename.pdf | 12:37.23 |
| chrisl | kubilayrd: the exact location will be distribution dependent, so it's hard to be sure | 12:37.39 |
| kens | SHould open the PDF file using the (default) display dfevice | 12:37.40 |
| kubilayrd | kens, out put of gs filename.pdf: | 12:38.24 |
| kens | Of course that assumes you have X installed | 12:38.26 |
| kubilayrd | Ghostscript 8.70: Cannot open X display `(null)'. **** Unable to open the initial device, quitting. | 12:38.33 |
| kens | OK so you don't hve X installed | 12:38.44 |
| kubilayrd | should I upgrade first? | 12:39.00 |
| kens | Try gs -sDEVICE=jpeg -sOutputFile=out%d.jpg filename.pdf | 12:39.08 |
| | THat should produce one JPEG file for each page in the PDF | 12:39.19 |
| Robin_Watts | I would consider upgrading first to be wise. | 12:39.49 |
| kubilayrd | kens, last command returns: | 12:40.11 |
| | Error: /syntaxerror in pdfopen | 12:40.12 |
| Robin_Watts | 8.70 was current when I joined the company (so 4.5 years ago) | 12:40.16 |
| kubilayrd | Operand stack: ... | 12:40.19 |
| | Execution stack:... | 12:40.25 |
| | Dictionary stack:... | 12:40.28 |
| | GPL Ghostscript 8.70: Unrecoverable error, exit code 1 | 12:40.30 |
| kens | kubilayrd : well that's probably the source of your ImageMagick problem | 12:40.33 |
| | That version of Ghostscript can't openyour PDF file | 12:40.43 |
| tor8 | Robin_Watts: so, any pointers to how to update the android JNI bufferstream for the new stream interface? :) | 12:40.56 |
| kens | I'd try the current version, if it still doesn't work we'll need to see the PDF file, so your best bet would be to open a bug report. | 12:41.15 |
| Robin_Watts | tor8: Urm... change 'read' or 'fill' to be next. | 12:41.20 |
| | next should set rp and wp as appropriate. | 12:41.41 |
| kubilayrd | kens, does centos have it or I need to build from src? | 12:41.46 |
| kens | kubilayrd : not a clue, but I would guess not. | 12:41.58 |
| Robin_Watts | There is no buffer in the fz_stream any more, so if you need one, put it in the private stream state. | 12:42.07 |
| kens | IIRC Centos is, umm, slow about putting out updates | 12:42.12 |
| | So you would needto build from source | 12:42.21 |
| kubilayrd | kens: okay, it has 8.70-xx; I'm installing from the src now | 12:42.29 |
| Robin_Watts | Then you return with if (rp == wp) return EOF else return *rp++; | 12:42.35 |
| kens | bows out and leaves that to chrisl | 12:42.38 |
| chrisl | Eh?? | 12:42.52 |
| tor8 | Robin_Watts: I'm guessing I have to add a buffer somewhere though... | 12:43.00 |
| kens | building from source on Linunx | 12:43.01 |
| kubilayrd | by the way, the pdf files that I have this problem about are created in acrobat. I have no problem with pdfs created with indesign | 12:43.04 |
| kens | chrisl^^ | 12:43.06 |
| tor8 | this thing uses (*env)->GetByteArrayRegion(env, array, stream->pos, len, buf); | 12:43.17 |
| Robin_Watts | tor8: If you can get access to the underlying data buffer, just make rp and wp point into that. | 12:43.22 |
| chrisl | kubilayrd: I assume you're build the 9.14 src? | 12:43.24 |
| kubilayrd | chrisl: like you said, yes | 12:43.38 |
| Robin_Watts | tor8: Does it get the byteArrayRegion, then copy the data out, then release the region before returning? | 12:43.49 |
| tor8 | Robin_Watts: I have no idea how JNI works, much less this stuff. | 12:43.50 |
| | it looks like it copies data out of the java side into a C buffer | 12:44.04 |
| Robin_Watts | If so, just put a char buffer[4096] into the private stream strate structure. | 12:44.19 |
| | That will replace the one in the fz_stream. | 12:44.28 |
| | tor8: If you want, I'll have a look after lunch for you. | 12:44.39 |
| tor8 | the stream state already holds the 'globals' struct | 12:44.47 |
| Robin_Watts | tor8: right, so it should have globals + a buffer. | 12:45.14 |
| henrys | tor8:you brought up cmake a few days ago and I hadn't looked at it in years, I didn't realize how large the following had become. Some important projects are using it. | 12:45.26 |
| Robin_Watts | I'll fix it after lunch. | 12:45.33 |
| tor8 | Robin_Watts: okay. I'll leave it alone for now then. | 12:47.47 |
| | Robin_Watts: I'll leave a WIP patch on tor/master with some other minor android fixes that are needed though | 12:48.37 |
| Robin_Watts | tor8: Great, thanks. | 12:49.05 |
| tor8 | I've got two ndk-build warnings that might be new | 12:49.19 |
| Robin_Watts | Clocks have changed, so we have 1:40 before the meeting, right? | 12:49.21 |
| tor8 | draw-scale-simple.c: Warning: conditional infixes are deprecated in unified syntax | 12:49.39 |
| pedro_ | nods - that's what I thought | 12:49.46 |
| tor8 | something with the inline asm | 12:49.47 |
| | Android NDK: WARNING:jni/Android.mk:mupdfcore: LOCAL_LDLIBS is always ignored for static libraries | 12:49.53 |
| Robin_Watts | tor8: That's known about. | 12:49.54 |
| | The assembler stuff is down to gcc being stupid. | 12:50.09 |
| tor8 | which sounds scarier, as if something big has changed with the build system... | 12:50.18 |
| Robin_Watts | If you use old format assembly you get a warning. | 12:50.36 |
| tor8 | Robin_Watts: right. | 12:50.43 |
| Robin_Watts | If you use new format assembly it doesn't work on old stuff. | 12:50.47 |
| | so we're living with the warning. | 12:50.54 |
| | The LOCAL_LDLIBS thing is new to me. | 12:51.05 |
| | I'll grab some lunch then have a look. | 12:51.11 |
| kubilayrd | chrisl: is this file the right one? I couldn't configure and make. http://downloads.ghostscript.com/public/binaries/ghostscript-9.14-linux-x86_64.tgz | 12:52.40 |
| chrisl | kubilayrd: that's a binary, not source | 12:53.00 |
| kubilayrd | http://downloads.ghostscript.com/public/ghostscript-9.14.tar.gz | 12:53.15 |
| | is that it? | 12:53.17 |
| chrisl | Yes, that's the source. TBH, you can probably use the binary to test whether the PDF works now..... | 12:53.44 |
| kubilayrd | gs-914-linux_x86_64 filename.pdf | 12:55.10 |
| | like this? | 12:55.11 |
| chrisl | Better to use the jpeg command line options kens posted above | 12:55.36 |
| kubilayrd | well it says no such command like this | 12:55.48 |
| chrisl | ./gs-914-linux_x86_64 -sDEVICE=jpeg -sOutputFile=out%d.jpg filename.pdf | 12:56.33 |
| kubilayrd | Processing pages 1 through 51. Page 1 >>showpage, press <return> to continue<< | 12:57.00 |
| | means it worked? | 12:57.08 |
| chrisl | Yes | 12:57.11 |
| jogux_mac | wonders if there would be any hope of getting epage to build sanely with cmake. | 12:57.14 |
| kubilayrd | great | 12:57.18 |
| | chrisl, do I need to remove the old version before installing 9.14? | 12:57.32 |
| kens | Of course, presuading ImageMagick to use that binary is a different problem | 12:57.44 |
| kubilayrd | okay, on it | 12:58.07 |
| chrisl | kubilayrd: I would suggest not, since it will also remove any packages that depend on gs, like ImageMagick | 12:58.10 |
| kubilayrd | yeah it just warned me about it, removing for dependencies section | 12:58.54 |
| chrisl | Are you just going to use the binary you just tried? | 12:59.27 |
| kubilayrd | chrisl, I doubt it, shouldn't I install? | 12:59.51 |
| | I was about to build the source | 12:59.59 |
| chrisl | If you install from source, it will almost certainly not overwrite the existing installation (which is a good thing!), but it will mean you'll need to add the location of the new gs executable to your PATH | 13:01.00 |
| kubilayrd | export $PATH=... ? | 13:02.11 |
| chrisl | Yeh, you'll probably want something like: export PATH=/usr/local/bin:$PATH | 13:02.44 |
| kubilayrd | okay, if that would be enough. | 13:03.20 |
| chrisl | Well, that only sets PATH for the current terminal instance. | 13:03.52 |
| kubilayrd | What should I do for persistency, then? | 13:04.35 |
| chrisl | To make the PATH change permanent, you should add PATH=/usr/local/bin:$PATH to the end of the file ~/.profile | 13:05.06 |
| kubilayrd | okay, that explains a lot to me about some commands I was trying to exec from PHP :-) | 13:05.43 |
| | thanks, I'm making the source atm | 13:05.51 |
| chrisl | *Hopefully*, ImageMagick will pick up the gs executable from your PATH, so will get the new one - but if it doesn't you'll need to ask the ImageMagick people about it | 13:06.30 |
| kubilayrd | yeah, that's my concern | 13:07.05 |
| | so there is no way to remove ghostscript without touching to imagemagick? | 13:08.02 |
| chrisl | kubilayrd: is there a reason you're using ImageMagick? | 13:08.13 |
| kubilayrd | I am converting PDF files to JPG images with high density | 13:08.31 |
| | that's the only reason | 13:08.40 |
| chrisl | So, you could really just use GS directly, and forget about ImageMagick | 13:09.14 |
| kubilayrd | Am I be able to set the resolution or density to higher values? | 13:09.53 |
| kens | resoltuoin, yes, what do you mean by 'density' ? | 13:10.07 |
| kubilayrd | It's the option I use with ImageMagick, etc the input file's resolution is 467âÃâ680; when I convert with --density 200 option, it gets up to 1299âÃâ1890 without quality loss | 13:11.09 |
| | Which is deadly important to me | 13:11.27 |
| kens | DOn't use JPEG then :-) | 13:11.49 |
| kubilayrd | I have to use an image format eventually, it doesn't matter whether it's JPEG or PNG | 13:12.21 |
| kens | Given that GS is doing the conversion, it seems likely that you can get whatever you wnt, unless IM is doing some post-processing | 13:12.21 |
| | JPEG is lossy, PNG need not be | 13:12.31 |
| tor8 | "density" is imagemagick's (non-standard) terminology for resolution | 13:12.42 |
| | so --density 200 is equivalent to -r 200 in gs | 13:12.57 |
| kens | Use PNG (or TIFF, whartever) not JPEG where quality is important | 13:12.57 |
| | so there you are, use -r to set the resolution | 13:13.07 |
| tor8 | and what kens said, please use PNG not JPEG | 13:13.09 |
| kubilayrd | how about file size? these images will be downloaded to mobile devices | 13:13.26 |
| chrisl | God I really dislike ImageMagick...... | 13:13.32 |
| tor8 | kubilayrd: pick one: quality or file size :) | 13:13.48 |
| kens | kubilayrd : then there's a tradeoff, size/quality | 13:13.51 |
| | If size is most important, use JPEG, if quality is most important, use PNG | 13:14.23 |
| tor8 | a lower resolution png is better than a high resolution jpeg; especially if you have text... jpeg really messes up text | 13:14.28 |
| kens | yeah that's really true | 13:14.41 |
| kubilayrd | I do have text... Okay it seems I need to try both and see the results. | 13:15.02 |
| tor8 | but if your PDF files are mostly photos and very light on text content, you could get away with JPEG | 13:15.03 |
| kubilayrd | So ImageMagick uses Ghostscript to do what Ghostscript does? | 13:15.16 |
| | I didn't get that part | 13:15.21 |
| kens | IM doesn't handle PDF files, it uses GS to convetrt them to an image format | 13:15.33 |
| kubilayrd | tor8: they contain both, like magazines | 13:15.34 |
| tor8 | kubilayrd: one thing to try is to render to 600dpi with gs and then downsample the results for improved anti-aliasing | 13:15.37 |
| kubilayrd | So IM does 2 jobs while GS doing one? | 13:15.53 |
| kens | No GS is doing all the work there | 13:16.05 |
| | IM only handles bitmap formats | 13:16.13 |
| tor8 | IM calls GS to do the job for converting PDF to an image | 13:16.19 |
| kens | For PostScript and PDF, IM uses GS to render to an image format | 13:16.36 |
| kubilayrd | A PDF file (90 MB) conversion took more than an hour with IM, does it change with GS? | 13:16.36 |
| kens | Probably its faster, only one way to find out | 13:16.50 |
| tor8 | in this case IM adds nothing of value compared with calling GS directly | 13:16.51 |
| | and IM is probably going to be slower (I expect IM calls GS to convert to some uncompressed image format, and then recompress that as JPEG) | 13:17.23 |
| kens | You *could* use IM to post-process an image file, you may get better image processing that way, but for rendering, GS does the work | 13:17.34 |
| kubilayrd | I have nothing to do with image processing, I just need to split a PDF file to image files page by page | 13:17.57 |
| kens | Then use GS | 13:18.03 |
| | Especially because it can do it in one pass, IM will do multiple passes, on per page | 13:18.20 |
| | one per page* | 13:18.31 |
| | I've answered questions on that one before on Stack Overflow | 13:18.44 |
| kubilayrd | Yeah, seems like I've chosen the wrong path once | 13:19.03 |
| kens | Many, many people do, because they are familiar with ImageMagick, but don;t realise that 'under the hood' the work is being handed off | 13:19.29 |
| kubilayrd | I did make too many searches but all it felt like IM was the only way | 13:19.35 |
| kens | Nothgin wrong with that, of course.... | 13:19.40 |
| kubilayrd | Yeah, seems like so, probably because of image processing | 13:20.01 |
| | Which is I don't need at all... | 13:20.09 |
| kens | Oh you could also use MuPDF for the same task | 13:20.11 |
| tor8 | kubilayrd: you could also try mupdf with the command line 'mudraw' tool, to convert from PDF to PNG. | 13:20.11 |
| kens | Its probably even faster | 13:20.17 |
| | LOL echo.... | 13:20.24 |
| tor8 | kens: indeed :) | 13:20.43 |
| kubilayrd | Is it a tool of... what? OS? | 13:20.45 |
| tor8 | kubilayrd: install the mupdf package, and run 'mudraw' command | 13:20.58 |
| kens | Its another Artifex open source product | 13:20.59 |
| chrisl | If the version of Centos only has gs 8.70 packaged, I wouldn't get my hopes up that mupdf is packaged for it...... | 13:21.35 |
| kubilayrd | It's Amazon | 13:21.56 |
| | but I doubt it too | 13:22.03 |
| | Okay, options are two, I'll try them both then | 13:22.58 |
| norbertj | henrys: I added a new patch for the jpeg decoder. I tested it and the few differences I see with clusterpush seem to me progressions (i.e. the stems in some characters seem more equal, also for barcodes). | 13:29.33 |
| henrys | norbertj: I'm looking at your bmpcmp results on the regression dashboard | 13:39.00 |
| kens | 50 mnuters to meeting ? | 13:39.51 |
| | Or even minutes | 13:40.04 |
| henrys | kens: yes | 13:40.11 |
| kens | thanks (our clocks changed) | 13:40.21 |
| henrys | norbertj: there are large shifts in images are these bmpcmp results current? | 13:40.45 |
| norbertj | yes. | 13:41.19 |
| | Is it possible to see the complete bitmap? This bmpcmp looks like a part of it. | 13:41.44 |
| henrys | norbertj: no typically we grab the file and render it ourselves | 13:42.38 |
| paulgardiner_lap | tor8: for printing non-pdf docs on iOS there looks to be only one way to proceed to get acceptable results and that's looking an infeasible amount of work. | 13:42.54 |
| Robin_Watts | norbertj: bmpcmp cuts the bitmaps down to only show the relevant areas. Otherwise the bitmaps are too large to see on a screen, and too large to store and too large to transfer between nodes. | 13:43.28 |
| paulgardiner_lap | tor8: the way I'm thinking is to create pdf in all cases, using the pdf device when the source isn't pdf | 13:43.37 |
| Robin_Watts | paulgardiner: Can airprint not take a bitmap as input? | 13:44.15 |
| | PWG specifically? | 13:44.29 |
| | norbertj: The bitmaps all look plausible to me, if they were dropping data before... | 13:45.11 |
| paulgardiner_lap | Robin_Watts: yes, but this is supposed to be a mobile solution, so I would have thought the size of image necessary to get reasonable printing result would be infeasibly large to transfer (and even temorarily store for files of many pages) | 13:45.39 |
| norbertj | henrys: I took the last job (mango_jpg.pxl) and printed, resulting bitmap looks ok to me (nicely centered, while previously it wasn't). | 13:45.51 |
| henrys | norbertj: on the dashboard choose test files, then you want tests_private/customer_tests/mango_jpg.pxl | 13:46.07 |
| | oh you already figured that out ;-) | 13:46.15 |
| Robin_Watts | henrys: You can click on the link in the bmpcmp too. Or in the web view of the report. | 13:46.34 |
| henrys | norbertj: great if it's all good let's commit | 13:46.36 |
| | norbertj: you have a pretty extensive test file suite as well I guess it passed. | 13:47.18 |
| kubilayrd | chrisl: No such file ~/.profile, should I change /etc/profile? | 13:48.12 |
| norbertj | I'm going to run today, (for that I had to first have a clean idea of the fix, which I now have). But it will take several hours to complete, so results tomorrow. | 13:48.28 |
| henrys | paulgardiner: did you ever find an AirPrint printer? | 13:48.50 |
| chrisl_r500 | kubilayrd: really? That's odd... give me a second | 13:49.06 |
| | kubilayrd: do you have a ~/.bashrc file (assuming you are using bash)? | 13:50.07 |
| paulgardiner_lap | henrys: jogux_mac tested on his and it seems to work, but I'm currently supporting only the printing of pdf docs, and tor8 pointed out that most of the moaning on the app store is about not being able to print XPS | 13:50.22 |
| kubilayrd | chrisl_r500: yes, contains: # .bashrc # Source global definitions if [ -f /etc/bashrc ]; then . /etc/bashrc fi # User specific aliases and functions | 13:50.52 |
| jogux_mac | henrys : the print simulator built into xcode seems to be pretty good too. (better than cups's airprint, anyway :-) ) | 13:51.07 |
| paulgardiner_lap | henrys: Also it seems the simulator has facility for testing printing so I should be able to test here too now | 13:51.09 |
| jogux_mac | [in the version of cups I have] | 13:51.13 |
| chrisl_r500 | kubilayrd: you should be able to add the PATH definition to ~/.bashrc then | 13:51.33 |
| jogux_mac | henrys : btw, been playing with ms office on the ipad. I think their UI may be a hell of a lot slicker than ours :-( | 13:51.58 |
| kubilayrd | chrisl_r500: this line to end? export PATH=/usr/local/bin:$PATH | 13:52.07 |
| chrisl_r500 | Yes | 13:52.20 |
| kubilayrd | still, gs command returns 8.70 | 13:52.36 |
| henrys | jogux_mac: in the mobile world I have to think we are very dated. | 13:52.52 |
| chrisl_r500 | kubilayrd: did the new gs actually install in /usr/local/bin? | 13:53.11 |
| kubilayrd | yep, I even tested the binary | 13:53.34 |
| jogux_mac | didn't kubilayrd say he was doing php (presumably inside a web server)? .bashrc may not affect that. | 13:53.37 |
| | henrys : yeah :-( | 13:53.51 |
| chrisl_r500 | kubilayrd: you do have to start a new terminal for that to take effect | 13:54.12 |
| kubilayrd | k, on it now | 13:54.19 |
| | chrisl_r500: Now it happened | 13:54.51 |
| | So, I've been looking to MuPDF documentation but couldn't find how to convert | 13:55.30 |
| chrisl_r500 | kubilayrd: that's good. Now, are you going to be calling gs (directly or indirectly) from your php scripts? | 13:55.44 |
| kubilayrd | chrisl_r500: Yes, exec("gs filename.pdf"); | 13:56.07 |
| Robin_Watts | kubilayrd: What do you want to do with mupdf? | 13:56.23 |
| kubilayrd | Robin_Watts: Convert PDF to JPEG or PNG | 13:56.49 |
| Robin_Watts | mudraw -o out%d.png in.pdf | 13:56.53 |
| kubilayrd | Robin_Watts: Can I give additional parameters for resolution? | 13:57.28 |
| chrisl_r500 | Right, so usually, php scripts run under dedicated user (often call "php" or similar). You either need to change your system wide configuration file (under /etc) or change the settings file for the php user | 13:57.41 |
| kubilayrd | chrisl_r500: You mean /etc/profile? | 13:58.23 |
| chrisl_r500 | Yes. Be *very* careful, though, an error in there could prevent you logging into the system | 13:59.01 |
| kubilayrd | chrisl_r500: How about giving the absolute path in PHP command? | 13:59.28 |
| | chrisl_r500: I was doing so with IM... | 13:59.37 |
| chrisl_r500 | kubilayrd: in that i | 13:59.51 |
| | in that case it will probably work - I really know almost nothing about php | 14:00.09 |
| kubilayrd | Is this MuPDF source wrong for Unix? I can't configure https://code.google.com/p/mupdf/downloads/detail?name=mupdf-1.3-source.tar.gz&can=2&q= | 14:00.45 |
| Robin_Watts | We don't need no stinkin' configure. | 14:01.15 |
| kubilayrd | OK, processing with make... | 14:01.30 |
| | Robin_Watts: Can I give additional parameters for resolution in MuPDF? | 14:01.57 |
| Robin_Watts | mudraw -o out%d.png -r res in.pdf | 14:02.06 |
| | mudraw -o out%d.png -w width in.pdf | 14:02.18 |
| | mudraw -o out%d.png -h height in.pdf | 14:02.26 |
| kubilayrd | Robin_Watts: One last thing: Does it convert each page to seperated files? | 14:02.40 |
| Robin_Watts | Yes, that's what the %d in the -o filename does. | 14:02.55 |
| kubilayrd | Robin_Watts: Thank you very much for the help, I'll return if anything goes wrong. | 14:03.15 |
| | Also, thank you guys, you've been very helpful. kens, chrisl_r500, tor8 | 14:03.54 |
| kens | You're welcome | 14:04.00 |
| chrisl_r500 | kubilayrd: no problem | 14:04.09 |
| henrys | Robin_Watts: what was it blissful deadlock? - I was ROFL never heard that before. | 14:04.45 |
| ray_laptop | kubilayrd: chrisl: kens: I was going to mention that -o out-%d.jpg is often less confusing since it doesn't prompt for showpage... | 14:04.53 |
| kens | ray_laptop : yes but we wanted to establish the problem first | 14:05.16 |
| tor8 | paulgardiner: yes, using pdfwrite would be the ideal long-term solution but that device doesn't support text yet IIRC | 14:05.23 |
| ray_laptop | kubilayrd: it's equivalent to -sOutputFile=out-%d.jpg -dBATCH -dNOPAUSE | 14:05.24 |
| | kens: right. | 14:05.36 |
| tor8 | paulgardiner: I would imagine anyone printing off a phone would be doing so on a WiFi network (being near the printer in question), not a 3G | 14:05.52 |
| Robin_Watts | holy deadlock :) | 14:05.56 |
| paulgardiner_lap | tor8: yes exactly and why I said it was an infeasible amount of work | 14:05.57 |
| henrys | Robin_Watts: ah that's right | 14:06.14 |
| tor8 | paulgardiner: yeah. still, bitmaps for XPS would be better than nothing. | 14:06.21 |
| ray_laptop | tor8: except for the speed | 14:06.50 |
| kubilayrd | Okay, before using these commands, I need to go through installation... X11/Xlib.h: No such file found. | 14:06.55 |
| | What's missing? | 14:06.58 |
| paulgardiner_lap | Yeah I suppose so. So what dpi should I aim at? Should I try to match the printer? | 14:06.58 |
| kens | kubilayrd : X windows ? | 14:07.20 |
| ray_laptop | paulgardiner: matching the printer is good | 14:07.23 |
| paulgardiner_lap | henrys: Am I okay working on iOS a bit longer. Thought I was nearly done, but it seems not. | 14:07.32 |
| kubilayrd | kens: platform/x11/x11_main.c:3:22: fatal error: X11/Xlib.h | 14:07.36 |
| tor8 | kubilayrd: you're missing xorg-dev (if you're on debian or ubuntu) | 14:07.49 |
| | kubilayrd: have you downloaded the 1.3 source or checked out the latest git? | 14:08.12 |
| ray_laptop | paulgardiner: but, of course, that can make the bitmap larger than might be needed. It's just for text that people want (near) full res | 14:08.27 |
| Robin_Watts | kubilayrd: HAVE_X11=no make build=release | 14:08.39 |
| tor8 | kubilayrd: make build=release NOX11=1 if you have 1.3, make build=release HAVE_X11=no if you're working from git | 14:08.41 |
| paulgardiner_lap | ray_laptop: just worried that, given as all the images have to be created up front, I might run the device out of space | 14:08.56 |
| ray_laptop | paulgardiner: Oh, all of the pages have to exist prior to sending to the printer ? | 14:09.25 |
| Robin_Watts | paulgardiner: Can you not do one job per page for the bitmap casee? | 14:09.26 |
| kubilayrd | tor8: downloaded the 1.3 src | 14:09.44 |
| tor8 | then NOX11=1 on the command line to make should do the trick | 14:10.00 |
| henrys | paulgardiner_lap: continue with iOS I'll fend off miles if he gets wind of it ;-) | 14:10.03 |
| paulgardiner_lap | henrys: :-) | 14:10.14 |
| kubilayrd | Robin_Watts: Now trying again with: HAVE_X11=no make build=release | 14:10.16 |
| | tor8: HAVE_X11=no make build=release wouldn't work? | 14:10.43 |
| tor8 | kubilayrd: no. we changed the magic variable name recently, so the 1.3 release still uses the awkward NOX11=1 | 14:11.08 |
| paulgardiner_lap | Robin_Watts: That's an idea, but I'm not sure it would help. | 14:11.19 |
| kubilayrd | tor8: Re-making now. | 14:11.51 |
| paulgardiner_lap | I guess I should give it a try. At least I should be able to write the images to disc and not need to keep them all in memory. | 14:12.11 |
| ray_laptop_ | paulgardiner: if it's a matter of know how many pages there will be in the print job, you know that without needing the bitmaps | 14:12.19 |
| tor8 | paulgardiner_lap: one would hope that the frameworks writes the pages to disk as PDF before firing off the job | 14:12.43 |
| ray_laptop_ | paulgardiner: what disk does an iPhone have ? | 14:12.49 |
| chrisl_r500 | paulgardiner_lap: can't you stream the bitmaps as they are created? | 14:12.52 |
| tor8 | paulgardiner: given that I think a lot of apps use plenty of bitmaps in printing | 14:13.02 |
| paulgardiner_lap | ray_laptop I think they vary. | 14:13.31 |
| tor8 | henrys: ray_laptop: no word from mujs.com yet, I take it? | 14:13.57 |
| paulgardiner_lap | I'll give it a go anyway | 14:14.06 |
| henrys | I haven't heard anything no tor8 | 14:14.18 |
| ray_laptop | tor8: no, no update from Miles | 14:14.51 |
| | I'll send Miles a reminder email to bug the guys (or at least give us the info from the purchase so I can bug them) | 14:15.40 |
| | email sent | 14:17.05 |
| marcosw1 | Robin_Watts: is the ~testdocs directory on peeves complete or is there somewhere else I should be transferring the files from? | 14:19.51 |
| | ^peeves^peeved | 14:19.58 |
| Robin_Watts | marcosw1: As far as I know, that's the most complete copy we have. | 14:20.10 |
| marcosw1 | Robin_Watts: thx. | 14:20.16 |
| Robin_Watts | We didn't get a better one from the admins did we? | 14:20.25 |
| jogux_mac | not that I saw anywhere | 14:20.38 |
| | I don't think there was anywhere big enough for 500G+ of data to be hiding without us noticing :) | 14:20.57 |
| Robin_Watts | yeah :( | 14:21.04 |
| | If it was going to be there it would have had to be somewhere other than /mnt/picsel | 14:21.32 |
| marcosw1 | jogux_mac: I've ordered two 1 TB drives which should arrive tomorrow; I'll starting transfering the files from peeved to one of my machines. | 14:22.03 |
| jogux_mac | is pretty sure it's not on that NAS, anyway. | 14:22.10 |
| | marcosw1: great. | 14:22.20 |
| marcosw1 | ray_laptop: what's the uplink speed from peeved? I want to use a reasonable bwlimit so as not to swamp your connection. | 14:22.55 |
| Robin_Watts | and if we do find a copy of a newer testdocs, we should still be OK as the picsel policy was never to change a file once it had gone into testdocs. | 14:23.22 |
| ray_laptop | marcosw: AFAIK it is 768Kb, but other than regression testing, I don't care | 14:23.32 |
| jogux_mac | marcosw1: assuming this'll be a slow-ish transfer, could you do testdocs/{pptx,docx,xlsx} first please, then 'ole', then everything else. [that's based on our work priorities] | 14:23.44 |
| ray_laptop | marcosw: if it is going to take too long, I can throw the stuff on a drive and overnight it up to you. | 14:24.53 |
| marcosw1 | jogux_mac: sure | 14:24.59 |
| ray_laptop | ISTR that it is about 600Mb | 14:25.06 |
| marcosw1 | ray_laptop: let me get it started and see how it goes, the stuff that's a priority might not be too bad. | 14:25.38 |
| jogux_mac | the first set is about 15G, ole 220G. | 14:26.29 |
| ray_laptop | I don't know if this is old news: http://www.electronicproducts.com/Software/System/Microsoft_finally_launches_Office_Suite_for_iPad_and_its_free_ish.aspx | 14:27.21 |
| Robin_Watts | freeish = 70 quid a year :) | 14:27.39 |
| henrys | ray_laptop: yes we've been discussing it. | 14:27.40 |
| marcosw1 | so the first set should be here in hours, the second set in a couple of days. probably not worth shipping a drive. | 14:28.17 |
| pedro_ | cool - that sounds ideal | 14:28.40 |
| jogux_mac | robin_watts : they did make the iphone one free for 'home' use (whatever that means) at the same time | 14:28.42 |
| henrys | ray_laptop: so our current planned response will be a free read-only smart office. | 14:29.10 |
| jogux_mac | robin_watts : and ipad one is 'free' for me :-) | 14:29.15 |
| henrys | paulgardiner_lap: feel free to expense a new AirPrint printer if you want a printer. I assume you weren't worried about that and just didn't want to get another printer but just in case... | 14:30.01 |
| | meeting time. | 14:30.17 |
| | I do think we should all take a look at cmake and at least find some reasons to shoot it down. chrisl thoughts? | 14:31.08 |
| chrisl_r500 | I think we shouldn't | 14:31.22 |
| Robin_Watts | henrys: For what products? | 14:31.28 |
| kens | henrys, its somethgin else WIndows users would have to download and build | 14:31.48 |
| henrys | Robin_Watts: everything | 14:31.51 |
| kens | We eliminated other products for the same reason | 14:32.05 |
| chrisl_r500 | henrys: I very much doubt that cmake can deal with the multiple build steps required for all our products - or at least, for gs and co | 14:32.29 |
| paulgardiner_lap | Thanks henrys. Might be handy. | 14:32.46 |
| Robin_Watts | henrys: I'd be more inclined to tell windows users that they would need to download and use gmake. | 14:32.46 |
| | s/gmake/gnu make/ | 14:32.59 |
| | but as it is, we have a solution that works well for windows/linux. Why go changing? | 14:33.20 |
| ray_laptop | seconds Robin_Watts | 14:33.34 |
| | (and chrisl) | 14:33.40 |
| Robin_Watts | and by everything I really hope you're not including SOT in that. | 14:33.46 |
| jogux_mac | robin_watts : I would kind of like to see the back of the FBS if that was at all feasible... | 14:34.13 |
| henrys | Robin_Watts: that's the first thing I'd consider actually | 14:34.16 |
| marcosw1 | ray_laptop: did you say 768 kb/s or kB/s? I'm only seeing ~120 kB/s. | 14:34.26 |
| ray_laptop | henrys: we have enough problems to fix without starting a new build system, don't we ? | 14:34.47 |
| paulgardiner_lap | We should use the FBS for gs and mupdf :-) | 14:34.53 |
| chrisl_r500 | jogux_mac: I suspect any non-dedicated system would struggle to cope with the vast array of configuration options needed for SOT | 14:35.09 |
| marcosw1 | it's called the Fantastic Build System for a reason... | 14:35.11 |
| pedro_ | notes that there were around 3 re-jigs of the SOT build system... | 14:35.14 |
| ray_laptop | marcosw: K bits per second (is that KB/s ?) | 14:35.16 |
| Robin_Watts | jogux_mac: I'd love to see the back of the FBS, but moving to cmake seems like changing one pair of comedy clown shoes for another. | 14:35.17 |
| jogux_mac | robin_watts : you may know more about cmake than I do (or less about FBS :-) ) | 14:35.35 |
| marcosw1 | ray_laptop: bits is b, afaik | 14:35.41 |
| pedro_ | depends largely how many separate targets we want to hit; seems feasible to concentrate on a small number of platforms/configs | 14:35.48 |
| henrys | Robin_Watts: is there some problem with cmake? | 14:35.53 |
| ray_laptop | marcosw: since it is counter intuitive, and since I've been using Windoze, my case sensitivity has declined | 14:36.28 |
| kens | So at the moment, GS can be buitl using standard tools on any platform, and we're going to ask *everyone* to downlaod and buld cmake before they ubild GS ? | 14:36.32 |
| chrisl_r500 | pedro_: I thought we were committed to at least a configuration per OEM customer? | 14:36.38 |
| ray_laptop | henrys: the biggest problem is that Windows users have to download it. | 14:36.56 |
| Robin_Watts | henrys: Is there some problem with the existing VS solution? | 14:37.02 |
| kens | ray_laptop : so would Linux users (unless its standard) | 14:37.07 |
| marcosw1 | ray_laptop: it probably will be necessary to send a drive, unless jogux doesn't made getting the docx/pptx/xlsx in days and everything else in weeks. | 14:37.21 |
| henrys | kens:it doesn't work, it does not have *real* project files and it still uses nmake. | 14:37.35 |
| marcosw1 | Robin_Watts: where were the testdocs files transferred from? | 14:37.42 |
| chrisl_r500 | It's not standard on Linux, but is mostly available as a package - but it would bugger Solaris, AIX, HP-UX users, etc..... | 14:37.44 |
| Robin_Watts | The only problem I can see with the current VS solution (for gs) is the lack of parallel builds. | 14:37.46 |
| pedro_ | chrisl_r500: most of them are simple feature config at present, on android, w8, w8mobile and iOS afaics - we used to have a whole host of arm,mips.x86 targets, various products, various features etc to support | 14:37.55 |
| kens | henrys it doeas work, and nmake is a standard part of VS, therefore its a standard tool | 14:37.58 |
| paulgardiner_lap | I think the problem with sot isn't so much the FBS as the number and complexity of the builds and that a lot is done just with free-form python programs | 14:38.09 |
| pedro_ | now it seems workable to reduce that and have a more comprehensible build system | 14:38.10 |
| mvrhel_laptop | sorry I am a bit late.. | 14:38.16 |
| chrisl_r500 | pedro_: and the "aliens" for individual customers? | 14:38.26 |
| jogux_mac | paulgardiner_lap : I'd perhaps also see it as a chance to clean up some of the mess :-) | 14:38.33 |
| ray_laptop | henrys: but it's good enough for who it's for. How does cmake cure having "real project files" for Visual Stufio ??? | 14:38.36 |
| pedro_ | for me its a question of pain/gain | 14:38.37 |
| henrys | mvrhel_laptop: do you think we are lacking for not having real project files in gs? | 14:38.46 |
| kens | cmake wouldn't improve the project files at all | 14:38.49 |
| Robin_Watts | I think we are having 2 separate conversations here. | 14:39.01 |
| kens | We *could* have real project files for VS and GS, but its a maintenace problem | 14:39.11 |
| henrys | kens:I thought cmake produces visual studio projects? | 14:39.19 |
| ray_laptop | Robin_Watts: gs builds are fast enough without parallel builds (IMHO) | 14:39.24 |
| jogux_mac | thought the same as henrys. | 14:39.29 |
| kens | henrys I wasn't aware of that | 14:39.36 |
| pedro_ | chrisl_r500 - not specifcally; there were initially, but we pushed that onto the customers; what we have in the build system is more dev and product aliens; we'd supply the library to customers and they'd write app/alien | 14:39.37 |
| Robin_Watts | 1) Changing gs over to cmake, and 2) Changing SOT away from the FBS. | 14:39.54 |
| | For 1) I question the need to mess with what we have. | 14:40.06 |
| ray_laptop | a full gs elease build is 146 seconds | 14:40.08 |
| kens | I question the gains for 1) also | 14:40.24 |
| henrys | if you haven't looked at in a long time you shouldn't be so quick to dismiss it. There are large serious projects now using cmake. | 14:40.31 |
| ray_laptop | debug slightly less, and incremental builds are on the order of 10 seconds | 14:40.38 |
| kens | We do have project files and they work well enough | 14:40.39 |
| Robin_Watts | For 2) I'd be all in favour of moving away from FBS if it can be done in reasonable time/effort/reliability. | 14:40.39 |
| jogux_mac | ray_laptop : I've never heard an engineer complain that the build was too fast :-) | 14:40.41 |
| ray_laptop | jogux: no, but having to support our customers and tell them "now you need cmake" doesn't appeal to me. | 14:41.21 |
| kens | nods | 14:41.27 |
| Robin_Watts | Fundamentally, we want to be able to change builds (add files etc) in just 1 place. And we have that for gs already. | 14:41.35 |
| ray_laptop | if nothing else, cust 532 would be a pain -- they integrate our build into the build of their entire printer software | 14:41.53 |
| henrys | ray_laptop: the no parallel build is torture if you spend a little time on a mac or linux and then try an visual studio build - you just haven't used it. My dial up modem is just fine ;-) | 14:42.03 |
| ray_laptop | and they use make | 14:42.04 |
| Robin_Watts | The problem with cmake is that people will inevitably edit the VS files themselves rather than the CMake sources for the VS files. | 14:42.04 |
| mvrhel_laptop | I am fine with the way gs is set up in visual studio now | 14:42.12 |
| Robin_Watts | And then there will be problems. | 14:42.20 |
| mvrhel_laptop | have the facade of files really causes me no issues | 14:42.30 |
| Robin_Watts | henrys: What we really want is a parellelised nmake then ? | 14:42.42 |
| henrys | well anyway I won't push it further... | 14:42.43 |
| chrisl_r500 | cmake just seems like a whole new set of problems to deal with.... | 14:42.49 |
| Robin_Watts | There is an nmake clone out there that claims to do parallel stuff, but it fails on gs. | 14:43.04 |
| | I'd rather see our engineering time spent on fixing that personally... | 14:43.15 |
| ray_laptop | Robin_Watts: that sounds like the real issue | 14:43.20 |
| henrys | anyway something to chew on - I won't push it more. | 14:43.25 |
| | tor8: mupdf release? | 14:43.55 |
| ray_laptop | Robin_Watts: not surprising that it fails on gs. But having a VS project that uses 'pre' and 'post' steps might let the build of the C files be built in parallel | 14:44.39 |
| Robin_Watts | ray_laptop: A pure VS project would be a retrograde step. | 14:45.05 |
| | Cos then people would need to maintain 2 build systems. | 14:45.18 |
| chrisl_r500 | ray_laptop: It takes a lot more than that - there's a lot of interdependencies as well as the pre/post steps | 14:45.29 |
| henrys | Robin_Watts: are you up on when the release will be? | 14:45.34 |
| Robin_Watts | henrys: AIUI, things to do for the release are: | 14:45.53 |
| | 1) Let paul finish the airprint stuff. | 14:46.03 |
| | 2) fix the android build. | 14:46.07 |
| | I'm doing 2) now. | 14:46.10 |
| ray_laptop | Robin_Watts: probably, in that it might end up compiling a bunch of devices that aren't needed | 14:46.27 |
| Robin_Watts | It would be nice to announce mujs at the same time, so it would be nice to have mujs.com sorted... | 14:47.04 |
| henrys | Robin_Watts: miles is bugging me for the release because he wants to do the SOT announcement in the newsletter. | 14:47.14 |
| | and we usually do the newsletter after the releases | 14:47.29 |
| Robin_Watts | I understand. | 14:47.34 |
| paulgardiner_lap | Do we need to hold off the main release for airprint, or could an updated app be submitted a little later? | 14:47.44 |
| Robin_Watts | traditionally appstore releases have not been synced with version releases. | 14:48.09 |
| | so we could release sooner. | 14:48.34 |
| paulgardiner_lap | That's what I thought. | 14:48.44 |
| henrys | paulgardiner_lap: estimate on airprint? | 14:48.55 |
| Robin_Watts | But I'm out of the loop on this slightly (have been wearing my SOT hat), so tor8 may know of other things. | 14:48.57 |
| henrys | Robin_Watts: perhaps the time change got tor8 | 14:49.37 |
| paulgardiner_lap | henrys: I'd guess 2-3 days, but I'm off for a week. Also another thing jogux pointed out is we have no share button to allow things like emailing altered docs | 14:50.17 |
| henrys | paulgardiner_lap: well keep going with it but it may not make it in. Of course if tor8 is not ready for some other reason it doesn't matter. | 14:51.46 |
| Robin_Watts | pedro: "arsing the ooxml" Nice :) | 14:52.01 |
| chrisl_r500 | Official project name?? | 14:52.15 |
| henrys | jogux_mac, pedro_: do you guys have anything for the meeting? | 14:52.28 |
| pedro_ | that's actually a better description of how it feels ;) | 14:52.31 |
| henrys | mvrhel_laptop: I didn't forget about xpswrite compression - just getting a bunch of stuff from guillaume - ugh | 14:53.17 |
| marcosw1 | ray_laptop: you will have to send me a drive, just the docx/pptx/xlsx portion is going to take ~36 hours (that's based on real transfer times). Since that's only 3% of the total we are talking weeks for everything. | 14:53.37 |
| jogux_mac | henrys : can't recall much. still plodding away at the various things; ATS is advancing but slowly. | 14:53.43 |
| | henrys : been talking about marcosw about getting more disc space so we can run the bulk da tests | 14:54.02 |
| marcosw1 | my estimate of ~4 hours when I thought you said bytes was pretty close though :-) | 14:54.02 |
| ray_laptop | marcosw: OK. I'll go get a drive today and xfer it | 14:54.04 |
| mvrhel_laptop | henrys: no problem. I am working on text selection UI right now in gsview | 14:54.15 |
| chrisl_r500 | marcosw1: do you have something other than jbig2dec that can read jb2 diles? | 14:54.22 |
| pedro_ | henrys: on the ooxml bugs (mainly word just now) we have a number of issues where we lose content or render incorrectly compared to MSWord; OpenOffice does the same as SOT and I can see no evidence in the xml internals which explains how word renders some of these features (watermarks, invisible orders etc | 14:54.34 |
| ray_laptop | marcoswI: I'll let you know, but I should be able to get it to you by tomorrow. | 14:54.42 |
| mvrhel_laptop | once I have that done, then I just need to add in a couple more save as options and clean up a couple things here and there | 14:54.43 |
| kubilayrd | I can convert my PDF to JPEG with both GS and MuPDF. Thank you guys all! Will try combinations now. See ya. :-) | 14:54.54 |
| henrys | ray_laptop: you're customer email was a bit honest for me, sounded like something I'd say ;-) | 14:54.58 |
| marcosw1 | chrisl_r500: I don't think so. I looked for something when the jbig2 testing came up but that was a while ago. | 14:55.07 |
| ray_laptop | marcosw: Is you address in Joann's employees doc correct ? | 14:55.13 |
| chrisl_r500 | marcosw1: oh well, you can just deal with the inverted jbig2 output for now..... | 14:55.45 |
| mvrhel_laptop | ray_laptop: nice response to the customer about doing his own transparency handling in his customer display list | 14:55.56 |
| jogux_mac | ray_laptop : it's not a huge rush as it'll take us a couple of days to configure the tests etc too (just don't want you going out of your way panicing thinking it's a huge rush!) | 14:56.03 |
| pedro_ | we can leave them on the basis we believe they're not standard ooxml or try to figure what msword is doing to infer behaviour. There are a few simle bugs, but the most critical issue I can see is the one I've been spending most time on; SOT is missing a whole bunch of layout and styling from the OOXML parsing | 14:56.05 |
| mvrhel_laptop | custom display list that is | 14:56.12 |
| marcosw1 | ray_laptop: yes. I'll be out of town for a few days starting on Friday, so if you can't get the drive to me by tomorrow Thursday would be good. | 14:56.14 |
| henrys | pedro_, jogux_mac any sense of how compatible microsoft's product is? | 14:56.16 |
| mvrhel_laptop | oh I will be out of town next week | 14:56.24 |
| pedro_ | this leaves a lot of misplaced, wrongly-sized or missing content. | 14:56.27 |
| mvrhel_laptop | already let joann know but I will send a email out | 14:56.35 |
| Robin_Watts | paulgardiner and I are away from Thursday->Tuesday. We'll miss next weekends meeting, probably. | 14:56.55 |
| | s/weekends/tuesdays/ | 14:57.04 |
| henrys | Robin_Watts: right | 14:57.13 |
| ray_laptop | mvrhel_laptop: yeah, but somehow I don't think I'll convince those guys :-( | 14:57.16 |
| mvrhel_laptop | Robin_Watts: where are you guys off to? | 14:57.17 |
| Robin_Watts | mvrhel_laptop: Southern Spain. | 14:57.33 |
| mvrhel_laptop | ray_laptop: I just hope we don't get sucked into a lot of work | 14:57.36 |
| jogux_mac | henrys : I mostly played with the ipad ui. pedro_ may have played more with the guts. | 14:57.38 |
| mvrhel_laptop | Robin_Watts: nice | 14:57.41 |
| | Robin_Watts: you should stop in and see Marti | 14:57.57 |
| marcosw1 | chrisl_r500: I've clearly missed an email (or irc discussion), but what inverted jbig2 issue are you going on about? | 14:57.57 |
| pedro_ | henrys: I think MSWord read most ooxml just fine, but it also writes mso-xxx properties which can override ooxml attributes | 14:58.00 |
| Robin_Watts | mvrhel_laptop: Hadn't thought of that. We could claim the trip on expenses! :) | 14:58.22 |
| tor8 | henrys: the android build is the main thing holding up a release | 14:58.24 |
| chrisl_r500 | marcosw1: if you look at the commit message for my jbig2 device fix, it's explained in there | 14:58.26 |
| mvrhel_laptop | exactly | 14:58.27 |
| tor8 | and having mujs.com be live would be nice for the release announcement | 14:58.33 |
| jogux_mac | pedro_ : I think henrys might've been asking about office for ipad (and I'm not sure that's what you're talking about)? | 14:58.45 |
| pedro_ | henrys: haven't played too much with it yet, but I didn't spot any obvious issues in the docs I tried | 14:58.47 |
| tor8 | and we're also waiting on airprint, since that would be sweet to have, but we can release the app built from later source as we've done in the past | 14:59.06 |
| pedro_ | jogux: was still talking about our SOT-OpenOffice-MSOffice compatibility | 14:59.17 |
| jogux_mac | nods at pedro_. too many parallel topics :) | 14:59.32 |
| marcosw1 | chrisl_r500: thx. haven't had my morning latte yet; was up late last night chatting with kens on irc :-) | 14:59.57 |
| kens | ROFL | 15:00.03 |
| | henrys marcosw what should we do with bug numbers 695121 to 695123 ? These expose the problems with rasterops and pdfwrite. I'm inclined to close as 'WONTFIX'. | 15:00.14 |
| ray_laptop | just like in person staff meetings -- people talking amongst themselves while henry tries to follow the agenda ;-) | 15:00.41 |
| henrys | kens:I left a comment and I assumed wont fix would be the outcome. | 15:00.58 |
| chrisl_r500 | marcosw1: well, it was your bug, so.... anyway, it's probably not hard to fix if we pin down the problem to our output | 15:01.16 |
| pedro_ | was just too slow typing my initial comments and was overtaken by a similar sounding question ;) | 15:01.18 |
| tor8 | Robin_Watts: have you got one of those mupdf changes since last release lists typed up somewhere? | 15:01.20 |
| kens | henrys, that's fine I'll do that then, I did read the comment but wasn't sure if you wanted an enhancement raised instead | 15:01.21 |
| marcosw1 | kens: that's fine with me. I'll let the customer know. | 15:01.25 |
| Robin_Watts | tor8: I have not. | 15:01.32 |
| | I can make one for you if you want. | 15:01.37 |
| tor8 | Robin_Watts: okay, I sort of thought you kept a rolling log somewhere from past releases, but no worries | 15:01.49 |
| marcosw1 | chrisl_r500: they are all my bugs :-( | 15:01.52 |
| henrys | we are past the 1/2 hour anything else meeting-wise? | 15:02.00 |
| marcosw1 | or at least sometimes it feels that way | 15:02.02 |
| Robin_Watts | tor8: Nah, I "git log" and type stuff into an emacs window :) | 15:02.18 |
| tor8 | Robin_Watts: aha! :) | 15:02.32 |
| marcosw1 | I did? | 15:02.37 |
| tor8 | well, I can do that as well then, if you're busy fixing the android build | 15:02.47 |
| pedro_ | marcosw: probably worth assigning all the GhostDocs bugs to me just now and I can reprioritise/update status for them. Sound sane? | 15:02.53 |
| Robin_Watts | tor8: I think I have it fixed. | 15:02.58 |
| marcosw1 | pedro_: sure. I could also just add you to the ghostdocs-bugs email alias. | 15:03.39 |
| tor8 | Robin_Watts: Fab! Have you tested it as well, to make sure the javascript stuff works? I'm not 100% confident I got the JS setup all right with the android changes. | 15:03.40 |
| kens | fetches more coffee | 15:04.16 |
| Robin_Watts | tor8: No, I just got the native part building. I'll push a fix commit for this onto robin/master in a mo, then test it a bit. | 15:04.21 |
| | If you could test it too, that would be good. | 15:04.36 |
| tor8 | Robin_Watts: I will. I'd like to test paul's iOS app too, before we release it onto the app store, but are we waiting for that to hold up the release or should we just get the android stuff working then release? | 15:05.32 |
| | the docs need some updating as well | 15:05.48 |
| | the examples have to get the fz_register_document_handlers call in there, and the build instructions for V8 could be excised | 15:06.21 |
| paulgardiner | Don't hold off for the iOS app. Hopefully it wont take long but just in case | 15:06.36 |
| | I need to nip out. biab | 15:07.15 |
| tor8 | paulgardiner: alright. I'll see if we can get the RC out before you two take off (thursday through tuesday next week was it?) | 15:07.29 |
| Robin_Watts | tor8: this thursday to next tuesday, yes. | 15:08.21 |
| kens | sounds a bit tight for a release | 15:08.48 |
| Robin_Watts | kens: That's just for the RC. | 15:09.43 |
| kens | ah ok | 15:09.48 |
| Robin_Watts | tor8: I've pushed an updated commit. | 15:10.05 |
| henrys | jogux_mac: so a timed save for this? http://bugs.ghostscript.com/show_bug.cgi?id=695128 | 15:10.07 |
| jogux_mac | henrys : or save when we enter background | 15:10.59 |
| Robin_Watts | Can we refuse to enter the background? | 15:11.20 |
| jogux_mac | but I'm not sure if it should be saving to a temporary file or updating the 'real' file (I guess the former) | 15:11.23 |
| | robin_watts : no | 15:11.25 |
| Robin_Watts | temp file then :( | 15:11.34 |
| jogux_mac | so we have to fiddle to reload the temporary file when we get restarted | 15:11.55 |
| | henrys : where does that prioritise in amongst everything else? | 15:12.07 |
| henrys | jogux_mac: I'm rather new to this but I'd say more important than the other bugs. I'd find that maddening. | 15:12.53 |
| jogux_mac | I still need to actually confirm that's why it's breaking; but it's a really good theory :-) | 15:13.10 |
| | I think paul raised a really annoying sounding one about external keyboards too | 15:13.34 |
| Robin_Watts | external keyboards cursor keys don't work. | 15:13.50 |
| jogux_mac | that's the one | 15:13.54 |
| tor8 | Robin_Watts: I suspect that will leak 'bs' on an exception in fz_open_document_with_stream | 15:14.21 |
| jogux_mac | I have vague memory of complaints in appstore reviews about that too. | 15:14.23 |
| henrys | jogux_mac: yes another P1 bug as we call them. | 15:14.26 |
| Robin_Watts | tor8: fz_new_stream should guarantee to close the stream it's passed on exception, I think. | 15:14.48 |
| jogux_mac | henrys: ah, right. | 15:15.04 |
| tor8 | Robin_Watts: not the stream, the stream state struct, in case fz_new_stream throws | 15:15.27 |
| henrys | paulgardiner_lap: can you make a p1 bug for that and I'll change the memory problem to p1 | 15:15.42 |
| tor8 | oh, wait, nvm. | 15:15.44 |
| Robin_Watts | tor8: s/the stream/the stream state/ | 15:15.45 |
| henrys | paulgardiner_lap: ah you did make a bug | 15:16.24 |
| tor8 | Robin_Watts: right. so the close(stream) in the always cleans up after any errors in fz_open_document_with_streams | 15:17.13 |
| Robin_Watts | yes. | 15:17.21 |
| tor8 | but fz_new_stream could throw (on a malloc error) and then 'bs' would leak | 15:17.36 |
| Robin_Watts | fz_new_stream only mallocs within an fz_try. | 15:18.06 |
| tor8 | and rethrows | 15:18.17 |
| Robin_Watts | and the fz_catch specifically calls close on the state. | 15:18.19 |
| tor8 | eww :) | 15:18.42 |
| henrys | jogux_mac: I made those P1 along with the set fault. | 15:18.44 |
| Robin_Watts | basically the ownership of bs passes in. | 15:18.45 |
| tor8 | that's not obvious :) | 15:18.49 |
| jogux_mac | henrys : how much of the P1s do we need to get in before release? Is it still urgent to do a release without EG? | 15:18.52 |
| marcosw1 | henrys: okay to add pedro_ and jogux_mac to the private attachments group on bugzilla? | 15:18.54 |
| henrys | s/set/seg | 15:18.54 |
| tor8 | but yes, in that case, LGTM | 15:18.55 |
| jogux_mac | henrys : thanks | 15:19.07 |
| henrys | jogux_mac: no we're square with eg | 15:19.11 |
| Robin_Watts | tor8: Do you have an android device with mail on ? :) | 15:19.19 |
| henrys | jogux_mac: short term - for this release at least | 15:19.45 |
| tor8 | Robin_Watts: yes, why? | 15:19.46 |
| Robin_Watts | tor8: Cos it's opening a PDF attachment from gmail that exercises this code I think. | 15:20.03 |
| | and I don't think I have that set up. | 15:20.15 |
| ray_laptop | darn. I forgot that I have to have the "extras=..." arguments to bmpcmp | 15:20.28 |
| henrys | pedro_: the watermark stuff is alarming - we have no idea where it comes from? | 15:20.39 |
| jogux_mac | henrys : ah, okay. so all P1s before release? | 15:20.44 |
| tor8 | Robin_Watts: right. I'll see if I can get that to work then. | 15:21.10 |
| marcosw1 | brb | 15:21.21 |
| henrys | jogux_mac: hopefully ;-) | 15:21.27 |
| ray_laptop | henrys must have been taking online management classes. Of course everything is your top priority ! ;-) | 15:21.31 |
| henrys | jogux_mac: if the keyboard thing is going to take a long time we won't wait. | 15:22.06 |
| ray_laptop | does the touch screen still work for cursor movement when you have the external KB ? | 15:22.48 |
| | (not that that is a very convenient work around) | 15:23.17 |
| tor8 | Robin_Watts: oh FFS, what is with people and this obsession with only providing 32-bit linux binaries?! | 15:24.17 |
| ray_laptop | tor8: well, they still run on a 64-bit sytem, right ? | 15:24.53 |
| tor8 | the "32 and 64-bit" tarball *only* has 32-bit binaries | 15:25.08 |
| | ray_laptop: nope. only if you install multi-arch and make your system a hybrid 32/64-bit | 15:25.21 |
| ray_laptop | tor8: that must be the default for ubuntu then -- I don't recall having to do anything special | 15:26.15 |
| Robin_Watts | tor8: dies with javascript. | 15:26.53 |
| tor8 | default ubuntu was still 32-bit last I checked (which may have been a while ago) | 15:26.54 |
| | Robin_Watts: stuck on not being able to compile due to the insanity of the android SDK being 32-bit only | 15:27.16 |
| ray_laptop | tor8: not for me when I set up peeves and peeved (and peeved was > 4 years ago). If you have a 64-bit capable processor and > 4Gb RAM it seems to automagically use 64-bit SMP | 15:28.25 |
| | s/peeved was/peeves was/ | 15:28.52 |
| tor8 | ray_laptop: interesting. last time I installed ubuntu I got a 32-bit PAE kernel | 15:28.55 |
| ray_laptop | tor8: was that in a VM ? | 15:29.12 |
| tor8 | ray_laptop: no, on a laptop so it may have been the <4G of RAM | 15:29.41 |
| marcosw1 | can someone refresh my (clearly failing) memory of why I needed the jbig2 device to be working? Something to do with testing of jbig2 output presumably but that's not as helpful as it might be :-) | 15:30.15 |
| ray_laptop | tor8: that's probably why | 15:30.52 |
| chrisl_r500 | marcosw1: it was for testing our luratech encoder integration - we had some case where it didn't work..... | 15:31.29 |
| pedro_ | henrys: I've made some degenrate files with simply a 1 page doc with watermark and 1 word of text; can't see any section styling to indicate how MSWord thinks there's a watermark so far | 15:31.43 |
| | henrys: microsoft do use special internal shapeIDs (just magic numbers to ooxml) so it may be one of them | 15:32.39 |
| ray_laptop | chrisl: you beat me to it | 15:32.40 |
| henrys | pedro_: for what it's worth I found the latest libre office to be a bit better than open office for some of these bugs. | 15:33.40 |
| marcosw1 | chrisl_r500: okay, I think that's enough for me to go on. | 15:33.43 |
| pedro_ | henrys: mm, I have SOT,OpenOffice, LibreOffice and 3 versions of Office for reference ;) | 15:34.24 |
| henrys | pedro_: well okay then | 15:34.40 |
| chrisl_r500 | marcosw1: so it's something to the weekly luratech tests - probably just a small subset of test files will be more than enough | 15:34.57 |
| | s/to the/to add to the | 15:35.12 |
| pedro_ | henrys: LibreOffice definitely does seem to be better maintained these days | 15:36.15 |
| marcosw1 | chrisl_r500: is it necessary to do it on a weekly basis? I was planning on doing a one time test and then presume we wouldn't break anything... | 15:36.31 |
| chrisl_r500 | marcosw1: I thought the point of this was that we did break something.... maybe not weekly - monthly? Pre-release? | 15:37.20 |
| marcosw1 | chrisl_r500: there isn't any mechanism to run the weekly tests with a subset of test files against a particular device, so without annoying changes we'd end up running all the test files with the jbig device. I suppose there isn't anything wrong with that... | 15:40.24 |
| | better too many tests than not enought. | 15:40.44 |
| chrisl_r500 | I suspect doing it as part of the release testing would be sufficient | 15:41.31 |
| henrys | pedro_, jogux_mac : Reading through pedro_ 's analysis in bugzilla my fear is that layout compatibility is not going to be possible with current resources. Maybe I'm not reading this right, but most problems are not very well understood or point to seeming daunting undertakings. | 15:41.36 |
| Robin_Watts | henrys: Yes, exact layout compatability is never going to happen. | 15:42.17 |
| | I've got a deal simple file here, just 1 paragraph of text on an A4 page, all one font size. | 15:42.52 |
| | We split the first line a word earlier than Libre Office does. | 15:43.13 |
| pedro_ | henrys: I think there's a long way we could go by going through the ooxml DAs and comparing against the spec to make sure everything's implemented as expected, and at the same time handling the missing mso_xxx properties; some of the reuired changes are akin to floats in html | 15:43.17 |
| | but it is going to be difficult to guarantee the same rendering | 15:44.01 |
| | (even MSWord on different machines/OS variants can produce different results) | 15:44.33 |
| Robin_Watts | I plan to follow the code through to try to understand why, but ultimately it's going to come down to "our measurement of the width of the text overflows our measurement of the width of the box". Without being able to compare it to what LibreOffice is doing, it's going to be very hard to spot where the actual mistake is. | 15:44.38 |
| | pedro_: I think the best we can realistically hope for is that we don't lose content, and that on the whole, we match (allowing for slightly different line breaks etc). | 15:45.19 |
| pedro_ | nods | 15:45.30 |
| Robin_Watts | It's never going to be "print from MS, print from us, compare, no diffs" | 15:45.49 |
| | And that realisation is going to require a shift in the way we (as traditionally being a print oriented company) think about things. | 15:46.30 |
| ray_laptop | we should just install the free MS Office view-only version and use that to generate the bitmaps we show ;-) | 15:47.02 |
| marcosw1 | ray_laptop: I'm giving up on the rsync of testdocs from peeved, will wait for you to ship me a drive. | 15:48.50 |
| ray_laptop | darn. the bmpcmp compare of 4-bit gray vs. 8-bit gray pgms works OK on my local machine, but the regression run gives me entirely /dev/fd/63: Unrecognised image type | 15:49.47 |
| | marcosw: sounds like a fine idea | 15:50.04 |
| jogux_mac | gotta go get poorer / pick up car from garage. back in a bit... | 15:50.12 |
| ray_laptop | I guess I've done all I can as far as testing 4-bit fast thresholding. No segfaults at least. | 15:51.03 |
| marcosw1 | ray_laptop: sounds like the bmpcmp the cluster uses is out of date, but I'm not sure how that can be since it's afaik it's compiled from source for every cluster run. | 15:51.11 |
| Robin_Watts | ray_laptop: That's... odd. | 15:51.26 |
| | tor8: Is there a magic incantation needed to get js support any more? | 15:52.29 |
| marcosw1 | the /dev/fd/63 instead of a file name is caused by the way bmpcp is called: bmpcmp <(gunzip file1.gz) <(gunzip file2.gz) ... | 15:52.44 |
| Robin_Watts | marcosw1: But I don't understand the Unrecognised image type error. | 15:53.24 |
| kens | I've seen that if sourece or candidate failed on a page | 15:53.52 |
| ray_laptop | marcosw: Robin_Watts: trying the one in ~marcos/cluster ... | 15:54.07 |
| kens | Eg if I fixed a crash | 15:54.08 |
| marcosw1 | Robin_Watts: yeah, that's why I was thinking it was an old version and couldn't deal with 4bpp bmp files (are there such things, or are they 8bpp with the bottom 4 bits set to 0?). | 15:54.40 |
| Robin_Watts | kens: ah, that would make sense, cos there would be no data to read. | 15:54.41 |
| pedro_ | has to pop out for 20 mins | 15:56.39 |
| ray_laptop | marcosw: Robin_Watts: I tried the version is ~marcos/cluster/users/ray/ghostpdl (date stamp corresponds to my last run) and it works as on Windows | 16:00.41 |
| Robin_Watts | ray_laptop: I wonder if I've inadvertently got some fseeks in there. | 16:01.10 |
| | That would fail on a piped input, and might give the error? | 16:01.27 |
| ray_laptop | Robin_Watts: I didn't try with the < syntax | 16:01.46 |
| Robin_Watts | ray_laptop: Right, so that would fit with my theory? | 16:02.12 |
| marcosw1 | I have to run to uni. back online later. | 16:03.23 |
| ray_laptop | Robin_Watts: what's the actual syntax I can use ? Or do I have to fire up a perl shell ? | 16:06.40 |
| Robin_Watts | ray_laptop: It's a bash thing. | 16:06.52 |
| | ./bmpbmp < (cat input1) < (cat input2) out 0 100 | 16:08.23 |
| | that assumes that input1 and input2 aren't compressed. | 16:08.38 |
| | marcosw uses (gunzip -c input1.gz) etc | 16:09.03 |
| ray_laptop | Robin_Watts: ./bmpcmp < (cat /tmp/x.pgm) < (cat y.pgm) xy (with bash) gives me: bash: syntax error near unexpected token `(' | 16:09.22 |
| Robin_Watts | Sorry, it may be important that there is no space between < and ( | 16:09.41 |
| ray_laptop | Robin_Watts: OK: This works: ./bmpcmp <(cat /tmp/x.pgm) <(cat /tmp/y.pgm) xy | 16:13.42 |
| Robin_Watts | That works as in, runs and gives no errors ? | 16:14.06 |
| ray_laptop | gives me the .meta and 3 .bmp files | 16:14.11 |
| | and they look the same as on windows | 16:14.20 |
| Robin_Watts | OK, so it's something cluster specific. | 16:14.24 |
| ray_laptop | Oh, well. I'm not going to worry about it. I'll just get gobs of differences I already expect. | 16:15.07 |
| Robin_Watts | ray_laptop: I'll try and look into it at some point if I see it again. | 16:15.28 |
| | tor8: Android mupdf is still dying on trying to open a passworded file and cancelling out of the dialogue box. | 16:16.51 |
| henrys | mvrhel_laptop: I mean to ask about the viewer. It might be nice to say something about it in the newsletter or should I hold off until next release? | 16:17.15 |
| | s/mean/meant | 16:18.10 |
| mvrhel_laptop | henrys: gosh. It may be better to wait until the next newsletter | 16:18.29 |
| | As you know sometimes that last 10% takes a bit of time to finish | 16:18.48 |
| henrys | mvrhel_laptop: that's fine | 16:20.02 |
| kens | heads into the sunset | 16:20.25 |
| | goodnigth all | 16:20.29 |
| chrisl | also off now - 'nite folks! | 16:23.13 |
| ray_laptop | Robin_Watts: what I really need is to compare 4-bit w/o fast thresholding to to 4-bit WITH fast thresholding, so don't bother about the bmpcmp issue. | 16:26.08 |
| | There isn't a way to use a different set of parameters for the baseline | 16:26.40 |
| Robin_Watts | ray_laptop: No. | 16:28.10 |
| | ray_laptop: You could go 'old school' and use htmldiff.pl ? | 16:28.41 |
| ray_laptop | Robin_Watts: I'm just going to run the 72 dpi differences locally -- there are *only* 2692 files to look at :-) | 16:32.15 |
| | or at least until I get tired | 16:32.29 |
| Robin_Watts | ray_laptop: That's what htmldiff.pl will do for you. | 16:32.31 |
| | It's basically a local bmpcmp. | 16:32.49 |
| | It's what I used before we had bmpcmp. | 16:32.55 |
| | gs/toolbin/htmldiff.pl | 16:33.14 |
| ray_laptop | Robin_Watts: right, I've used it before. | 16:33.16 |
| Robin_Watts | tor8: ping | 16:34.20 |
| | tor8: Fix for android crashing on robin/master | 16:34.51 |
| | Still no idea why javascript stuff is dying though :( | 16:35.06 |
| | heh. My phone is set to download my mail, but it's pre spam filter. | 16:36.04 |
| | All the attachments appear in the Download folder. | 16:36.22 |
| | This means that all the jpeg test files over time become viagra ads :) | 16:36.41 |
| mvrhel_laptop | oops forgot that I needed to take my car in today. bbiab | 16:37.35 |
| Robin_Watts | tor8: I think it's dying in javscriptSupported | 16:55.40 |
| tor8 | Robin_Watts: hmm. does it call javascriptSupported before opening a document? | 17:01.33 |
| Robin_Watts | tor8: Not sure. | 17:01.46 |
| tor8 | because that probably won't work... | 17:01.55 |
| Robin_Watts | Trying some debug now. ndk-gdb is... unhelpful. | 17:01.56 |
| tor8 | it doesn't look like it from grepping, javascriptSupported is only called in MuPDFPageView.passClickEvent | 17:03.01 |
| Robin_Watts | MuPDFCore_javascriptSupported is called with globals being NULL. | 17:03.12 |
| tor8 | get_globals on a class object might not work? | 17:04.11 |
| Robin_Watts | get_globals won't work within a static object. | 17:04.30 |
| tor8 | might need to make that a function on MuPDFCore objects rather than the class, and use mCore.javascriptSupported? | 17:04.56 |
| Robin_Watts | yes. | 17:05.08 |
| | That fixes it. | 17:08.08 |
| | tor8: OK, various fixes on robin/master | 17:13.49 |
| | Was there anything else I was supposed to do before returning to SOT duties? | 17:15.47 |
| | Did you want the list of changed features? | 17:15.59 |
| Gin | Hi | 17:29.15 |
| | I've been stuck getting mupdf to build for android for days now. ndk-build results in an error: jni/mupdf.c:383:21: error: 'fz_stream' has no member named 'bp' What am i missing? | 17:30.35 |
| Robin_Watts | Gin: You need the commits that are currently waiting here: | 17:31.36 |
| | http://git.ghostscript.com/?p=user/robin/mupdf.git;a=summary | 17:32.02 |
| | You need "Various android fixes", "Fix Android build w.r.t mujs." and "Solve crash ..." | 17:32.29 |
| | They should hopefully make it into the trunk today or tomorrow. | 17:32.43 |
| | Careful, they are almost 20 mins old :) | 17:32.57 |
| Gin | cool, how do i get them? I'm new to gitly | 17:33.34 |
| Robin_Watts | Never used gitly :( | 17:34.09 |
| Gin | I meant git | 17:35.39 |
| Robin_Watts | oh, well, maybe: git add remote robin http://git.ghostscript.com/user/robin/mupdf.git | 17:36.43 |
| | THen git fetch robin | 17:36.49 |
| | Then you can git cherry-pick them across. | 17:36.58 |
| tor8 | Robin_Watts: no, that's fine. I'll compile the changelist, you go on into SOT hell :) | 17:39.30 |
| Robin_Watts | tor8: Want what I've got so far? | 17:39.41 |
| | tor8: If you can review the stuff on robin/master, we can push it, and Gin can download it directly. | 17:41.42 |
| | pedro: Where can I find a copy of the OOXML specs? | 17:51.05 |
| pedro_ | has them locally, but I'll see if I can dig out the link | 17:51.34 |
| Robin_Watts | And presumably you've got some cunning way of seeing the contents unpacked? | 17:51.38 |
| pedro_ | its as cunning as copy the docx, rename to .zip and extract :) | 17:52.04 |
| | there are tools in the openxml sdk though | 17:52.15 |
| Robin_Watts | OK, so it's akin to xps. Ta. | 17:52.16 |
| pedro_ | http://www.ecma-international.org/publications/standards/Ecma-376.htm | 17:54.38 |
| Gin | woot! Thanks Robin, works a charm :) | 17:54.55 |
| pedro_ | Robin_Watts> I have OLE format docs too, but there's better (newer) stuff on the MS website for that | 17:55.11 |
| Robin_Watts | pedro_: Thanks. | 17:55.14 |
| pedro_ | will stick the specs up on casper sometime soon | 17:56.08 |
| | or stick them on the wiki - may be better | 17:56.26 |
| Robin_Watts | was about to suggest wiki. | 17:56.34 |
| | pedro_: You've probably twigged to this already, but if you 'start' the XML, it gets shown nicely in IE. | 17:59.25 |
| | At last, a use for IE :) | 17:59.32 |
| pedro_ | yup - quite nice to have it formatted - its a single line xml :) | 18:01.06 |
| | Robin: addd links to the ECMA specs and to MSDN site descriptions for XML and OLE files for office (GhostDocs Wiki) | 18:06.52 |
| pedro_ | disappears for food - biab | 18:07.03 |
| Robin_Watts | pedro_: Ta | 18:07.11 |
| | We should all upvote this: https://bugreports.qt-project.org/browse/QTCREATORBUG-10834 | 18:30.27 |
| Jogux | robin_watts : starting with you? :-) | 19:17.23 |
| Jogux | has voted, anyhow | 19:18.15 |
| Robin_Watts | Jogux: I created it. Not sure I can upvote it :) | 19:18.34 |
| Jogux | ah, yes, don't think you can actually | 19:19.59 |
| tor8 | Robin_Watts: yes, sure. | 19:34.38 |
| | Robin_Watts: 4 commits on robin/master LGTM | 19:35.50 |
| Jogux | htf have I managed to get a whole bunch of merge commits using rebase to squash my 2 latest commits together?! | 19:45.57 |
| henrys | tor8, Robin_Watts are the Canva bugs relevant to anyone else? My impression is they are not. | 19:47.24 |
| Robin_Watts | henrys: How many are there? | 19:48.12 |
| | There's the "off by 1" in the color component one. | 19:48.25 |
| henrys | they want 01,00 695085 and 694623 fixed | 19:49.13 |
| | so the other one is the one I'm not sure about. | 19:49.24 |
| Robin_Watts | bug 695085 is probably simple to fix (famous last words) | 19:49.49 |
| | The other one is hard. I have no idea how to solve it. | 19:50.12 |
| Jogux | whatever henry just said has made ircii go crazy, all the text is flashing now! | 19:50.32 |
| henrys | Jogux: it was a copy paste ? | 19:51.20 |
| | the bug numbers | 19:51.26 |
| Jogux | henrys : hm, do you have the same mac problem I was having the other day? switching system preferences -> keyboard -> text -> "use smart quotes and dashes". off solved that for me | 19:51.42 |
| | the logs say they bug numbers have a whole bunch of funny characters before/after them | 19:51.57 |
| Robin_Watts | Jogu: I see them as having a different background colour. | 19:52.20 |
| | I mean, I could maybe code something that antialiased the edge of images, but it would have problems with | 19:52.21 |
| mvrhel_laptop | Robin_Watts: do you have a sec? | 19:52.27 |
| Robin_Watts | abutting images. | 19:52.28 |
| | mvrhel_laptop: Sure. | 19:52.32 |
| | tor8: THanks. | 19:52.40 |
| mvrhel_laptop | simple question so in the fizt block/lines/spans of text, are the spans thought of as different columns in the same line? | 19:53.16 |
| | or am i reading to much into that | 19:53.36 |
| Robin_Watts | IIRC, the spans all share the same style. | 19:53.38 |
| | So if have some text with /italics/ in it, the spans would be: | 19:54.01 |
| mvrhel_laptop | ah ok so if I had a line, that had different fonts etc, then you | 19:54.05 |
| | right | 19:54.08 |
| Robin_Watts | (or might be...) | 19:54.10 |
| mvrhel_laptop | ok that is fine then | 19:54.18 |
| Robin_Watts | "So if I had some text with" "italics" "in it, the spans would be:" | 19:54.27 |
| | After analysis, it's possible that the spans could represent different columns. | 19:54.45 |
| | I think I annotate the spans with column numbers, maybe. | 19:54.56 |
| mvrhel_laptop | ok. I have enough to dig myself into a deeper hole with now | 19:55.22 |
| marcosw | Robin_Watts and ray_laptop: the bmpcmp issues are not with the bmpcmp program itself, the input files are not being generated due to the following error in the ghostscript command line: Error: /undefinedfilename in (extras=-dGrayValues=16) | 20:02.14 |
| | it looks like the extras= clusterpush command isn't correctly being processed by the clustermaster. | 20:02.50 |
| Jogux | robin_watts : new version of my release script in my repo; I've simplified the parameters (release.py appstore-ios works now), and you can throws flags on the end if you want. and I fixed it to allow comments anywhere. | 20:23.21 |
| | and, instantly, I decide "wouldn't it be great if it had a #include type operator" | 20:25.31 |
| | I presume no one objects if I generate myself a promo code for SO+? | 20:26.42 |
| | [does anyone else want one?] | 20:29.14 |
| henrys | Jogux: we can find all the answers to our problems here: http://www.computerhistory.org/atchm/microsoft-word-for-windows-1-1a-source-code/ ;-) | 21:28.59 |
| spanners | "Remember, this was a time when a typical personal computer might have an 8 Mhz processor, 1 megabyte of memory, a 20 megabyte hard disk, and a floppy disk drive. How did Word accomplish so much with so little?" | 21:34.33 |
| | ISTR plenty of other word processors, equally capable, which didn't run like dogs. >:-| | 21:35.08 |
| Robin_Watts | Jogux: #include fills me with dread. | 22:51.56 |
| | It means you can't review the history of a build by watching a single file. | 22:52.15 |
| | You have to watch a web of them. | 22:52.20 |
| | Forward 1 day (to 2014/04/02)>>> | |
IRC Logs