| <<<Back 1 day (to 2013/05/29) | 2013/05/30 |
robin_watts_mac | sebras: My nook is api-7 :) | 00:18.13 |
| I haven't run tor8's whitespace stuff, no. This was all done on the plane, so ignore the whitespace for now. | 00:18.49 |
| sebras: I'm expecting the fz_pixmap_set_resolution to be batted about a bit before it goes in. | 00:19.17 |
| basically PCL needs the resolution, and we have resolution fields in the pixmap, but we don't have any nice way to set them programmatically. | 00:19.48 |
| possibly we need to set x and y independently. | 00:19.57 |
| the foreach patch is tor8's. I'm generally not in favour of such macroisation. | 00:20.46 |
| but I'm prepared to be convinced on a case by case basis. | 00:21.00 |
| and the subdiv stuff is a nice idea, but it's flawed. | 00:21.24 |
| But our big customer likes it. | 00:21.46 |
| so we could maybe do with finding a way to have it in as an option. | 00:22.58 |
errorist | I use scanimage command to get image from a scanner. I would like to get the output to a nice pdf file. how should I do that? A4 pagesize. all standard. Seems like scanimage can scan to two filetypes; ppm and tiff. any idea? | 08:56.24 |
kens | ImageMagick ? | 08:56.43 |
errorist | ah, good idea! didnt know imagemagick did pdf | 08:57.30 |
kens | I don't know if it does for sure. | 08:57.50 |
| Google 'Convert TIFF to PDF' results in numerous hits | 08:58.18 |
| Another solution, using Ghostscript : | 08:59.57 |
| http://stackoverflow.com/questions/15211428/conversion-tif-to-pdf-ghostscript | 08:59.57 |
| libtiff apparently provides a tiff2pdf tool | 09:01.06 |
errorist | nice!! strangely convert works, but mogrify dont. | 09:01.47 |
| tiff2pdf is a good idea too :) | 09:02.08 |
| convert foo.ppn bar.pdf works but mogrify foo.ppn bar.pdf dont :p | 09:02.42 |
kens | Can't help with that... | 09:02.57 |
tor8 | errorist: mogrify -format pdf foo.ppn | 09:03.33 |
| (or some variation of the theme) | 09:03.58 |
sebras | tor8: to whom do I send info like the ones from late yesterday evening..? | 09:24.10 |
kens | sebras, probably us, though you can contact Miles as well. It might be easier if you let us know and we tell Miles. | 09:25.12 |
sebras | kens: alright, then I was doing this correctly. :) | 09:26.01 |
tor8 | sebras, I was going to say Scott, but probably Miles would be better for judging whether to send off nasty emails or not | 09:26.17 |
kens | I would say so, yes. But I'm not sure. You think better to tell Miles directly tor ? | 09:26.40 |
tor8 | probably better to filter it through us (or cc us) | 09:27.07 |
| is the artifex staff list moderated? | 09:27.21 |
kens | I htought you and Robin oculd at least do a quick pass and determine whether they are likely commercial customers, and/or a quick vet of GPLness | 09:27.55 |
| I don't hink any of our lists are moderated are they ? | 09:28.13 |
sebras | kens: tor8: is it better to send you guys emails or to post here? | 09:53.04 |
kens | sebras both are good, but I suspect email is safer, just in case | 09:53.25 |
sebras | ok. | 09:53.32 |
kens | tor8 will you write up sebras's findings of last night for Miles ? Or should I do it ? | 09:53.56 |
tor8 | kens: I'll do it, unless you want to | 10:02.25 |
kens | I'd rather not :-) | 10:02.36 |
tor8 | alright, then I'm on it :) | 10:03.19 |
kens | :-D | 10:03.26 |
sebras | tor8: if you want to check the .apk's I can pass you the files tonight. | 10:04.34 |
| tor8: I sniff off of the wifi while installing them on my phone. well... my employers phone. | 10:05.04 |
| s,rs,r's, | 10:05.28 |
tor8 | sebras: the perfectviewer pdf plugin is open source as well, so I believe they're in the clear | 10:10.19 |
| seeing as it's a separate install and all | 10:10.33 |
| sebras: if you have the foobnik and anttek apks available, I'd like to take a look at them | 10:14.09 |
robin_watts_mac | There is a google chrome extension that lets you download apks direct from google play. | 10:19.35 |
| When you visit the page a new icon appears in the bar that downloads the apk. | 10:20.03 |
tor8 | hi robin! | 10:22.31 |
| up early? | 10:22.33 |
robin_watts_mac | tor8: yeah. just going for a run. | 10:22.58 |
tor8 | robin_watts_mac: re whitespace, just run "bash scripts/gitsetup.sh" and then rebase and whitespace will be fixed | 10:23.45 |
robin_watts_mac | tor8: Is there a similar simple invocation for removing the setup for when I horribly break it? | 10:24.29 |
tor8 | rm .git/hooks/pre-commit | 10:24.51 |
| robin_watts_mac: is it "APK Downloader"? | 10:29.07 |
robin_watts_mac | tor8: might be. | 10:29.40 |
| I don't have it on the laptop. | 10:31.26 |
tor8 | robin_watts_mac: so happy I ditched chrome... | 10:31.47 |
| google has blacklisted the extension, refuses to install it | 10:32.06 |
| it gives me this helpful message. https://support.google.com/chrome_webstore/answer/2664769?p=crx_warning&rd=1 | 10:32.56 |
| jerks. | 10:32.58 |
robin_watts_mac | odd. I have a version that works for me still on the pc at home. I will remote desktop in and look later | 10:35.13 |
tor8 | can't install the newest version from the website | 10:35.34 |
| the version in the chrome web store is outdated and doesn't work | 10:35.45 |
| google has really cranked up the evil meter of late. soon they will surpass apple in evilness. | 10:36.36 |
robin_watts_mac | only when their leveraged takeover of North Korea goes through. | 10:38.28 |
| http://apps.evozi.com/apk-downloader/ | 10:41.03 |
chrisl | thinks tor8 has been possessed by LPD: he's adding crazy macros with lower case names, and has been spotted talking about garbage collection (of sorts)....... | 10:45.35 |
tor8 | robin_watts_mac: yeah, that's the one that chrome refuses to install. | 10:47.02 |
sebras | tor8: legendary pink dots! | 10:50.36 |
tor8 | http://www.barnesandnoble.com/w/mupdf-upsilon-studio/1112525841 that one's not ours I believe... | 11:24.31 |
paulgardiner | I think I may have created a cyclic dependency: crypto-pkcs7-openssl.c (which I've added to fitz) relies on generated. Does generated rely on fitz? | 11:29.20 |
tor8 | paulgardiner: no, generated is independent | 11:30.25 |
| cmapdump requires some of fitz, but includes it itself via #include, not by using the library | 11:30.47 |
paulgardiner | Oh okay. I must have messed up something in generating the quoted CA file. | 11:32.18 |
tor8 | paulgardiner: I fixed it locally when testing your old openssl branch like this: | 11:33.18 |
| $(OUT)/signatures.o : $(ADOBECA_HDR) | 11:33.26 |
| don't depend on the "generated" pseudo-target itself | 11:33.55 |
paulgardiner | Right. thanks | 11:34.11 |
tor8 | that pseudo-target is there primarily for use when cross compiling and you don't want/need the apps | 11:34.39 |
| like the ios build script, which calls "make generate" with the host compiler, and them "make libs" with the cross compiler | 11:35.13 |
paulgardiner | I don't think I added any dependency. I was just assuming something ensured generated was handled first | 11:35.14 |
| Actually. I cattempted to copy what had worked for pdf_util.js | 11:37.07 |
tor8 | I'm pretty strict about keeping makefile dependencies working | 11:37.51 |
| but not necessarily accurate, a few false positives are okay (due to header files mostly) | 11:38.14 |
| but I can't abide missing dependencies... too many times have I been bitten by broken builds | 11:38.34 |
paulgardiner | I see it now. When looking at JAVASCRIPT_HDR, I missed a line just like the one you've just suggested | 11:39.08 |
tor8 | the one major annoyance I have is when a libfoo.a library has old object files from sources that have been renamed | 11:39.15 |
| which I think I just wrote a patch for :) | 11:45.37 |
paulgardiner_ | Fell off the internet for a minute there | 11:47.21 |
tor8 | paulgardiner: nothing of import happened :) | 11:47.51 |
| paulgardiner: I've given the commits on paul/master a brief review | 11:48.04 |
| 1) you introduce the adobe_ca : crypt_pkcs7_openssl.c dependency before you introduce crypt_pkcs7.c | 11:48.32 |
| also, the way the makefile is set up now it'd be easiest to have just a crypt_pkcs7.c with a #ifdef HAVE_OPENSSL around the implementation | 11:49.10 |
| (so combined _openssl and _null in one file) | 11:49.20 |
| I'm heading out for lunch now, I'll back in an hour or so | 11:50.26 |
paulgardiner | tor8: ping me when you get back | 11:52.24 |
tor8 | paulgardiner: ping. | 12:40.20 |
paulgardiner | Yeah, I could swap to using an ifdef if you like. That possibly wouldn't be sp good if ever we added another version that used something other than openssl. | 12:41.56 |
tor8 | paulgardiner: I don't think that'll ever happen, to be honest. openssl seems to be the one and only open source solution, and I doubt we'll want to hack our own certificate toolchain. | 12:42.58 |
paulgardiner | Okay. I'll swap it over | 12:43.15 |
tor8 | and if we do, changing back won't be hard :) | 12:43.17 |
| better to have simplicity now than suffer lots of small annoyances by being too "general" too early. | 12:43.41 |
robin_watts_mac | tor8: I think I fixed the whitespace issues in the pcl commit. It's probably not quite ready as is, because it doesn't do color yet, and yet the hooks are all there for color, but it would be good to get your opinion on the approach to the way it handles options etc. | 13:33.50 |
tor8 | you're also contaminating the namespace by not prefixing a lot of macros with FZ_ | 13:34.17 |
| I'll look it over. I think it's about time to split the header files into something more manageable. | 13:34.46 |
robin_watts_mac | tor8: yeah, I lifted the macros direct from gs. | 13:38.26 |
| We *could* hide the macros in the .c, and only expose preset printer combinations, but that precludes people adding more printer combos of their own. | 13:39.06 |
henrys | robin_watts_mac:are you at the show? | 13:40.05 |
tor8 | robin_watts_mac: maybe we should add a generic "option string" mechanism for the output formats | 13:42.47 |
| robin_watts_mac: also, I think a "fz_print" (without the trailing f) for just printing a string without printf formatting would be nice, so we don't have to worry about malicious input (and processing overhead) | 13:44.45 |
| or fz_puts | 13:44.58 |
robin_watts_mac | henrys: show starts in just under 2 hours. | 13:46.25 |
| tor8: is that not fz_write ? | 13:46.45 |
| well, fz_write with strlen. | 13:46.56 |
sebras | tor8: I vote for fz_puts() for consistency. | 13:47.00 |
robin_watts_mac | I agree | 13:47.16 |
henrys | robin_watts_mac:you should go to paul revere's house and tell 'em you've arrived. | 13:49.30 |
tor8 | regular stdio puts adds a newline, which I don't think is very useful for us | 13:51.03 |
| but that may be surprising if we mismatch | 13:51.23 |
robin_watts_mac | tor8: hmm, not keen on adding\n. I'm sure gs equivalent doesn't. | 13:52.44 |
tor8 | robin_watts_mac: well, fputs doesn't add a newline. hello for consistency... | 13:54.34 |
paulgardiner | tor8: updated signature commits on paulg/master. Cluster seems happy with it | 13:55.33 |
tor8 | paulgardiner: looks good superficially. I haven't looked too closely at the openssl specific code but I'll trust you on that :) | 14:01.11 |
| want me to push? | 14:01.13 |
paulgardiner | Yes please | 14:01.30 |
tor8 | paulgardiner: done. | 14:02.57 |
paulgardiner | ta | 14:03.12 |
robin_watts_mac | paulgardiner: The 2 android commits on robin/master are good to go. | 14:17.55 |
| tor8: the fz_puts commit is there now too. | 14:18.05 |
| I'll update the PCL stuff to use it. | 14:18.22 |
paulgardiner | robin_watts_mac: will we need to keep adding back the "debuggable=true" line for development? | 14:20.15 |
robin_watts_mac | paulgardiner: Supposedly, no. | 14:20.34 |
tor8 | robin_watts_mac: a bash script to automatically update version numbers etc in the strings and manifest xml file, would that be something desirable? to automate the production of all the different apks as well. | 14:20.47 |
robin_watts_mac | In newer SDKs doing a debug build automatically sets the debuggable flag. | 14:20.55 |
| tor8: yes, that would probably be desirable, but non trivial. | 14:21.09 |
| I can't call ndk-build under bash. | 14:21.18 |
| so I have to swap back and forth between cmd window (or cygwin) and bash. | 14:21.32 |
tor8 | robin_watts_mac: ick. | 14:21.38 |
paulgardiner | robin_watts_mac: the two android commits look good to me | 14:21.50 |
tor8 | bat file that runs bash with the setup script and then ndk-build? | 14:21.56 |
chrisl | robin_watts_mac: cygwin runs bash | 14:22.19 |
robin_watts_mac | chrisl: But I can't run ant under cygwin. | 14:23.20 |
| tor8: If you can make it work, I'll be all for it! :) | 14:23.33 |
chrisl | robin_watts_mac: oh, that wasn't how I read it above...... | 14:24.05 |
tor8 | robin_watts_mac: first three on robin/master look good. want me to push them? | 14:24.50 |
robin_watts_mac | Gotta swap location to the exhibition hall. bbiab. | 14:24.50 |
| tor8: please. Thanks. | 14:24.57 |
chrisl | robin_watts_mac: you can run run windows cmd from inside cygwin | 14:25.14 |
tor8 | robin_watts_mac: Core2.mk is that still required, or has the newen NDK fixed the bug? | 14:57.27 |
robin_watts_mac | tor8: pass. | 15:13.08 |
| Where the hell is itunes on my mac? I want to copy files to the ios version of mupdf. | 15:13.40 |
| can't see for looking. | 15:13.55 |
tor8 | robin_watts_mac: you can also open a pdf from safari, click the "open in other app" button thing, and select mupdf there and the pdf will be copied into the mupdf library | 15:14.44 |
robin_watts_mac | tor8: Ah, thanks. | 15:17.40 |
| where did I go? Has my house burned down? | 16:35.32 |
ray_laptop | grr... somehow my VS 2008 won't build anymore :-( It looks like 'link' is running something else (like it's a batch file). Too bad cmd doesn't have a 'which' command. | 16:53.30 |
| it happens with either a cmd nmake OR the VS "Build" | 16:54.04 |
| oops. Found it. Somehow a 'link.cmd' was laying around in my gs/ directory (I forgot to get rid of it after making a custom link step) | 16:57.12 |
ray_laptop | blushes | 16:57.17 |
robin_watts_mac | did someone mess with the mupdf build files recently? I could build yesterday, but today it can't fins X11/Xlib.h | 17:13.33 |
tor8 | robin_watts_mac: I did. which system? | 18:09.00 |
| paulgardiner: compiling crypt_pkcs7.h on macosx gives lots of deprecated warnings | 18:16.05 |
| void X509_STORE_free(X509_STORE *v) DEPRECATED_IN_MAC_OS_X_VERSION_10_7_AND_LATER; | 18:16.24 |
paulgardiner | That's going to be a pain. Some of that code is altered bits taken from other areas of openssl. I guess we could always create our own build for MAC OS | 18:18.22 |
tor8 | I wonder why it's been deprecated though | 18:18.49 |
| okay, a quick search later has the answer: they're deprecating openssl in favour of apple's own "CommonCrypto" library | 18:21.20 |
| apparently that one is open source as well | 18:22.48 |
robin_watts_mac | tor8: On macos, doing debug builds. | 18:30.07 |
tor8 | there's a patch on tor/master to fix it | 18:30.21 |
robin_watts_mac | tor8: Thanks. | 18:49.33 |
| Both look good. | 18:50.59 |
| Shall I push them? | 18:51.03 |
tor8 | robin_watts_mac: sure. | 18:53.19 |
robin_watts_mac | Thanks. | 18:53.27 |
sebras | tor8: I managed to get quadwheel to compile and ran the testsuite just now. | 20:21.46 |
| tor8: 16 of 53 tests are failing. but I'm working on it. :) | 20:22.18 |
robin_watts_mac | tor8: how odd. Previously when I built on macos I had to link with the ia32 version of v8. | 20:25.57 |
| Now I'm having to link with the x64 version. | 20:26.09 |
henrys | robin_watts_mac: I take it the show has pretty good wireless? | 20:26.34 |
robin_watts_mac | yeah, seems to. | 20:26.47 |
| I had a nice chat with some guys at Nvidia earlier. | 20:27.06 |
| They are keen on getting people to use their path rendering on a gpu stuff. | 20:27.22 |
henrys | robin_watts_mac: ah that's interesting | 20:27.35 |
robin_watts_mac | so if we ever want to look at HW acceleration using GPUs, they are a good point of contact. | 20:27.51 |
| apparnetly they've been talking to FoxIt to help accelerate their PDF rendering on multicore devices. | 20:28.11 |
henrys | robin_watts_mac: it really is an area I've wanted to explore time permitting. | 20:28.15 |
robin_watts_mac | yeah, I've seen it on several staff meeting agendas I think. | 20:28.34 |
| or it's been mentioned at least. | 20:28.43 |
henrys | robin_watts_mac: any potential customers? | 20:29.15 |
| at the show | 20:29.49 |
| I worry that if foxit does work in that area we will be less competitive. | 20:30.56 |
robin_watts_mac | yeah, maybe. | 20:31.07 |
sebras | robin_watts_mac: is the performance improvement significant? | 20:31.24 |
robin_watts_mac | though when I ran mupdf on their device we couldn't see any slowdown anywhere. | 20:31.41 |
| sebras: They say "It's just incredible" | 20:32.10 |
| but the things they were talking about were 'making stuff save to run multithreaded' etc. | 20:32.37 |
sebras | robin_watts_mac: maybe on torture_test.pdf (the chip design)... | 20:32.38 |
robin_watts_mac | which we already have. | 20:32.52 |
| I suspect the path rendering on GPU is more for desktop GPUs. | 20:33.08 |
henrys | robin_watts_mac:I did think apple, google and friend were trying to push folks into the diminutive opengl | 20:34.13 |
robin_watts_mac | customers - we had a guy asking about wanting to handle PDF files with 3d data in. | 20:34.19 |
sebras | robin_watts_mac: U3D? | 20:34.34 |
robin_watts_mac | sebras: maybe - I forget the name. | 20:34.51 |
henrys | OpenGL es that is. | 20:35.05 |
robin_watts_mac | and we met a PWG guy from Xerox. | 20:35.06 |
sebras | robin_watts_mac: I believe U3D is the one that adobe put in the spec. | 20:35.06 |
robin_watts_mac | henrys: yea, I think the path rendering stuff *may* be device specific | 20:35.32 |
henrys | robin_watts_mac: and he wants this on a phone? | 20:35.38 |
| 3d? | 20:35.46 |
robin_watts_mac | Yes. | 20:35.50 |
| I said that we could certainly get the data out of a PDF for him, but we had no code to handle it at the moment. | 20:36.20 |
henrys | robin_watts_mac: maybe something for the agenda. It would be interesting to work on. | 20:36.51 |
robin_watts_mac | it would. | 20:36.58 |
henrys | noted | 20:37.10 |
robin_watts_mac | The problem with an OpenGL ES pipeline is that once you're *into* textures, it's very hard to get the pixels back. | 20:37.40 |
| which means you have to do *everything* in opengl es (past the point at which you first use it). | 20:38.12 |
henrys | so what is their formula for transparency, do they support all the operations? | 20:38.24 |
robin_watts_mac | He said yes, they supported everything as it's all shader based. | 20:38.41 |
henrys | robin_watts_mac: is there anything different between an open gl es pipeline and an nvidia pipeline? | 20:39.50 |
robin_watts_mac | nvidia has extra extensions I think. | 20:40.06 |
henrys | why would we tie ourselves to a vendor? | 20:40.07 |
sebras | tor8: after fixing some more whitespace stuff only 6 tests are failing. | 20:40.10 |
robin_watts_mac | henrys: I'm not sure. | 20:40.16 |
henrys | say hi to Miles for me. I have a quick errand to run bbiab | 20:40.58 |
robin_watts_mac | I don't know if the path rendering via GPU is nvidia specific, or whether it's a cunning use of generic ops. | 20:41.04 |
| but often these vendor specific extensions wind up in the next version of the general spec (if they are useful). | 20:41.42 |
sebras | tor8: 3... | 20:59.21 |
| two of them is documented to be failing though. | 20:59.47 |
| and the third one is testing expression evaluation. and I messed around with the handling of doubles... | 21:00.32 |
robin_watts_mac | tor8: Makefile tweak up for review. | 21:17.02 |
sebras | there are quite a few leaks as well. hm... with some more work it might be possible to get quadwheel to a useable state. | 21:17.41 |
robin_watts_mac | BB10 stuff looks really cool. | 21:21.35 |
| There looks to be a PDF sized hole in their software offerings though. | 21:22.06 |
tor8 | robin_watts_mac: pushed. | 21:45.05 |
henrys | robin_watts_mac: it would be great if they could stage a comeback. | 21:45.43 |
| Forward 1 day (to 2013/05/31)>>> | |