| | <<<Back 1 day (to 2014/08/24) | 2014/08/25 |
| kens | OK so it seems generating an appearance for a missing Tx field in an AcroForm just doesn't work. Now to figure out why :-( | 07:55.11 |
| jogux | it'd probably make sense to package the iOS mupdf as a proper iOS framework, should avoid questions like the one over the weekend. packaging it as a cocoapod might also be good, lots of people use those to import third party sources into their apps and it makes sure all the compiler setting are right etc... | 10:29.17 |
| kens | didn't understand any of that :-) | 10:29.38 |
| Robin_Watts | either. | 10:36.15 |
| paulgardiner | Coo, I did... but would have no idea how to do it :-) | 11:04.10 |
| | Sounds well worth doing though | 11:04.28 |
| Robin_Watts | I vaguely understood it as "It would be nice to package mupdf into a form that other ios users could easily reuse", and that does indeed sound nice. | 11:07.40 |
| | If that's just a "take the existing unchanged code and wrap it into a specific format xcode project" then great. | 11:08.36 |
| | Except for the obvious fact that xcode is a huge steaming pile of shite that seems to break everything between every release, | 11:09.13 |
| | If it's a "just write this new layer of interface code to cope with apples latest ideas of how to interface between things" then fuck that. | 11:09.47 |
| chrisl | Now, Robin_Watts, I'm tired of you holding back, I do wish you'd just say what you mean..... ;-) | 11:10.53 |
| Robin_Watts | chrisl: Don't get me started on ms word... | 11:11.39 |
| kens | No need, seems the SF authors already did | 11:12.10 |
| chrisl | WordStar is where it's at! | 11:13.46 |
| tor8 | Robin_Watts: two commits on tor/master ready to go, two others up for discussion | 11:21.38 |
| ken488 | Hi guys. I have a question for you about MuPDF | 11:26.18 |
| | If you load a PDF page with some text on. Is it possible to remove that text, and possible save a .png file without the text in but just the picture?? | 11:26.58 |
| Robin_Watts | ken488: Using vanilla MuPDF, no. | 11:27.44 |
| | But using a slightly hacked version, sure. | 11:27.50 |
| tor8 | ken488: if you're just looking to get the images from a pdf, you can extract them using "mutool extract" | 11:28.09 |
| | but rendering a page without the text, not with vanilla, but it should be an easy hack to modify it to do so | 11:28.35 |
| Robin_Watts | tor8: I only see 2 new commits. | 11:28.39 |
| ken488 | I know, thanks :) I know how to extract text etc. But I would like to know if it is possible to remove the text from a PDF page, so you only have the picture left, where the text was placed.. But it seems like that is not possible | 11:29.06 |
| | I will have a look into it :) Thanks! | 11:29.25 |
| tor8 | Robin_Watts: 0afbc91 Fix uninitialized variable, d291833 Add a paragraph, 6287c81 Revise test-device, cf81a4b (tor/master, master) WIP test device | 11:29.27 |
| | maybe you already approved the first two before the weekend and I forgot to push | 11:29.48 |
| Robin_Watts | I did. | 11:29.54 |
| tor8 | Robin_Watts: ah, thanks. then ignore those and we can fight about the test device :) | 11:30.19 |
| Robin_Watts | The revised test-device. You don't allow a configurable threshold. | 11:31.05 |
| tor8 | I don't, that's in the following commit which I'm unsure of | 11:31.27 |
| Robin_Watts | And the hardwired threshold you have put in is applied as being 'differences between components' not 'difference from greyscale'. | 11:31.41 |
| tor8 | it adds complication and I'm not convinced it needs to be configurable | 11:31.43 |
| | Robin_Watts: that ends up being equivalent | 11:32.08 |
| Robin_Watts | The customer has specifically requested a threshold. | 11:32.24 |
| tor8 | Robin_Watts: it's the same test (and threshold) as is hardcoded in the equivalent ghostscript test | 11:32.24 |
| Robin_Watts | Right, and gs is being updated to have a user specified threshold I think. | 11:32.47 |
| tor8 | Robin_Watts: right, in that case we should merge in the WIP commit which adds the threshold | 11:33.15 |
| Robin_Watts | The WIP one looks wrong to me. | 11:34.05 |
| | threshold is a float, so presumably between 0 and 1. | 11:34.20 |
| | so it's application in is_rgb_color_u8 is wrong. | 11:34.33 |
| tor8 | it's pre-multiplied by 255 before the call to is_rgb_color_u8 in fz_test_fill_image | 11:35.05 |
| Robin_Watts | tor8: Hmm. My code worked with threshold being the sum of the squares of the component differences. | 11:36.28 |
| | So it's a more accurate difference. | 11:36.51 |
| tor8 | mine is the same as gs, the absolute color value difference from grayscale to trigger | 11:36.57 |
| Robin_Watts | Right, but gs is 'suboptimal'. | 11:37.18 |
| tor8 | the distance from equivalent grayscale you had is strictly better, but it's harder to explain to non-computer-graphics people | 11:37.30 |
| | in any case, I feel we should be using the same math in both places | 11:38.10 |
| Robin_Watts | I think you don't need much maths before you appreciate that the difference between things is calculated using pythagorus. | 11:38.34 |
| tor8 | and if we're reworking gs then maybe we should have the squared distance threshold rather than single-component difference | 11:38.45 |
| Robin_Watts | I'd contend that mine is better cos it's at least documented in the header files :) | 11:38.47 |
| | My code was also more flexible; it could be used to calculate the actual 'greatest distance from greyscale' for a page, rather than just "does this page exceed a given threshold". | 11:40.04 |
| | fz_new_test_device is broken in that it seems to call fz_new_device twice. | 11:40.26 |
| tor8 | Robin_Watts: oh, oops. yeah, that's a typo. | 11:40.53 |
| | Robin_Watts: I feel that a greatest distance from grayscale is too much precision, without actual accuracy. (red space character, etc) | 11:41.43 |
| | if you want that detail, get a histogram of a render | 11:41.53 |
| Robin_Watts | Well, this feels like a step backwards to me. | 11:42.16 |
| | At the very least it needs documentation in the form of some header comments. | 11:42.40 |
| | Mine was constructed the way it was,because I was hoping to offer a way for the device to bale out as soon as it had made the 'color' determination using the cookie. | 11:44.00 |
| | but that wasn't possible, so possibly pushing everything through a single function to record thresold differences was overkill. | 11:44.40 |
| | I think we should talk to ray/michael and see what their plans are for updating the gs device. | 11:45.48 |
| | If they plan to move to squared sum of differences, then we should adopt the same. | 11:46.03 |
| | squared sum of differences sounds better to me, but I'll go with whatever the majority decides. | 11:46.29 |
| tor8 | squared sum of difference would be my preference as well | 11:46.53 |
| | but michael may have a better grasp on the color science involved | 11:47.22 |
| Robin_Watts | I'm pretty sure that squared sum of difference would be consistent with how lcms interpolates colors etc. | 11:50.12 |
| tor8 | Robin_Watts: we should be using the proper rgb-to-grayscale conversion math then as well ( r * 0.3f + g * 0.59f + b * 0.11f) | 11:51.13 |
| Robin_Watts | For some definition of proper. But, yes, probably. | 11:51.41 |
| | tor8: Did you see the reply from the epub potential customer? | 12:02.52 |
| tor8 | Robin_Watts: yes. they wanted the moon on a stick by tomorrow, as henry put it | 12:03.28 |
| Robin_Watts | that was me, I think. | 12:03.39 |
| tor8 | oh, maybe it was you :) | 12:03.47 |
| | this push for EPUB 3 and "fixed layout" seems backwards to me ... why not just use PDF in that use case? | 12:03.59 |
| Robin_Watts | tor8: Yeah. | 12:04.17 |
| tor8 | but I guess it's all the HTML madness ... and media overlays and javascript interactivity and videos in ... books. | 12:04.26 |
| | *sigh* | 12:04.27 |
| Robin_Watts | I can understand (possibly) the idea of wanting to offer multiple fixed layouts. | 12:04.35 |
| | So as people change the font size, content like tables stays optimally laid out. | 12:05.08 |
| | but having a single layout seems bonkers. | 12:05.19 |
| tor8 | is that even possible? the fixed layout in EPUB 3.0 comes from the newer version CSS they include by reference with position: absolute | 12:05.19 |
| | but maybe that's possible with different media selectors in CSS | 12:05.36 |
| Robin_Watts | Right, I don't know what EPUB does. | 12:06.07 |
| tor8 | anyway, it's all madness and none of the current crop of e-ink based devices can barely do plain static html layout | 12:06.14 |
| | Robin_Watts: think of EPUB 3 as HTML5 in a zip-file | 12:06.30 |
| | with all the hype attached | 12:06.40 |
| | Robin_Watts: okay, somewhat cleaned up revised test-device on tor/master | 12:07.06 |
| | should be easy to change to use a different threshold metric should we decide to make gs based on a squared-distance metric as well | 12:07.47 |
| | Robin_Watts: so, if they want all this EPUB3 interactive madness, IMO just embed chrome or awesomium or firefox or something | 12:08.29 |
| Robin_Watts | Will fz_process_mesh accept NULL as a function argument. | 12:09.16 |
| | ? | 12:09.18 |
| tor8 | but if they want a good small-sized low-footprint EPUB 2 with "static" content we can deliver reasonably | 12:09.25 |
| | Robin_Watts: hm, actually, I think it might. if not, we could make it. | 12:09.38 |
| Robin_Watts | Possibly it should. And if we made it do so, you could pass NULL in. | 12:09.43 |
| tor8 | Robin_Watts: I'll make the necessary changes | 12:10.09 |
| Robin_Watts | tor8: random thought... indexed images... | 12:10.33 |
| tor8 | Robin_Watts: possible optimization ... too rare that I'll consider bloating the code with the special case? | 12:11.20 |
| Robin_Watts | I'd be tempted to say explicitly that threshold should be a value between 0 and 1. | 12:11.30 |
| kens | I don't think /IOndexed is rare | 12:11.41 |
| | Or even /Indexed | 12:11.47 |
| Robin_Watts | tor8: I don't care about speed. I was just wondering it we'd checked that it worked :) | 12:11.56 |
| kens | I'm fairly sure we have quite a number of example sin the test suite | 12:11.58 |
| Robin_Watts | but otherwise looks good. | 12:12.13 |
| tor8 | Robin_Watts: ah, right. well, they'll go through the cc.convert to rgb case | 12:12.45 |
| Robin_Watts | tor8: So possibly someone should write back to the potential customer saying what we feel we could offer. | 12:12.53 |
| tor8 | and the first non-gray color we hit will jump out of the loop | 12:13.04 |
| Robin_Watts | and why etc. | 12:13.06 |
| | Video and Audio aren't really our issues. | 12:13.40 |
| | Those would be down to the app writer. We can certainly give an API for them to get access to such data. | 12:14.00 |
| tor8 | Robin_Watts: yeah. I'm confident we could get fixed layout to work eventually, but in my estimate of 'by christmas' I was thinking of the more basic CSS layout stuff (just flowed text with floating images in the margins) | 12:14.05 |
| Robin_Watts | yeah. | 12:14.17 |
| tor8 | not any of the fancier new relative, absolute, madness rules | 12:14.25 |
| Robin_Watts | mmm. | 12:14.31 |
| | I think be honest, say what we feel we can deliver, and see what they say. | 12:14.55 |
| | (might be worth drafting a reply and running it past henrys). | 12:15.12 |
| tor8 | yeah. but once they have video and audio, they'll then want javascript, and while we do have a javascript interpreter, I'm not willing to implement the full HTML DOM | 12:15.14 |
| | if that is on the table, we should just say no | 12:15.26 |
| Robin_Watts | Better to under promise and over deliver etc. | 12:15.47 |
| tor8 | given that their target is text books, I do believe them when they say they want EPUB 3 with the fixed layout stuff | 12:16.03 |
| | and embedded video | 12:16.11 |
| | and then javascript to make the questions at the end of the book atuo-verify and all that insanity | 12:16.34 |
| | Robin_Watts: yes, I agree. we should let them know what we feel we are confident we can deliver and when and then let them see if they want that or the pain of embedding a web browser using the readium extension. | 12:17.27 |
| kens | chrisl Robin_Watts re bug #695444 I can't see any reason why we should do what is requested, this should just work with Unicode filenames, right ? What is required when using gsapi_init_with_args ? | 12:18.06 |
| Robin_Watts | kens: It should work. | 12:18.49 |
| | kens: gsapi_init_with_args always takes the args as 8 bit strings. | 12:19.20 |
| chrisl | Isn't he calling the API? So would need to convert to UTF-8 first | 12:19.20 |
| kens | Already, a difference of opinion :-) | 12:19.40 |
| Robin_Watts | What format the args is in depends on what gsapi_set_arg_encoding has been set (or some function like that whose name I forget) | 12:19.48 |
| kens | Hmm, let em look in the main WIndows source | 12:20.08 |
| chrisl | My first question would be: does it work in the gswin32/64 executable? | 12:20.41 |
| kens | Or cansomeone point me at documentation ? | 12:20.43 |
| | chrisl I bet it does | 12:20.49 |
| | Yes, it works fine | 12:21.40 |
| Robin_Watts | cor, I got the function name right. | 12:21.54 |
| | OK, so from the docs: | 12:22.04 |
| kens | Docs would be ideal | 12:22.10 |
| | I can point hiim at that then | 12:22.16 |
| Robin_Watts | By default we | 12:22.19 |
| | expect args to be in encoding 0 (the 'local' encoding for this OS). | 12:22.21 |
| | On Windows this means "the currently selected codepage". On Linux this | 12:22.23 |
| | typically means utf8. This means that omitting to call this function | 12:22.25 |
| | will leave Ghostscript running exactly as it always has. Please note | 12:22.27 |
| | that use of the 'local' encoding is now deprecated and should be | 12:22.28 |
| | avoided in new code. | 12:22.30 |
| kens | Which file Robin_Watts ? | 12:22.33 |
| Robin_Watts | API.htm | 12:22.41 |
| kens | OK thanks, I'll write something now. | 12:22.51 |
| Robin_Watts | He should call gs_set_arg_encoding to set UTF8 after calling gsapi_new_instance and before gsapi_init_with_args. | 12:23.20 |
| | and then obviously pass a UTF8 encoded string in to gsapi_init_with_args :) | 12:23.38 |
| jogux | oops. lost this window. robin_watts: packaging mupdf as a proper ios framework or cocoapod would essentially just be a packaging exercise, no new code needed I believe. | 12:58.19 |
| | we'd probably really want to add the standard 'ios appstore' exclusion to the license too though. (normal GPL & appstore T&Cs conflict otherwise, making it impossible to use mupdf in an opensource iOS app, you just need a clause saying "distribution on iOS appstore is not considered a violation" or something like that.) | 13:04.09 |
| kens | I'd be inclined to get Miles' rubber stamp on that before making such a statement..... | 13:05.53 |
| jogux | kens: oh, indeed, would definitely need to be approved by the appropriate parties. | 13:06.42 |
| tor8 | jogux: that clause would have to be carefully worded to not negate the GPL though | 13:07.50 |
| jogux | indeed. I believe there is a standard wording. | 13:08.04 |
| tor8 | OTOH, an "oh, you want it on the app store? pony up some $$$." is a compelling story for Miles | 13:08.32 |
| jogux | tor8: Yeah. I don't know where the line lies. I presume we'd probably want to encourage (or at least allow) mupdf to be used in opensource iOS apps, but I could be wrong. | 13:09.42 |
| tor8 | given how anti-open source the whole apple/ios ecosystem is these days, I don't feel too compelled to play nice with them | 13:11.37 |
| | but that's just me | 13:11.40 |
| henrys | tor8: zeniko said we should look at nvareader for infringement. Theyâve denied it, I also looked at the windows binary and didnât find anything, didnât check the mobile apps. Do you know anything about this? | 13:11.50 |
| tor8 | henrys: I haven't looked. | 13:12.06 |
| | henrys: I'd have to fire up a virtualbox to install the windows version to look | 13:14.44 |
| | the screenshots of their iOS app has elements that look *extremely* similar to our app though | 13:15.13 |
| | and the android app looks like a carbon copy of our android app | 13:15.34 |
| | so I'd say their denial is just so much hot air | 13:15.57 |
| kens | Want me to grab a copy ? | 13:16.14 |
| | Or maybe Robin woudl be better suited | 13:16.27 |
| tor8 | it's a paid app, so I think Robin or sebras would be best suited | 13:16.41 |
| henrys | tor8: I have the windows thing donât see any familiar symbols | 13:16.42 |
| kens | I htought there was a trial version | 13:16.55 |
| tor8 | henrys: the windows thing may indeed be different, let me try the install in a scrap winxp virtualbox | 13:17.07 |
| kens | Yes, thre's a free windows version | 13:17.15 |
| | The windows executable is certainly small | 13:17.52 |
| tor8 | kens: yeah. I was going to install that in a sandboxed environment to look at the exe file for suspicious strings | 13:17.54 |
| henrys | they may have just used our ui code for android and have their own core | 13:18.00 |
| tor8 | henrys: that sounds doubtful, given how tied our ui code is to the core | 13:18.13 |
| | the android app has the exact same annotation UI that we do (maybe different icons) but the page number overlays and preview of annotations looks spot on like ours | 13:18.47 |
| kens | Well, it won't load Bug692447.pdf... | 13:19.19 |
| tor8 | and the library view in the ios has the same delete icon, and the page scrubber on ios looks identical to ours | 13:19.20 |
| kens | The file selection dialog on Windows is frankly broken | 13:19.43 |
| | It says 'all supported document's and fails to find any. *,pdf shows loads though | 13:20.05 |
| | Ah, the Windows app only loads NVA files | 13:20.20 |
| henrys | tor8: some evidence from disassembly is preferable. Saying it looks like something is weak. | 13:20.29 |
| kens | I don't believe the windows app opens PDF files at all | 13:21.01 |
| jogux | I could have a quick look at the iOS app if wanted? | 13:21.10 |
| henrys | kens:it worked fine for me. | 13:21.32 |
| kens | Doens't open any PDF files for me | 13:22.02 |
| | http://forums.fofou.org/sumatrapdf/topic?id=3184315 | 13:22.04 |
| | NVA file avasilable here: | 13:24.05 |
| | http://www9.zippyshare.com/v/45868324/file.html# | 13:24.05 |
| tor8 | henrys: the windows app is a sumatrapdf reskin, that can't open pdf files | 13:24.11 |
| kens | just going to see if I can open that. | 13:24.15 |
| tor8 | but it handles CBZ files identically, the dialogs are the same, the menu options are the same | 13:24.27 |
| kens | The NVA file is 'basically' a PDF file it seems | 13:25.33 |
| | Changed %PDF to %NVA | 13:25.43 |
| henrys | kens:that post is from july I wonder if they obfuscated since then I donât see anything | 13:25.53 |
| pedro_mac | Failed to unpack Reader_NVA.pdblibmupdf.pdb | 13:26.13 |
| | thatâs from the binary for nvareader | 13:26.21 |
| tor8 | henrys: the windows exe has *tons* of mupdf strings | 13:26.23 |
| pedro_mac | mm | 13:26.37 |
| kens | henrys, that's entirely possible. It *looks* like MuPDF, and teh file format is certainly a mangled PDF file | 13:26.39 |
| tor8 | ..\mupdf\source\fitz\pixmap.c etc | 13:26.46 |
| kens | Well that seems like a giveaweay to me. | 13:26.56 |
| tor8 | libmupdf.pdb | 13:27.02 |
| | etc etc etc | 13:27.04 |
| kens | Seems it evne ahs some errors: | 13:27.58 |
| | Failed to unpack Reader_NVA.pdblibmupdf.pdbReader_NVA-no-MuPDF.pdbFailed to unpack libmupdf.pdb or Reader_NVA-no-MuPDF.pdb | 13:27.58 |
| henrys | tor8: give me a a specific string? | 13:28.47 |
| kens | \mupdf\source\pdf\pdf-lex.c | 13:29.00 |
| | Or just search for mupdf | 13:29.07 |
| | In the binary | 13:29.11 |
| henrys | kens: I did I must have a different binary | 13:29.21 |
| kens | I have the free windows version | 13:29.31 |
| henrys | NVA Reader.exe is the name of mine | 13:29.34 |
| kens | "NVA Reader.exe" | 13:29.44 |
| tor8 | henrys: I've uploaded the installed exe on casper in ~tor/NVA\ Reader.exe | 13:29.56 |
| | not the installer, so the copy on casper can be grepped for strings | 13:30.09 |
| kens | henrys 5,651,968 bytes, the installer was downloaded 10 minutes ago | 13:30.20 |
| | This is the installed version, not the NSIS installer which is what gets downloaded | 13:30.46 |
| tor8 | strings NVA\ Reader.exe |grep mupdf | 13:30.53 |
| | strings NVA\ Reader.exe |grep -i sumatra | 13:31.09 |
| | strings NVA\ Reader.exe |grep cannot | 13:31.14 |
| | for our unique error strings | 13:31.20 |
| henrys | oh silly me Iâm grepping the installer | 13:31.38 |
| kens | There is absolutely no acknowledgement of MuPDF, no GPL licence, no indication that its GPL, and no way to rebuild the app even if all the above were not true, so in no way GPL cvompliant | 13:31.47 |
| henrys | okay Iâll follow up with Scott and Miles | 13:32.01 |
| kens | Note from their blog: | 13:33.06 |
| | "We are planning a big update on Windows version with more functions. The structure of the application will be changed totally because we will use a NEW CORE." | 13:33.06 |
| tor8 | kens: you think they found a stolen copy of acrobat? ;) | 13:33.34 |
| kens | I suspect they are lying actually :-) | 13:33.51 |
| Robin_Watts | tor8: No. gsview. | 13:33.55 |
| kens | They seem to be Vietnamese, so probably the best we cna do is get them pulled from the app stores | 13:34.13 |
| tor8 | kens: we should equip robin with a big pointy stick and send him on paid vacation ;) | 13:34.57 |
| kens | :-) | 13:35.21 |
| Robin_Watts | Hmm. NVA = North Vietnamese Army, right? :) | 13:36.17 |
| kens | Ah, so its riddled with ,alware as well then. Best delete that VM | 13:36.44 |
| henrys | there are 100 of them so Iâd be careful with that stick | 13:38.37 |
| kens | No, they *say* there are 100 of them...... | 13:38.50 |
| tor8 | henrys: so, make it a multi-pronged stick? | 13:39.27 |
| henrys | Iâm pretty sure I could trade you guys in for about 100 of them so that seems about right ;-) | 13:39.54 |
| tor8 | henrys: what? only 100 of them!? I'm disappointed... | 13:40.22 |
| kens | We're too cheap clearly | 13:40.33 |
| Robin_Watts | https://www.youtube.com/watch?v=hw4wzwYeZ0Y | 13:41.05 |
| Robin_Watts | remembers spending hours making that run fast enough/properly on picsels flash player. | 13:41.35 |
| kens | So... You've been practicing with your stick and are confident of taking on 100 of them? | 13:41.51 |
| Robin_Watts | That's Darins job, surely? :) | 13:42.38 |
| henrys | Iâll have to check the numbers but a few years back I met a guy at a show and he was farming out a team of 100 Vietnamese engineers. | 13:42.41 |
| kens | Crumbs, an error messag from the GS PDF interpreter that I've never seen before O.O | 13:42.58 |
| tor8 | henrys: did you see our earlier discussion about epub? | 13:43.28 |
| henrys | I have a sneaking suspicion these are the same guys that stole cups and made it into a printing module ⦠but thatâs a skype discussion | 13:43.56 |
| Robin_Watts | henrys: That was china not vietnam, I believe. | 13:44.16 |
| henrys | no I think the company in china used these guys | 13:45.13 |
| Robin_Watts | oh... | 13:45.20 |
| henrys | tor8: no I didnât read the logs, I will look now | 13:45.57 |
| kens | continues playing whack-a-mole with the GS PDF interpreter | 13:46.48 |
| henrys | Robin_Watts, tor8 : yeah letâs tell them what we could do by christmas, what we can do in the near future, and what weâll never do. | 13:52.07 |
| | tor8: do you still have a mac, it seems the ibook implementation is pretty good. | 13:52.58 |
| | ? | 13:53.02 |
| tor8 | henrys: no, but I still have my iPad | 13:53.14 |
| | in an emergency, I could take my mac back from my sister | 13:53.46 |
| henrys | tor8: okay, we might ask them if they had some target of emulation - a product that we have to match but I donât know if that is important here. | 13:54.08 |
| tor8 | iBooks is probably the best of the best in terms of ebook readers available | 13:54.29 |
| | a shame it's mac/ios only and tied to the P.O.S. that is iTunes | 13:54.43 |
| mvrhel_laptop | mornign | 14:12.25 |
| | or morning rather | 14:12.37 |
| henrys | tor8: weâll try them again, I reread scottâs mail and itâs bad questioning. He just asks if they are distributing MuPDF⦠to which they could answer no. Iâve worded something better for him to send. | 14:22.53 |
| rayjj | I saw the logs about the gray detection and threshold. I agree that gs should be updated to have a dynamic threshold (parameter) and using the 'distance' instead of a component to component delta | 15:04.05 |
| | but having mupdf search the entire document for 'maximum difference from gray' seems like it would take longer on a color document | 15:05.09 |
| | I think tor mentioned 'first non gray color' jumps out and classes the page as 'color'. Is that how it's ending up ? | 15:05.59 |
| | oops. forgot to mention names (the above was for Robin_Watts and tor8) | 15:07.01 |
| Robin_Watts | Currently, when the page is found to be colorful, we set a flag to say it's colorful. | 15:07.32 |
| | And all the tests say if (!colorful) { ... check this new object } | 15:07.54 |
| | so once we detect that a page is colorful, subsequent objects are skipped quickly. | 15:08.13 |
| tor8 | and when the flag is set, we don't bother decompressing images or converting any other colors | 15:08.14 |
| | what robin said | 15:08.23 |
| Robin_Watts | We do NOT stop the enumeration of objects though (cos we can't in our device mechanism yet) | 15:08.38 |
| tor8 | s/yet// if I have anything to say about it :) | 15:09.08 |
| Robin_Watts | I did a version of the device that was capable to being asked to find the maximum difference from greyscale, but tor8 has dropped that in the latest version. | 15:09.29 |
| rayjj | Robin_Watts: tor8: OK. Thanks. I'll open a reminder bug to make gs operate similarly (maybe even use -T :-) ) | 15:14.25 |
| tor8 | we should decide which difference metric to use in both mupdf and gs, so we can keep them the same | 15:15.25 |
| rayjj | tor8: agreed. I don't know color theory well enough to know if the more time consuming 'distance from neutral' in RGB space is appreciably better than the simple threshold on the difference | 15:24.02 |
| Robin_Watts | rayjj: I suspect it makes relatively little difference when the threshold is small, and more when it's larger :) | 15:26.27 |
| mvrhel_laptop | unless you are willing to do distance measures in a perceptual color space, doing a norm 1 (i.e. abs value component distance difference) is as good as any heuristic way | 15:27.55 |
| rayjj | Robin_Watts: sounds reasonable. So, why not just stick with the threshold on the delta then | 15:28.15 |
| | mvrhel_laptop: thanks. | 15:29.04 |
| | mvrhel_laptop: yeah -- I wasn't going to bring up delta e :-) | 15:29.32 |
| mvrhel_laptop | :) | 15:29.39 |
| | on top of that differences in images pixel by pixel would really use a different measure compared to solid graphical objects | 15:30.45 |
| Robin_Watts | mvrhel_laptop: We're always doing distance comparisons in rgb. | 15:30.46 |
| mvrhel_laptop | if you wanted to get picky | 15:30.58 |
| sebras | tor8: henrys: pong! do you still want me to take a look tonight? do you have a URL? | 15:36.42 |
| tor8 | sebras: we've looked, but if you're curious it's nvareader.com | 15:39.11 |
| mvrhel_laptop | this is too funny | 15:40.39 |
| | https://readreidread.wordpress.com/tag/cat-prevents-house-fire/ | 15:40.40 |
| Robin_Watts | Thanks for that image. :) | 15:42.06 |
| mvrhel_laptop | hehe | 15:42.17 |
| rayjj | Robin_Watts: what UI did you have to the threshold -T # -- I notice that tor8's has a fixed threshold of 0.02f ) | 15:52.36 |
| Robin_Watts | rayjj: I didn't expose the threshold value in mudraw. | 15:54.08 |
| | It was exposed at the api level, but not in the mudraw level. | 15:54.20 |
| rayjj | Robin_Watts: I see -- so tor8 has that also in fz_new_test_device as the third param | 16:00.19 |
| Robin_Watts | yes. | 16:00.44 |
| kens2 | congratulations Robin, you'e inherited the idiot..... | 16:06.57 |
| henrys | my UK calendar says itâs a holiday | 16:07.31 |
| kens2 | Your calendar is correct | 16:07.39 |
| | late August bank holiday | 16:07.54 |
| | BTW, you're still writing weird things for apostrophes henrys | 16:08.53 |
| | The one above came out acircumflex, Euro, TM for me | 16:09.16 |
| jogux | henrys: only in England ;-) | 16:09.30 |
| kens2 | Really ? My calendar usually lists UK wide and regional holidays, and it has today as UK wide | 16:10.20 |
| mvrhel_laptop | henrys: i got the package today from company M | 16:10.23 |
| henrys | kens2: yes the irc logs show other problems also. | 16:10.25 |
| | ? | 16:10.26 |
| mvrhel_laptop | did you talk to Miles? | 16:10.36 |
| jogux | anyone know when MarcosW is next working btw? (I sent him an email) | 16:10.37 |
| | kens2: yeah. or actually just except scotland - it's england/wales/NI I believe. | 16:10.52 |
| kens2 | marcos sent an email saying he'd be back 'in about 6 hours' I think that measn about 8pm UK time | 16:10.56 |
| | He was in JFK at the time | 16:11.05 |
| jogux | kens2: oh, of course, he's been on hols hasn't he, I recall moaning about support now. ;) thanks. | 16:11.31 |
| kens2 | :-) | 16:11.38 |
| | henrys, yes the irc logs are even worse than my log here, even more weird characters | 16:13.03 |
| jogux | is that the irc log problem the mac os one? I figured out a fix for that I'm sure. | 16:14.02 |
| henrys | jogux: yes I did too, but I think I reinstalled and itâs back. Forgot what we did. | 16:14.44 |
| kens2 | jogux, it seems to be Mac specific yes | 16:14.50 |
| jogux | I think it was system preferences -> keyboard -> text tab and untick 'use smart quotes and dashes'. | 16:14.56 |
| kens2 | THat sounds about right | 16:15.07 |
| henrys | let's see what that does | 16:17.34 |
| | kens2: apostrophe okay? | 16:18.00 |
| kens2 | Yes, looks great! | 16:18.09 |
| | Much, much easier to read | 16:18.15 |
| Robin_Watts | kens2: Get a decent irc client :) | 16:18.29 |
| henrys | thanks jogux | 16:18.38 |
| jogux | robin: it breaks the logs too :-) | 16:18.39 |
| | henrys: np! | 16:18.44 |
| kens2 | Precisely :-) | 16:18.45 |
| Robin_Watts | jogux: Well, the logging scripts are a horrible hack :) | 16:19.10 |
| robbyman | Hey everyone! | 16:19.36 |
| Robin_Watts | ETOOMUCHENTHUSIASM. | 16:20.02 |
| jogux | robin: that's not a real error :P | 16:20.19 |
| | hello robbyman | 16:20.29 |
| kens2 | It shoudl be a warning at least | 16:20.33 |
| Robin_Watts | EENTHUSIASMOVERFLOW ? | 16:20.34 |
| robbyman | can I use ghostscript + imagemagick to convert illustrator (ai) to png? | 16:20.56 |
| kens2 | WHy use ImageMagick ? | 16:21.05 |
| | And the answer is 'it depends' | 16:21.13 |
| robbyman | or not! | 16:21.19 |
| | I'm trying to make a web app that converts ai, psd, or pdf to png | 16:21.48 |
| kens2 | .ai files are not PostScript, they look like PostScript, they can often be treated as PostScript, but its not guaranteed. Oh, except for the ones that look like PDF files | 16:21.49 |
| | For files which work (the majority) you just use the png device and Ghostscritp will consume the file and produce one or more ONG files. | 16:23.15 |
| | For those files which don't work, you get an error. | 16:23.30 |
| | Those can only be handled by opening in Illustrator and saving sa 'something else' | 16:23.48 |
| robbyman | kens2: I see, thank you | 16:24.03 |
| rayjj | and for psd, you are on your own (maybe imagemagick handles those, I don't know) | 16:24.07 |
| robbyman | I think imagemagick handles those | 16:24.19 |
| kens2 | I suspect ImageMagick will open at least the simple cases .f Photoshop files | 16:24.25 |
| henrys | Robin_Watts: nice 5 "As I said before"'s in one email. certainly a record ;-) | 16:24.44 |
| kens2 | Yeah, but they don't listen. | 16:24.55 |
| rayjj | robbyman: the .ai files that are problematic are those that are saved without the Adobe Illustrator "ProcSet". | 16:25.02 |
| kens2 | Wagers on how long it takes for them to ask the same questions again ? | 16:25.08 |
| kens2 | is off. night all | 16:26.23 |
| rayjj | robbyman: and if you want to get "tricky" (do some work on your own) it is possible to look at the problem .ai and having grabbed a copy of the ProcSet from a matching Illustrator file that was saved with it, load that into gs before running the .ai file | 16:26.57 |
| robbyman | kens2: gotcha. I need to make a quick prototype and see what happens.. thanks for all the help | 16:27.50 |
| rayjj | robbyman: that will probably require some reading parts of the PLRM and some experimentation. | 16:27.52 |
| robbyman | is there an alternative that might work better? | 16:27.58 |
| rayjj | nope | 16:28.02 |
| | other than Adobe Photoshop CS | 16:28.24 |
| robbyman | yeah | 16:29.25 |
| rayjj | I *think* PhotoShop CS can load Illustrator files (at least most of them). The CS products all seem to 'convert' as needed | 16:29.37 |
| robbyman | well it does open them, turns it into jpg or png | 16:29.55 |
| rayjj | robbyman: I mention that possibility of using gs with the AI ProcSet because I did something like that years ago (at a previous company) | 16:30.22 |
| robbyman | I'm trying to build something on the web that would convert ai/pdf/psd to png... pdf/psd are a breeze, but ai is tricky | 16:30.43 |
| | rayjj: appreciate the help | 16:33.20 |
| Robin_Watts | henrys: I hope they spot that :) | 16:33.33 |
| henrys | Robin_Watts: I wouldn't count on it. | 16:34.05 |
| rayjj | Robin_Watts: in order for them to spot those, they have to read, and that is doubtful | 16:48.41 |
| Robin_Watts | They started by thanking me for a detailed reply. I did wonder about asking "but did you actually read it?" | 16:50.02 |
| mvrhel_laptop | henrys: ping | 16:53.24 |
| henrys | mvrhel_laptop: yup | 16:57.18 |
| mvrhel_laptop | so I got the package from company M. did you get a chance to talk with miles? Just wanted to double check that I should forward it to Ray. I will do that today | 16:58.32 |
| henrys | mvrhel_laptop: I sent him email but did not get a response. I'll take responsibility for it. | 16:59.12 |
| mvrhel_laptop | ok. thanks. | 16:59.22 |
| | I will let ray know it is coming | 16:59.44 |
| henrys | mvrhel_laptop: thanks | 17:00.09 |
| Robin_Watts | Has anyone spoken to Miles since the earthquake? | 17:01.36 |
| | I think it was a bit closer to Miles that Scott was making out. | 17:01.51 |
| henrys | Robin_Watts: no he's out until tomorrow | 17:03.19 |
| | Robin_Watts: napa is 30 miles from miles | 17:08.06 |
| Robin_Watts | 30 miles doesn't sound a lot for a 6.1 earthquake? | 17:10.21 |
| henrys | Robin_Watts: according to the "shake map" he felt it weakly no damage | 17:27.56 |
| Robin_Watts | ah, cool. | 17:28.09 |
| jogux | that's good news. | 17:34.51 |
| mvrhel_laptop | bbiaw | 17:53.58 |
| Robin_Watts | heh: Big earthquake in Napa this morning. Locals described it as "vibrant and intense, but not overly flamboyant with a clean, earthy finish." | 18:07.00 |
| | oops: https://twitter.com/DavidSilverOak/status/503514636290105344/photo/1 | 18:07.48 |
| henrys | Robin_Watts: I guess we'll be seeing a lot of Napa blends soon | 19:31.36 |
| Robin_Watts | Has anyone else managed to enable their Office for Mac ? | 19:32.06 |
| | oh, works now. Stupid installer. | 19:33.48 |
| | Forward 1 day (to 2014/08/26)>>> | |
IRC Logs