| <<<Back 1 day (to 2015/09/09) | 20150910 |
srini | Mudraw html does not recognize decorations like underlines...? | 06:43.11 |
kens | chrisl, you having any network problems ths morning ? | 06:54.17 |
| I'm wondering if my ISP's DNS is shafted | 06:54.32 |
chrisl | kens: I haven't so far, but then I just sat down | 06:56.10 |
| Nope, everything seems fine | 06:56.56 |
kens2 | Weird | 06:57.02 |
| Shifting to the Google DNS server helps, in that I can now resolve the IP addresswes, but I can't ping a boat load of sites | 06:57.34 |
| Eghttp://www.bbc.co.uk is OK< buthttp://www.dilbert.com isn't | 06:58.08 |
| It does seem to be progressively improving, I suspect my ISP has a problem | 06:58.44 |
chrisl | dilbert.com works fine for me - actually, everything seems quite quick this morning | 06:59.07 |
kens2 | It is fine for me too, now..... | 06:59.28 |
| Oh it seems dilbert just doesn't respond to pings | 07:00.06 |
| I am having to use the Google DNS though | 07:00.26 |
chrisl | Well, lots of sites ignore pings these days | 07:00.50 |
kens2 | yeah, but I know it used to be OK, which is why I'm using it, they must have changed | 07:01.10 |
chrisl | Yeh, it's not responding to ping for me either. | 07:02.18 |
kens2 | Not to worry, now I switched DNS everything is working. I've had to do that before with my ISPs DNS | 07:02.46 |
srini | What is the best time to catch up with mupdf developers? | 07:11.23 |
kens2 | an hour or two from now | 07:11.33 |
srini | kens2, Thank you! | 07:11.47 |
tor8 | Robin_Watts: so you're finally doing that forwarding device? I remember running into some trouble doing that before. | 08:02.58 |
| the way paths and a few other things require back-and-forth communication | 08:03.20 |
| s/paths/patterns/ | 08:03.26 |
| but I guess enough time might have passed for those issues to have been fixed in the passing | 08:04.55 |
Robin_Watts | tor8: they do? | 08:34.07 |
tor8 | to be fair I was trying a 'tee' device, to feed two devices from the same source | 08:34.34 |
| that didn't fare well with the pattern device call having a return code and that the interpreter can change behaviour based on | 08:35.01 |
Robin_Watts | tor8: Ah, yes. | 08:35.10 |
tor8 | a straight up forwarding/decorating device shouldn't have the same troubles | 08:35.33 |
| when I get some time over I would like to expose the device and interpreter interfaces to javascript, so we could script this kind of stuff without having to recompile | 08:36.17 |
| a "mutool run script.js" | 08:36.35 |
Robin_Watts | tor8: Yeah, I have some prototype code somewhere that exposes the device interface into android java. | 08:43.51 |
tor8 | Robin_Watts: is there a quick way to get the bounding box for a display list? | 08:44.15 |
| I'm thinking of adding two fz_rects to the display list struct | 08:44.33 |
| one for the mediabox passed to the fz_begin_page call and another for the actual inked area | 08:44.43 |
| as would be calculated by the bounding box device | 08:44.59 |
Robin_Watts | What if you have more than one beginpage? | 08:45.03 |
tor8 | you shouldn't, ever | 08:45.12 |
Robin_Watts | Suppose you open a display device, and then fz_run_contents, then fz_run_annotations. | 08:45.41 |
tor8 | but yes, that is a fair question. I guess we could take the union. | 08:45.43 |
| fz_run_annotations wouldn't call beginpage | 08:45.51 |
Robin_Watts | So fz_end_page is not the last thing in a display list? | 08:46.07 |
tor8 | only for actual pages | 08:46.41 |
Robin_Watts | Or suppose you want to 'merge' 2 pages, by running 2 pages into the same device with different CTMs ? | 08:46.55 |
tor8 | fz_run_page wraps the page contents and the annots with begin/end page calls | 08:47.06 |
| that's a better scenario | 08:47.30 |
Robin_Watts | tor8: Right, so fz_run_page gives different results from running the contents and annots separately. | 08:47.34 |
tor8 | and in that case, taking the union is the right thing to do, but I haven't really considered how to handle that case on the backend side with the pdfwrite device etc | 08:48.12 |
Robin_Watts | tor8: I agree that taking the union would be the right thing to do. | 08:48.38 |
| I'm disquieted by the difference in behaviour between fz_run_page and fz_run_contents/annots. | 08:49.06 |
| Holding a couple of top level rects doesn't offend me massively. | 08:49.38 |
tor8 | if you're calling fz_run_page_contents, I expect you to call fz_begin_page manually as part of the contract for doing things the low level way | 08:49.41 |
Robin_Watts | I worry that computing the second one might slow down display list collection. | 08:50.00 |
tor8 | I'm looking through the code and wondering what the different rects that the display list collection calculates are | 08:50.30 |
| if it might be easy to just stash them away at an opportune spot into a high level rect | 08:50.58 |
Robin_Watts | pdfapp.c does not call fz_begin_page, end_page then. | 08:51.35 |
tor8 | Robin_Watts: on tor/master is the main reason for this ... a few high level functions to wrap the most common ways of creating and using devices | 08:51.46 |
| Robin_Watts: and in adding a few more to that I ran into the issue with the bounding box missing from the display list. | 08:52.33 |
| I want to call fz_new_display_list_from_page() and then fz_new_pixmap_from_display_list() | 08:52.53 |
| without having to query the page mediabox manually, so the display list needs a way to bound itself for the latter call | 08:53.14 |
Robin_Watts | Nor does platform/android/jni/mupdf.c | 08:53.18 |
tor8 | preferably using the mediabox | 08:53.19 |
| those bits of code predate the addition of the begin_page/end_page calls | 08:53.39 |
| it might be that we should call begin_page/end_page for the run_page_contents call | 08:54.02 |
Robin_Watts | right, but it's an area that should be cleaned up. Not blaming anyone. | 08:54.03 |
tor8 | Robin_Watts: agreed. | 08:54.07 |
Robin_Watts | I agree that we'd like a way to get the media box out of the displaylist. | 08:55.07 |
| A top level rect seems reasonable. | 08:55.26 |
| which is the union of all the begin_pages in the list. | 08:55.40 |
tor8 | but if we're going to use the beginpage/endpage calls in high level output devices then getting the endpage before the annotations is going to be awkward | 08:55.49 |
| yes, that sounds ideal | 08:55.59 |
Robin_Watts | It's the other calculation (the inked area) which may be trickier to achieve without doing excess work. | 08:56.20 |
tor8 | but also means that if I make separate display lists for annotations, which bounding box do I use for those? | 08:56.34 |
Robin_Watts | I'd be tempted to just use fz_bound_page for that one, and let it work the usual way. | 08:56.40 |
tor8 | could be that each annotation has its own bounding box that I could feed into the begin_page calls when creating the display list for them | 08:57.07 |
Robin_Watts | Could we add a flag to begin_page where we can say if it's a page or an annotation? | 08:57.30 |
| Actually... do we ever want annotations to change the media size of the page ? | 08:57.54 |
tor8 | we don't | 08:58.16 |
| so going to be a bit of a hack | 08:58.42 |
Robin_Watts | So, there are 3 bboxes we need. | 08:58.52 |
| 1) the mediabox (the union of the page boxes) | 08:59.03 |
tor8 | I need the page mediabox and the annotation for rendering annotations to temporary bitmaps | 08:59.32 |
Robin_Watts | 2) the inked region (the union of all the marking operations for a page intersected with 1)) | 08:59.40 |
| 3) the bbox that completely covers both the page and the annotations. | 08:59.59 |
tor8 | I'm planning to do several separate bitmaps and then compose the final screen from those rather than re-rendering the page + annotations each update from display lists | 09:00.06 |
Robin_Watts | tor8: Right. | 09:00.32 |
tor8 | I don't think we need 3) | 09:00.37 |
Robin_Watts | Well, we might be able to live without 3) if we can fz_bound_annot. | 09:00.47 |
tor8 | ah, we do have a fz_bound_annot function | 09:01.03 |
Robin_Watts | Maybe hold separate displaylists for each annotation? | 09:01.13 |
tor8 | and I can live with just 1) for now | 09:01.16 |
| I was going to hold separate display lists for each annotation and the page | 09:01.25 |
| so I should just need to make sure to begin/endpage each display list with its appropriate bbox | 09:01.39 |
Robin_Watts | So each annot displaylist will have no 'mediabox', as it won't have any begin_page/end_page. | 09:01.49 |
tor8 | being the mediabox or annotation box | 09:01.50 |
| or as you said, add a flag to begin_page | 09:02.18 |
Robin_Watts | Either annotations have begin/end_page, or they don't. | 09:02.31 |
| If they do have it, then we can't get a proper mediabox from a displaylist generated by run_page. | 09:02.57 |
tor8 | how about making sure the devices can cope with nested begin/endpage calls? | 09:03.04 |
Robin_Watts | Not sure that helps. | 09:03.25 |
tor8 | take the union of the top level ones and intersect child beginpage boxes | 09:04.00 |
| we could pass a 'mediabox' to fz_new_display_list... | 09:04.49 |
Robin_Watts | I still can't see how that nesting scheme solves this situation, and it would require changes to the bbox device to properly clip, or it'd be nonsensical. | 09:05.33 |
tor8 | but I rather like the begin_page / end_page calls as they are | 09:05.54 |
Robin_Watts | tor8: But we don't necessarily know the mediabox at display list creation time. | 09:06.02 |
tor8 | yeah, ignore the nesting thing, that's just adding bad complexity | 09:06.12 |
| actually, I think we do | 09:06.29 |
Robin_Watts | I am tempted to say leave begin_page as it is (except properly insert it on run_page_contents). | 09:06.51 |
| We can keep the media box generated by that. | 09:07.01 |
tor8 | but that was one of the reasons for adding the begin_page/end_page calls | 09:07.02 |
Robin_Watts | We should either add a begin/end_annotation call (or add a flag to begin_page). | 09:08.17 |
| annotations don't include a mediabox. | 09:08.27 |
| Actually, let me start again. | 09:09.23 |
| Add a flag to begin_page to say if it's a page or an annotation. | 09:09.39 |
| We keep 3 rects in the displaylist. | 09:09.48 |
| 1) The mediabox of the pages. | 09:09.52 |
| 2) the mediabox of the annotations | 09:09.58 |
| 3) the union of the inked regions of the page | 09:10.43 |
| That gives us everything we could possibly need I think. | 09:11.06 |
| and every call is clearly defined. | 09:11.19 |
tor8 | Robin_Watts: okay. how do you envision the call nesting for the begin_page(is_annot) calls? | 09:12.55 |
Robin_Watts | tor8: every fz_run_annot starts with a begin annot and ends with an end_annot. | 09:19.03 |
| every fz_run_contents starts with a begin_page and ends with an end_page | 09:19.25 |
| So it's simple to see from the display list where each thing came from. | 09:19.38 |
tor8 | so basically: begin_page, page contents, end_page, begin_annot, ..., end_annot, begin_annot ... end_annot | 09:20.13 |
| I was otherwise thinknig begin_page ... begin_annot ... end_annot begin_annot ... end_annot end_page | 09:20.52 |
| Robin_Watts: on tor/master, a basic first step that just records page bounding boxes and doesn't attempt to cope with per-annotation separate display lists | 09:23.32 |
Robin_Watts | tor8: I prefer the former to the latter. | 09:24.27 |
tor8 | hm, maybe we shouldn't be passing the ctm to fz_bound_display_list | 09:24.48 |
| Robin_Watts: actually, now I remember another reason why I added the calls | 09:25.04 |
Robin_Watts | the latter makes it impossible to do fz_run_page_contents then fz_run_annot. | 09:25.06 |
tor8 | so that we can use them in the pdfwrite device so you can get natural page delineations without having to recreate the device for each page | 09:25.41 |
| which makes a lie of my earlier statement that I only expect to see one begin_page call | 09:26.04 |
| and means that we really ought not to call begin_page automatically in run_page_contents, since that would foul up any n-up style devices | 09:26.37 |
Robin_Watts | tor8: ok... just mulling that. | 09:29.25 |
| tor8: No... I think I disagree. | 09:30.14 |
| when we call fz_run_page, that must call fz_begin_page, ...contents ..., fz_end_page. | 09:30.33 |
| If I run that to a display list and then run the display list out again, I want to get the exact same sequence of events. | 09:31.03 |
tor8 | fz_run_page draws the whole page and wraps it with begin_page/end_page and that's fine | 09:31.18 |
Robin_Watts | If we have an n-up device, then that should swallow the begin_page/end_pages, and regenerate a new one. | 09:31.39 |
tor8 | but when calling run_page_contents, I think we should make the begin_page/end_page wrapping the clients responsibility | 09:31.41 |
Robin_Watts | ok, I can't immediately see an objection to that. | 09:32.17 |
tor8 | run_page_contents is for the guts of one page without the begin/end_page package | 09:32.24 |
| and n-up can be done in two ways here I think | 09:32.51 |
| 1) at the fz_document level, by calling run_page_contents with various matrices | 09:33.07 |
| 2) at the fz_device level by eating begin_page/end_page calls and decorating and forwarding calls | 09:33.39 |
kens | ah drat, I just created a branch on master I didn't mean to | 09:44.39 |
| tor8 or Robin_Watts would one of you mind killing the branch dynamic_PJL_allocs please ? | 09:45.34 |
Robin_Watts | kens: on golden? | 09:47.25 |
kens | yes pleae | 09:47.33 |
Robin_Watts | git push golden :dynamic_PJL_allocs | 09:47.37 |
srini | I need help in mupdf text extraction to html | 09:47.45 |
kens | Having screwed it up I don't thnk I trust myself to undo it | 09:47.45 |
srini | is there a way to get the text-decorations? Underline StrikeThrough etc.... and also is there any patch on h1 h2 h3 detection? | 09:48.24 |
Robin_Watts | done. | 09:48.25 |
kens | thanks Robin_Watts | 09:48.30 |
srini | font-color also | 09:48.41 |
Robin_Watts | srini: The text decorations aren't defined in PDF. | 09:48.44 |
srini | Robin_Watts, so we have not way to detect underlined text? | 09:49.11 |
| *no way | 09:49.32 |
kens | You could detect the presence of a thin line parallel to, and in close proximity to, a glyph | 09:50.10 |
| Which 'might' be an underlined piece of text, or it might be the top of the frame for a figure right underneath it | 09:50.39 |
| So you would have to further test that the path extended from the beginning of the text to the end of the text, and no further | 09:51.14 |
srini | How about h1 h2 h3 detections | 09:52.10 |
kens | No idea what you are talking about there | 09:52.22 |
Robin_Watts | srini: In the PDF file, all we get is 'put this glyph in this font, at this position'. | 09:52.30 |
| And things like underlines/strikeouts are added on later using line art. | 09:52.45 |
| There is no concept of 'structure'. | 09:52.55 |
| Hence we have no h1/h2/h3 information. | 09:53.03 |
srini | Hmmm ... got it... | 09:53.09 |
Robin_Watts | We don't even have paragraph information. | 09:53.11 |
| Now, when we do text extraction we do some funky heuristics that try to reassembly chars to words, words to lines, lines to paragraphs, etc. | 09:53.40 |
| but those are all prone to failure. | 09:53.47 |
kens | chrisl, looks like I was suffering from ths problem ths morning: | 09:53.51 |
| http://www.theregister.co.uk/2015/09/10/major_plusnet_outage_dns_goes_titsup/ | 09:53.53 |
srini | I actually tried a way sometime.... find most used font and size, considering this a base font size, continue searching fonts greater than base font size and mark them into h1 h2 h3 etc | 09:54.15 |
| but did not actually work well that way | 09:54.27 |
Robin_Watts | srini: That's exactly the kind of think you'd need to do. | 09:54.44 |
| your best bet is to operate on the structures you get back from the fz_stext_device. | 09:55.12 |
kens | All these things are, in the absence of dtaa, heuristic. If your heuristic doens't work well, then you need to refine it | 09:55.13 |
Robin_Watts | s/think/thing/ | 09:55.28 |
srini | Any official/unofficial patch available for text decorations? | 09:56.49 |
| And, how about font-color? | 09:57.17 |
Robin_Watts | nope. | 09:57.20 |
kens | I thought the current color was in the XML output (I coudl be wrong) | 09:57.41 |
Robin_Watts | kens: Not in the text extraction output, I thought. | 09:58.04 |
kens | Ah.... | 09:58.11 |
| Its been a while (ie more than 10 minutes) and therefore I cannot recall for certain | 09:58.34 |
Robin_Watts | oh, wait, you may be right. | 09:58.50 |
| the color might be in the style. | 09:58.56 |
srini | Robin_Watts, kens - many things are not available in the text extraction... they are there in the library | 09:59.01 |
Robin_Watts | no, the text color is passed to fz_lookup_text_style, but is not actually used. | 09:59.50 |
| You could extend that fairly simply if you wanted. | 09:59.58 |
kens | thnks PlusNet is havig a *very* bad day, I'm glad I'm not on their Hell Desk | 10:01.26 |
srini | I am not a seasoned hand on coding... decoration especially underline... can someone help me add code? | 10:03.33 |
Robin_Watts | srini: Forget about underline. That ain't gonna happen easily. | 10:04.22 |
kens | likewise strikethrough I'd have thought | 10:04.35 |
Robin_Watts | same deal, yes. | 10:04.41 |
| color should be a simple addition for anyone that knows C. | 10:04.53 |
kens | that one might be easier to detect, but its till a project | 10:04.54 |
Robin_Watts | kens: strikethough and underline are both basically the same problem (dot products with the baseline and examine the offset) | 10:05.47 |
kens | http://www.theregister.co.uk/2015/09/10/major_plusnet_outage_dns_goes_titsup/Actually, I wonder if the underline and strikethrough are done as part of the font definition, in which case they are inherently undetectable in PDF anyway | 10:05.54 |
| Robin_Watts : that assumes that teh work isn't done by definiing a type 3 font which first draws then underlying font glyph, then draws a line | 10:06.25 |
Robin_Watts | kens: true. | 10:06.38 |
kens | I'd be fairly certain that's how QAuark Xpress would do it, but I doubt if MS Office does (for insgtance), but I muight be wrong | 10:06.55 |
| There's a bunch of werid stuff in the MS PS output that I've never needed to look at | 10:07.23 |
srini | mostly I come across text merging issues.... they are generally because of very small space <less than general tab space> at the beginning of the paragraphs... | 10:25.46 |
| can we have a method to alter them from the command line... | 10:26.19 |
hj__ | is there a difference in handling file name between 9.16 and 9.07. we have a input file name using chinese characters and ghostscript 9.16 gives an open error | 13:27.14 |
kens | You probably need to use the latest code. THere may be a work-around, I don;t recall | 13:27.55 |
hj__ | in our debug it looks that "os_status" is called with the UTF8 name passed to stat. are there no specific os_status/os_rename etc for windows doing the wide char version | 13:31.07 |
| when converting the utf8 name in os_status to wchar and calling _wstat instead of stat it seems to work | 13:34.18 |
| a quick look in the code...the os_rename/os_status/os_delete should have a windows specific implementation | 13:35.14 |
chrisl | hj__: as kens said, that's been fixed, a few months ago | 13:40.35 |
hj__ | thanks. | 13:41.37 |
chrisl | hj__: here's the commit: http://git.ghostscript.com/?p=ghostpdl.git;a=commitdiff;h=1572afcd | 13:43.00 |
hj__ | chirshl thanks for the stat commit. do you know if the rename/remove also have got a similar change? | 13:46.52 |
| i just looked in the git source. the os_delete and os_rename just call the default c functions. this might need some changes for windows wide char | 13:51.54 |
kens | If you think you;'ve found a bug, feel free to open a bug report. Don;t forget to include a sample file and instrcutions for reproducing the bug. | 13:53.57 |
hj__ | oke | 13:54.20 |
henrys | kens: begrundginly I'm good with it, it was intended never to fail, only doing allocations at startup; the job manager should alway be able to run on a pcl printer, but I realize that's too inflexible. | 14:02.31 |
kens | henrys I could go with that if it wasn't for the fact that we don't action the PJL immediately but save it all up. TO do that, I need a bigger (potentially much bigger) allocation | 14:04.50 |
| By the way, it won't fail if it can't allocate a new string or tbale, it just keeps on using the old one | 14:05.36 |
| If it only runs at startup, a failure to allocate one of these small strings woud be only one step ahead of the whole interpreter faolding up anyway I'd have thought | 14:06.20 |
| Anyway, I'll move ahead to the distillerparams | 14:07.16 |
henrys | kens: sounds good. | 14:08.21 |
kens | Well, ths will be the more interesting part :) | 14:08.48 |
henrys | kens: where are the device parameter setting going to take place exactly as I'm rewriting much of plmain and friends. | 14:09.00 |
| ? | 14:09.09 |
kens | It'll be simple strings | 14:09.22 |
| I have yet to decide on delimiters for arrays and dicts, probably I'll just stick with the PS ones | 14:09.57 |
| henrys, what do you see with C705.bin and -dFirstPage=2 -dLastPage=4 ? | 14:10.28 |
| D'oh I clearly cannot read | 14:10.42 |
| I will plan to process the dsitillerparams at the same time as all the other devcie stuff | 14:11.05 |
| SO all the params will get assembled into a big set ofstorage, then we'll call some magic procedure which will turn it all into param lists and call putdeviceparams | 14:11.37 |
| IIRC this is pretty much what happens for the existing settings ? | 14:12.02 |
henrys | kens: I haven't looked at C705.bin but it can't work, it installs XL's device in pcl's state the page count is gone. | 14:14.56 |
kens | THe page count is held by the First/Last page device | 14:15.25 |
| Its independent of the interpreter | 14:15.31 |
| Err, I think..... | 14:15.50 |
| It does work on WIndows | 14:16.02 |
henrys | XL and PCL have their own device and their own subclass. | 14:16.28 |
kens | Well each should operate independently, assuming you pass hte device params to both | 14:17.09 |
| Like I said, I did try ths on Windows and it seemd to do what I expected | 14:17.36 |
henrys | oh maybe I should look again. | 14:18.52 |
kens | It *ddn't* work on Linux, but it didn't work wihtout -dFirstPage either | 14:19.12 |
| I have no clue why | 14:19.28 |
henrys | I will look at it, it was "a note to self" type bug. It can wait. | 14:21.52 |
kens | No problem, I'll leave it with you | 14:22.30 |
mvrhel_laptop | tor8: is there any particular reason that fz_keep_page does not have a prototype in document.h? | 16:48.14 |
| I was maybe going to make use of it in my page cache | 16:49.04 |
| or Robin_Watts if tor8 is not around... | 16:50.11 |
Robin_Watts | mvrhel_laptop: Sounds like an omission to me. | 16:50.37 |
mvrhel_laptop | ok | 16:50.52 |
| I will add it | 16:51.47 |
| Robin_Watts: the tiny minor commit is on my mupdf repos if you can take a 2 sec look | 17:00.07 |
Robin_Watts | looking. | 17:00.30 |
| Can't argue with that. | 17:01.01 |
mvrhel_laptop | I hope not :) | 17:01.10 |
| what day do you fly in to chicago? | 17:01.24 |
| Robin_Watts: are you guys coming early for the show? | 17:02.22 |
Robin_Watts | mvrhel_laptop: No, flying on wednesday. | 17:03.14 |
| We have to stay til sunday night, so that's quite long enough away from home (or so I'm told). | 17:03.35 |
| Who will deal with the spiders in my absence? :) | 17:03.44 |
mvrhel_laptop | Robin_Watts: :) | 17:03.50 |
Robin_Watts | mvrhel_laptop: I have tea. | 17:04.05 |
mvrhel_laptop | Robin_Watts: I may try to meet up with you all for dinner on Wed. what time do you arrive? | 17:04.05 |
| oh great. do you want anything like dark chocolate malt balls, or dark chocolate pb cups? | 17:04.28 |
Robin_Watts | We land at 12:30 | 17:04.30 |
mvrhel_laptop | oh early | 17:04.33 |
Robin_Watts | mvrhel_laptop: Definitely not! | 17:04.39 |
mvrhel_laptop | I need to bring something | 17:04.45 |
Robin_Watts | Still working off the holiday fat :( | 17:04.52 |
| Helen loved the last ones but said "Tell Michael never to bring these again!" | 17:05.09 |
mvrhel_laptop | they are my favorite | 17:05.26 |
rayjj | tor8: Scott is on the phone and a potential customer wants to know if we support EPUB 4 OS | 17:11.08 |
Robin_Watts | wtf is EPUB 4 OS ? | 17:11.23 |
mvrhel_laptop | it the one after 3 | 17:11.53 |
Robin_Watts | MuPDF supports epub v1. | 17:12.06 |
| v2 is 'fixed format' rather than 'reflowable' so moving towards pdf. We don't support that properly. | 17:12.32 |
rayjj | that's what I didn't know either. As far as I knew their is a v 3 (that we can't handle) | 17:12.58 |
Robin_Watts | hmm. Maybe we support v2, but not v3. | 17:13.54 |
| Certainly I've never heard of a v4. | 17:14.00 |
| Googling "EPUB 4 OS" is unhelpful. | 17:14.13 |
| I suspect the customer is parroting something requested by their engineering team, and filtered through scott it's arriving corrupted. | 17:14.53 |
| No idea where the corruption might be occurring. | 17:15.02 |
rayjj | Robin_Watts: it's a .jp company, so that doesn't help | 17:15.51 |
Robin_Watts | Get scott to ask them to forward him some links about what they want supported. | 17:16.21 |
| and we can look at those. | 17:16.30 |
rayjj | It might be a question about which OS's | 17:19.16 |
henrys | I hope our imiplementation is compatible with epub 2, that's what we are saying to the world | 17:19.50 |
Robin_Watts | and a link about the OS they want supported would make that clear. | 17:19.53 |
rayjj | Scott is asking them. I had him list that we support Windows, unix/linux, Mac OS/x, iOS and Android in case that answers their question | 17:20.31 |
Robin_Watts | henrys: Right. I think I was wrong earlier. We have a v2 implementation, but we don't support the new stuff in v3. | 17:20.32 |
henrys | yeah what we've said is 2 compatible and we'd add 3 features as time permits if they are sane | 17:21.25 |
rayjj | Robin_Watts: that's what I recalled as well. But I also hadn't heard about v4. V3 was just released 2014, so it's a little soon for a v4 :-) | 17:21.25 |
| henrys: Robin_Watts: thanks. That's what I told Scott. v2, not v3, but if they need v3 we'd have to discuss it (and get sample files) | 17:22.29 |
henrys | sounds fine rayjj | 17:22.46 |
nvs | Hi! Are there any mupdf guys here? | 17:33.02 |
mvrhel_laptop | bbiaw | 17:46.32 |
| Forward 1 day (to 2015/09/11)>>> | |