| <<<Back 1 day (to 2015/03/24) | 20150325 |
Robin_Watts | According to the cluster timings, the name rework saved 25%. | 00:19.02 |
| From 4:04 to 3:01 | 00:19.24 |
| Latest run has us at 2:57 | 00:20.14 |
mvrhel_laptop | Robin_Watts: nice! | 00:55.53 |
KunMyt | jogux: thank for reply :D | 01:20.36 |
| jogux: but when i call MuPDF activity i has though intent.setDataAndType(uri, "application/pdf"), so "application/pdf" different "PDF" | 01:24.26 |
| i got slove problem, when i read PDF by file when i convert it to byte[], the MuPDF.Core is still i read pdf with file. so wasOpenedFromBuffer= true... | 01:49.17 |
rayjj | Robin_Watts: how trustworthy are cluster timings ? | 01:51.06 |
| even on a single machine I usually have to run something several times and take the lowest, but on the cluster, you don't know which job will go to which machine and they are VERY different | 01:52.09 |
| darn, I started my job after marcos took his machines down, so it's taking longer :-( | 01:59.11 |
chrisl | kens: this tesseract font thing: wouldn't it be sufficient to have a font with a single glyph (notdef)? | 08:46.39 |
kens | chrisl I don't think so, because we'd need a cmap and so on. | 08:47.12 |
| We can't access a glyph in the font if it isn't present. | 08:47.24 |
| Now, you could have a CMAP which referenced ony the notdef glyuph | 08:47.38 |
| But that's not what they have | 08:47.45 |
| I suspect I know what they are trying to do, and I could probably come up with a minimalist font,but I don't think they know enough to do it themselves | 08:48.38 |
chrisl | If you build a font with only one glyph, the cmap and other tables ought to be correct for just that. As none of their glyphs actually display anything, I don't see why you couldn't just have every character code map to notdef | 08:48.56 |
kens | Yes, that's what I was suggesting, but its not what they do | 08:49.09 |
| At least one of their fonts has all the glyphs being 0 length for instance | 08:49.27 |
chrisl | So what I'm saying is, maybe that would be easier than trying to "fix" their current approach | 08:49.44 |
kens | Yes it probably would be. To be honest, there isn't a huge problem with their current approach (except that I don't see any way it can cover 'the entire multilingual plane'), except for the fact that the font only includes glyphs 0-116, but the PDF file references glyph 117 | 08:50.50 |
Robin_Watts | tor8: mujstest is screwed :( | 10:36.59 |
tor8 | how? | 10:46.27 |
Robin_Watts | cluster tests failing all over the place. | 10:47.45 |
| looks like memory corruption. | 10:47.52 |
| Looking into it now. | 10:47.55 |
| I forgot the INT_MAX/SIZE_MAX thing. | 10:48.04 |
tor8 | Robin_Watts: you didn't push the largefile thing yet did you? | 10:52.10 |
Robin_Watts | I did not. | 10:52.28 |
| Oh, that was where the UINT stuff was? Phew. | 10:52.43 |
tor8 | yeah :) | 10:52.57 |
Robin_Watts | I'm getting corruption of a freed block. | 10:56.50 |
| We're cocking up the refcounting on an old style PDF_Name | 11:01.25 |
| tor8: OK, think I found it. Silly mistake on my part. testing now. commit on robin/master | 11:57.00 |
| going for a run. | 11:57.14 |
| balls. I fear there are more problems lurking in mujstest. | 12:04.12 |
| paulgardiner, tor8: if you could review that fix, I can get it in and then look for the other problems when I get back. | 12:04.49 |
paulgardiner | Robin_Watts: I can take a look. MuPDF? | 12:05.24 |
tor8 | Robin_Watts: did you push your fix? | 12:47.24 |
| Robin_Watts: I don't like the dozens of (int) typecasts in the LARGEFILE commit | 13:14.30 |
| I'm going to split the LARGEFILE commit into two parts -- one introducing size_t and off_t, and the other introducing large file support | 13:14.54 |
kens | One for the MuPDF'ers: | 13:25.44 |
| http://stackoverflow.com/questions/29254240/create-image-annotation-of-pdf-using-mupdf-on-windows-8-1 | 13:25.44 |
| Possibly most relevant to paulgardiner | 13:25.57 |
paulgardiner | looking | 13:27.36 |
Robin_Watts | tor8: I have not pushed the fix. waiting for a review. | 13:38.26 |
| tor8: as for splitting it - go fior it. | 13:38.54 |
tor8 | Robin_Watts: I can't see any new fixes on robin/master | 13:40.51 |
| Robin_Watts: but I gotta go, dentist appointment. won't be back for a couple of hours. | 13:41.13 |
Robin_Watts | Fix reference counting mistake in forms support | 13:41.14 |
| sorry, my bad. pushed now. | 13:41.41 |
| will talk to paul about it. | 13:41.46 |
tor8 | Robin_Watts: looks like a small enough fix :) | 13:42.18 |
| but now I really gotta go | 13:42.21 |
Robin_Watts | i understand. | 13:42.30 |
paulgardiner | Robin_Watts: yeah, LGTM | 13:45.21 |
Robin_Watts | ta. | 13:45.26 |
| kens: I suspect the root cause for to that stackoverflow problem is that the pdf-device doesn't do images yet, does it? | 13:48.21 |
kens | Robin_Watts : I don't know enough about MuPDF to tell. | 13:50.38 |
Robin_Watts | kens, sorry, that started out as a statement to you, then I lost confidence, and it turned into a question to paulgardiner :) | 13:51.44 |
kens | :-) | 13:52.19 |
paulgardiner | Robin_Watts: yeah, I'm thinking the same. | 13:52.33 |
Robin_Watts | actually, we do do images. | 13:52.38 |
paulgardiner | Also I'm not sure a "Square" annotation should be used to display images. | 13:52.59 |
kens | I wasn't aware there was an 'image' annotation type | 13:53.21 |
| In fact there doesn't appear to be an 'image' annotation type at all | 13:54.16 |
| I don't see any annotation type which looks like it could be used for an image. Am I mistaken ? | 13:55.32 |
| OK it looks lik eyou could do an image as a Watermark, and put the image in the Appearance. | 13:57.00 |
paulgardiner | I remember (or misremember) there was some way to handle images that came up as part of forms. Maybe there's a flavour of widget annotation | 13:57.53 |
kens | I wondered about widgets, but those all seem to be form fields and other interactive elements | 13:58.14 |
| I suppose a widget with no action (or actions) and no highlight might do it, if you put the image in the appearance. | 13:59.53 |
| Its kind of icky but it might work | 14:00.19 |
| A stamp annotation would seem possible too | 14:01.53 |
| Oh, no, that has an associated popup | 14:02.18 |
paulgardiner | Using a square should work for most readers... just so long as they don't recreate the appearance stream | 14:02.51 |
kens | paulgardiner : Acrfobat *always* ignores the appearance and creates a new one. | 14:03.17 |
paulgardiner | Ah! | 14:03.32 |
kens | Thre may be some widgets for which it doesn't (or can't) do that, but I've proved in the past it does so for many types | 14:04.14 |
paulgardiner | Maybe what I have a weak recollection of was part of the JS API. I remember something to do with adding images that looked difficult to implement | 14:05.05 |
kens | No idea, I've never had ot look at JS | 14:05.25 |
Robin_Watts | Adding images to the doc stream itself is now possible, I think. | 14:08.00 |
kens | I suspect that's 'probably' what the poster really wants | 14:08.19 |
paulgardiner | Yeah, that sounds most sensible. | 14:10.25 |
| What does Acrobat to with Stamp annotations? Does it have predefined graphics for "Approved" "Experimental" etc.? | 14:12.48 |
kens | Yes I believe it does | 14:12.58 |
| For al the standard names mentioned in the PDF reference | 14:13.10 |
paulgardiner | If adding images to the doc stream is the way to go, I'll leave answering it to someone else. | 14:13.18 |
Robin_Watts | tor8: If you're back from the dentist... 2 commits on robin/master | 15:41.58 |
| crap. mutool is broken too. | 15:53.47 |
| by which I mean "I've broken mutool". | 15:54.03 |
| 3 fixes on robin/master | 16:28.46 |
| balls, ignore the last fix for now. | 16:33.07 |
| ok, 4 fixes on robin/master. | 16:39.34 |
henrys | Robin_Watts: is all this before the release? yikes | 16:52.53 |
Robin_Watts | henrys: Yeah. | 16:53.09 |
| That's why tor suggested leaving having the rc around for a few weeks. | 16:53.42 |
henrys | right makes sense should I tell Miles May? | 16:54.57 |
Robin_Watts | henrys: that would be good. | 16:55.21 |
| release before tor disappears on sabatical. | 16:55.39 |
henrys | will do | 16:55.42 |
| did you tell him I'm asking Miles to have the next meeting in Oslo? | 16:56.57 |
kens | June meeting ? WOuldnt' Copenhagen be better for Tor ? | 16:58.20 |
henrys | kens: the joke is tor is missing the next meeting because of his sabatical | 16:58.57 |
kens | THoughI think that would be bad for the US travellers | 16:59.02 |
kens | thinks Tor wouldn't care if it was Oslo :-) | 16:59.30 |
Robin_Watts | 5 fixes. | 17:44.23 |
tor8 | kens: Oslo is only 6 hours away by train+bus | 18:13.08 |
| kens: about the same distance as to stockholm :) | 18:13.59 |
henrys | tor8: well we'll have it in Lundh I had no idea it was that far out ;-) | 18:14.56 |
tor8 | Robin_Watts: 6 commits LGTM, though I'd probably rename outline->open to outline->is_open | 18:16.18 |
| henrys: the hotels in St. John's have great WiFi ;) | 18:16.59 |
Robin_Watts | tor8: is_open it shall be. | 18:21.40 |
| Only Air Canada flies to St Johns from LHR. 5 times a week. | 18:26.44 |
| 1 more fix on robin/master | 19:06.36 |
| another fix. | 19:38.01 |
tor8 | Robin_Watts: I think I can test the MINGW changes tomorrow. I'm too tired to do it now. | 19:57.29 |
| The incremental save fix LGTM | 19:57.46 |
rayjj | finally have cleaned up the problems with writing large patterns to band_ranges rather than always just to 'all_bands'. There were several places, including the pseudo-band logic that was not coping well, using band_range_max when it shouldn't | 20:06.48 |
Robin_Watts | tor8: Thanks. | 20:21.45 |
Tgiboi | what is that ? | 20:52.14 |
| Forward 1 day (to 2015/03/26)>>> | |