| <<<Back 1 day (to 2014/01/01) | 2014/01/02 |
kens | Robin_Watts : ping | 10:36.04 |
Robin_Watts | Morning kens | 11:21.08 |
kens | Hi RObin | 11:21.21 |
| Got a Git question | 11:21.28 |
Robin_Watts | Maybe henry is suggesting some way for high level streams to be passed through pdfwrite unchanged? | 11:21.32 |
kens | can I reset my local repo, and get rid of untrqacked files at teh same tim ? | 11:21.44 |
Robin_Watts | so images don't need to be recompressed or stuff like that? | 11:21.48 |
| kens: git clean ? | 11:21.55 |
kens | aha, that sounds right thanks | 11:22.02 |
Robin_Watts | You might need a -n or a -f or something. I never use git clean :( | 11:22.18 |
kens | I *think* the high level streams may have meant high level paterns, but I no longer remember | 11:22.24 |
kens | remembers to update repo | 11:30.54 |
paulgardiner | Robin_Watts: fix for Bug 694753 on paul/master | 12:54.20 |
Robin_Watts | looking now. | 13:22.29 |
| looks good to me. | 13:23.30 |
paulgardiner | Great. Thanks. | 13:23.40 |
| Robin_Watts: another couple of small commits on paul/master | 14:31.09 |
henrys | kens:high level pattern streams is what the bug refers to, search for have_pattern_streams, a field in the graphics state, for usage in the code) | 14:47.18 |
| Happy New Year everyone! | 14:47.39 |
kens | henrsy, right,but I think I added it to one of XPS/PCL already ? | 14:47.47 |
henrys | kens:I thought you did xps and not pcl, let me check | 14:48.32 |
kens | You could easily be correct.. | 14:48.40 |
henrys | kens:it is of little value to pcl as all patterns are either 300 or 600 dpi images | 14:49.01 |
kens | Its in pxink.c | 14:49.05 |
| and xpstop.c | 14:49.12 |
| henrys, but you repeat the images ? | 14:49.37 |
| I guess in PDF that's not a huge expense | 14:49.47 |
henrys | kens:I haven't quantified it, I'm just guessing. | 14:51.21 |
kens | OK well it looks like pxink.c will return 'remap' if the device supports high level patterns | 14:51.48 |
| So we end up capturing the pattern in pdfwrite | 14:52.00 |
henrys | kens:right so pcl remains right? | 14:52.25 |
kens | pxink.c is not PCL ? | 14:52.34 |
henrys | PXL | 14:52.39 |
kens | Aha | 14:52.42 |
| SO yes I guess PCL remains then | 14:52.47 |
| Any clues where I should start ? Or shall I fetch my ball of string :-) | 14:53.10 |
henrys | sure let me refresh my memory here for a mon | 14:53.35 |
| s/mon/mo | 14:53.40 |
kens | No rush | 14:53.46 |
kens | discovers the PS version is unreasonably complex looking | 14:54.15 |
henrys | kens:yes I think Igor's approach could have been simpler. | 14:55.30 |
kens | looks like pcpatrn.c is the PCL implementation | 14:58.39 |
henrys | kens:you did do this | 14:58.47 |
| the code is in gsptyp1.c | 14:58.55 |
kens | THere is code in there for 'something' :-) | 14:59.07 |
| Oh pixmap_remap_mask_pattern ? | 14:59.39 |
henrys | kens:well pcl uses gs_makepixmappattern and it looks like you fixed that to support the patterns | 14:59.45 |
kens | I htink I did yes | 14:59.53 |
| Its a bit hazy now.... | 14:59.59 |
| I don't suppose you know an example file I could test quickly ? | 15:00.10 |
henrys | kens:for me too | 15:00.12 |
kens | I must confess it looks complete | 15:00.41 |
henrys | kens:yea one second | 15:00.49 |
Robin_Watts | paulgardiner: Looking now. | 15:01.04 |
paulgardiner | ta | 15:01.12 |
kens | rebuilds ghostpcl | 15:01.27 |
Robin_Watts | tor8: So, I reckon all but one of the commits on robin/master are good to go now (subject to agreement on rebinding etc). Which ones are you happy with ? | 15:01.50 |
kens | UAC caught me out again today, ended up with a folder I couldn't delete/rename/take ownership of..... | 15:01.51 |
henrys | I thought that could be disabled | 15:02.20 |
kens | Quite true, it was hte only way I could delete the folder..... | 15:02.44 |
Robin_Watts | paulgardiner: canvas.drawCircle? Do we use canvas to draw the path as it's being constructed or something? | 15:03.01 |
kens | Because it was in my ghostpdl tree, rebase and things weren't working | 15:03.09 |
Robin_Watts | paulgardiner: All the others look good to me. | 15:03.48 |
henrys | paulgardiner: we used 23 and me. | 15:04.24 |
Robin_Watts | henrys: That's been closed down now, right? | 15:04.47 |
kens | 'suspended' I think | 15:04.55 |
henrys | paulgardiner: I did see a recent NY Times article criticizing them? | 15:04.56 |
Robin_Watts | paulgardiner: OK. even the canvas one is fine. | 15:04.57 |
paulgardiner | Robin_Watts: yeah, it's slightly strangely structureed | 15:05.36 |
| henrys: oh okay. I'll look into that if it's still going. | 15:06.11 |
henrys | paulgardiner: I'm afraid not: http://www.theverge.com/2013/11/25/5143464/fda-orders-23andme-to-stop-selling-dna-test-kit | 15:07.50 |
Robin_Watts | henrys: They will still sell the kits, they just won't 'interpret' the results vis.a.vis health | 15:08.24 |
paulgardiner | There site says they've suspended health related testing, but not ancestry tests | 15:08.24 |
henrys | paulgardiner: ah okay, but the new york times article was interesting - a single person getting different results from different companies which has not cast doubt (for me) on the entire business. | 15:09.57 |
| s/not// | 15:10.07 |
paulgardiner | henrys: Yeah I'm suspicious of it but thought I'd check it out a bit further | 15:11.59 |
Robin_Watts | http://www.nytimes.com/2013/12/31/science/i-had-my-dna-picture-taken-with-varying-results.html?pagewanted=all | 15:12.40 |
henrys | tor8:in the tradition of your countrymen I did a "polar plunge" yesterday. At least I think it's a scandinavian thing. | 15:13.21 |
| Robin_Watts: yes that's the article thanks | 15:13.38 |
| paulgardiner: it is cheap though | 15:14.00 |
Robin_Watts | henrys: I think that's those crazy Finns. | 15:15.19 |
paulgardiner | henrys: can't really imagine what's that like. The closest I've tried is cold plunges you sometimes find next to saunas. Those are strange because after the initial pain, you feel strangely warm inside,. | 15:17.33 |
henrys | paulgardiner: I don't recall any warmth | 15:18.00 |
| I guess that is the effect of the sauna | 15:19.41 |
paulgardiner | Yeah, the cold water closes the capilaries down and seals in the heat, although I can't remember whether there was a sauna or it was next to a swimming pool. Long time ago. Certainly never had a pressing desire to repeat the experience | 15:22.54 |
henrys | kens:I think the patterns in fts.0880 should be okay. As I recall any kind of raster op would mess that up so I was trying to find examples without rops | 15:23.41 |
Robin_Watts | henrys: The cvs2git failed again. | 15:29.35 |
henrys | Robin_Watts: ugh should just try and extract epage? | 15:30.05 |
Robin_Watts | I went digging in the files as to why, and it looks like there is a branch "nec-lookclub-branch" and another "nec-lookclub-branch^M" | 15:30.07 |
| I am just trying to extract epage. | 15:30.14 |
| So I've manually stripped the "nec-lookclub-branch^M"'s out, and am retrying now. | 15:30.37 |
henrys | Robin_Watts: surprising they weren't tempted by either svn or later git. | 15:31.42 |
Robin_Watts | Well, wurz had some funky thing that used to produce a git repo from it. | 15:32.04 |
| ahem. wurz = paulgardiner, sorry. | 15:32.18 |
kens | henrys, thanks. (was answering front door and showing fencing man our bropken fence) | 15:32.29 |
Robin_Watts | but paulgardiner stopped getting access to the git repo that was being kept up to date months ago. | 15:33.27 |
paulgardiner | I set up a git mirror using a ruby based utility and some scripts. That allowed a few of us at Picsel to use git rather than cvs, but as Robin says, I lost access recently. | 15:34.43 |
henrys | Robin_Watts: I wonder if cvs -> svn -> git would fair better. | 15:34.54 |
Robin_Watts | cvs2git is part of the cvs2svn project. | 15:35.16 |
| so it uses the same cvs repo parsing code. | 15:35.30 |
paulgardiner | The thing I used was called fromcvs/togit | 15:35.41 |
Robin_Watts | and that's where it's currently falling over, so cvs2svn would hit the same problems. | 15:35.51 |
henrys | Robin_Watts: oh okay I just assumed that path would have been exercised more. but sigh. | 15:35.53 |
paulgardiner | Well it was two things really. togit is a fromcvs plugin | 15:36.07 |
Robin_Watts | paulgardiner: How long did the conversion take? And did you convert the entire history ? | 15:36.22 |
paulgardiner | I used it because cvs2git didn't work | 15:36.27 |
| Yeah the entire history. I think it could take a few hours. It could do updates faster | 15:37.22 |
Robin_Watts | cvs2git doesn't work iteratively, so we can't do 'updates'. | 15:37.52 |
kens | henrys, that file produces 28 patterns, they all appear to be repeated 1 bit images in various colour spaces | 15:38.30 |
paulgardiner | I had to add some fiddles to avoid the top line of the cvs commit messages (which were all identical) ending up as the tag line | 15:38.41 |
Robin_Watts | paulgardiner: Oh, I bet I end up like that :( | 15:39.03 |
kens | henrys so it looks to me like you are correct and teh PCL work is also done | 15:39.10 |
Robin_Watts | paulgardiner: Do you have a record of what you did ? | 15:39.26 |
paulgardiner | I still have versions of the scripts, although possibly a fair bit out of date | 15:39.58 |
henrys | kens:1 bit images? They should be larger than that or do you mean 1 bit per component colors? | 15:40.09 |
kens | 1 bit per component | 15:40.17 |
| mostly 16x16 images I think | 15:40.24 |
henrys | kens:sounds like you can close the bug | 15:41.05 |
kens | THansk henrys | 15:41.10 |
| Wow bug count down to 64, I can't remember the last time I had so few open bugs | 15:42.43 |
henrys | kens: a sign we need to focus on new stuff ;-) | 15:43.33 |
kens | Probably true..... | 15:43.54 |
henrys | hi mvrhel_laptop Happy New Year | 16:16.31 |
Robin_Watts | mvrhel_laptop has been struck dumb from looking at the SmartOffice code :) | 16:24.09 |
paulgardiner | henrys: What do you think of http://bugs.ghostscript.com/show_bug.cgi?id=694709. Is that one we should sort out soon, or leave until a customer requires it? | 16:27.00 |
henrys | paulgardiner: is the OP saying it fails in adobe also? | 16:28.48 |
paulgardiner | No, I think it works in adobe | 16:29.05 |
Robin_Watts | paulgardiner: It strikes me as being something worth looking into. Crashes aren't nice. | 16:29.06 |
paulgardiner | The crash I can probably fix fairly easily. | 16:29.29 |
Robin_Watts | Fixing the crash would be good. But then it would be nice to be able to search for accented chars etc. | 16:29.55 |
paulgardiner | But actually allowing the text to be input would require more effort | 16:29.59 |
| search may be okay | 16:30.11 |
Robin_Watts | I think people can paste text in from other apps at the moment to get the accented chars. | 16:30.26 |
| So if they can paste it in, we should be able to search for it. | 16:30.46 |
henrys | paulgardiner: I still can't parse comment #1 - can you reproduce a crash in our code? | 16:31.06 |
| sorry comment #0 the description | 16:31.26 |
paulgardiner | I don't think this one is about search | 16:31.27 |
| Search may also be a problem though | 16:31.37 |
Robin_Watts | Oh, sorry, yes. | 16:33.07 |
paulgardiner | I think it's two problems. Can't insert accented chars into an empty form field. Attempt to edit a form field contatiing accented chars crashes | 16:33.29 |
henrys | paulgardiner: my read was we didn't do it and adobe crashes | 16:33.50 |
paulgardiner | I think he means he used adobe to insert the chars into the form, and then an attempt to edit with MuPDF crashed | 16:34.38 |
henrys | paulgardiner: but if our code crashes it should be fixed and the rest if probably wait for a customer | 16:34.47 |
paulgardiner | Great. That's sort of what I thought. | 16:35.00 |
henrys | paulgardiner: ah yes that makes sense. | 16:35.14 |
paulgardiner | I know where the crash is already | 16:35.15 |
mvrhel_laptop | good morning | 16:53.58 |
| Happy New Year to all of you | 16:54.10 |
kens | Hi Michael, happy new year also | 16:54.20 |
mvrhel_laptop | henrys: I am going to be out most of the morning today. | 16:55.34 |
henrys | mvrhel_laptop: okay no problem | 16:55.48 |
mvrhel_laptop | kens: did you get in any skiing over the holiday? | 16:55.49 |
| there has been no snow here. it is terrible | 16:56.02 |
| henrys: is there much snow in CO? | 16:56.09 |
| plenty of green though I see | 16:56.22 |
henrys | mvrhel_laptop: plenty in the mountains little near me, exactly as I like it ;-) | 16:56.34 |
kens2 | THere's snow where we are going: | 16:58.26 |
| http://webtv.feratel.com/webtv/?cam=5685&design=960v2&c0=0&bgcolor=246FAC&color1=ffffff&color3=ffffff&color4=ffffff&color5=246FAC | 16:58.26 |
henrys | mvrhel_laptop: vail says 38" base packed powder 4" inches falling today | 16:58.48 |
| kens2:when do you leave? | 17:00.11 |
Robin_Watts | http://www.shotdeadinthehead.com/nice-cup-of-scottish-tea-t-shirt.html | 17:00.45 |
kens | henrys, looks like we are off on Saturday 15th February, back on the 22nd | 17:02.46 |
| Hmm snow isn't great, only 40 cm at the top and 20 at the bottom | 17:06.32 |
| Its very warm too | 17:07.02 |
| Well, 6 weeks to go yet | 17:07.11 |
mvrhel_laptop | kens: I hope you get some snow. the place that I have season passes here has not even been able to open yet. last 4 years they opened in november | 17:09.35 |
kens | Well the resort is open, but I'd normally expect twice that amount of snow | 17:09.56 |
| And its supposed to be 8C on Monday O.O | 17:10.07 |
Robin_Watts | watched "Frozen" the other night. I recommend it to you skiing types :) | 17:10.48 |
henrys | kens:long term climate models do not look skier friendly | 17:11.30 |
kens | THat's the Disney thing ? | 17:11.34 |
Robin_Watts | Also, stay on piste. http://www.telegraph.co.uk/sport/motorsport/formulaone/michael-schumacher/10546682/Michael-Schumacher-remains-stable-but-critical-after-fourth-night-in-hospital-with-serious-head-injury.html | 17:11.41 |
kens | animated something or other | 17:11.41 |
Robin_Watts | kens: definitely not :) | 17:11.48 |
kens | Yeqah, we don't ski off-piste | 17:11.50 |
| Robin_Watts : 2 films called frozen, the new one is a Disney animation | 17:12.27 |
Robin_Watts | http://www.imdb.com/title/tt1323045/?ref_=nv_sr_3 | 17:13.31 |
| Of course, I also watched "Sharknado" and "Machete" over xmas. | 17:14.12 |
henrys | sharknado, seriously? | 17:14.39 |
Robin_Watts | Sharknado plumbs new depths. It's even worse than Battlefield Earth. | 17:14.42 |
| henrys: seriously. | 17:14.47 |
| Machete was spot on though. Robert Rodriquez doesn't miss a beat. | 17:15.29 |
| Oh wow! They are making "Sharknado: The Second One". Did they learn nothing? | 17:16.20 |
henrys | Robin_Watts: I saw the previews and that was plenty | 17:17.17 |
| we started newsroom, it's okay a bit preachy and political for me but well done. | 17:19.23 |
Robin_Watts | Ah, yeah, I like newsroom, but them I'm a Sorkin nut. | 17:21.15 |
| It's not a patch on the west wing though. | 17:21.36 |
kens | Night folks | 17:22.05 |
Robin_Watts | Night kens | 17:22.31 |
henrys | Robin_Watts: I don't know much about him, some parts I like but some of the rapid banter comes across as artificial. too complex for the dialog situations | 17:28.01 |
Robin_Watts | Yes, I felt it worked in the west wing, but in the newsroom, it often feels like everyone is just too clever too fast. | 17:28.53 |
| haha. I just heard from Scott. | 18:21.06 |
| Miles is making an offer on an airstream motorhome. If it's accepted, he and scott will be driving it back to San Fransisco. | 18:23.15 |
| Guess where it currently is... | 18:23.16 |
henrys | I'd assume Texas | 18:25.56 |
Robin_Watts | Orlando! | 18:26.44 |
henrys | Robin_Watts: I hope he's able to keep up with the current business during his "voyage" | 18:28.55 |
Robin_Watts | cvs2git is getting further. Through to pass 4 of 16 now and still going... | 19:06.46 |
| sebras, tor8: You here? | 19:07.11 |
| Hi zeniko | 19:40.57 |
zeniko | Robin_Watts: tor8: sebras: everybody: Happy new year! | 19:41.19 |
Robin_Watts | I put a patch on one of your bugs earlier. | 19:41.22 |
| and to you! | 19:41.24 |
zeniko | Thanks for the patch, | 19:41.54 |
| I'll see how much of a difference it makes for our use case | 19:42.13 |
| (and for curiosity, I'll compare it against just disabling the caching of non-requested objects in streams | 19:43.04 |
| which apparently 2010 me thought would help enough) | 19:43.28 |
| One thing I don't quite understand about your patch: | 19:43.50 |
| Why do you only drop objects with a refcount of 1? | 19:44.01 |
Robin_Watts | IF they have a refcount of > 1 then someone else is holding them in memory. | 19:44.34 |
| If someone else is holding them, then dropping them won't save memory immediately. | 19:44.53 |
| and it means that if we then come along and reload that page later, it'll create a new copy, so we'll end up 2 copies of the same object in memory. | 19:45.18 |
zeniko | Ah, that makes sense indeed. | 19:47.41 |
| So I'll look into testing that code when I get some time and then I'll get back to you about that issue. | 19:48.35 |
| On a different note: There are three more patches for you to look at on zeniko/mupdf when you've got some time to spare, | 19:49.32 |
| two simple fixes for memory leaks | 19:49.44 |
| and one suggestion for an enhancement of fz_load_system_font which tor8 also might want to consider. | 19:50.15 |
Robin_Watts | cool. I will look, thanks. | 19:52.33 |
| The memory leak in pdf_xref_size_from_old_trailer... | 19:53.49 |
| If odf_dict_gets throws, then we still leak. | 19:54.01 |
| so we probably need an fz_always block, I think? | 19:54.13 |
zeniko | pdf_dict_gets shouldn't be able to throw under these circumstances (no indirect references are involved, the resolution of which is AFAICT the only possible reason it could throw) | 19:55.31 |
Robin_Watts | Even so, it's nicer to allow for that, I think. | 19:55.57 |
| I have an amended version of the patch here. | 19:56.05 |
| The second one... do we ever actually allow ourselves to set a 0 0 obj? | 19:56.51 |
| but your fix seems fair. | 19:57.58 |
zeniko | Turns out that we do: If a (broken) document uses 0 0 R, we resolve that and create a pdf_new_null in pdf_cache_object. | 19:58.19 |
Robin_Watts | Possibly we should fix that. but then I wonder if there are broken docs out there that actually use 0 0 obj. | 19:58.49 |
zeniko | I wouldn't be surprised if there were, however the only cases I actually have are files broken through fuzzing. | 19:59.54 |
Robin_Watts | zeniko: If you can't see any problems with http://git.ghostscript.com/?p=user/robin/mupdf.git;a=commitdiff;h=159a038cfa6e8c70c2d4f065732857627249d24c I'll push those 2 fixes. | 20:03.26 |
zeniko | LGTM, thanks. | 20:04.34 |
Robin_Watts | Thanks. | 20:06.41 |
zeniko | Robin_Watts: Would you also take fixes for various compiler warnings? | 20:48.05 |
mavinman | looking for a ghostscript expert to help me with http://stackoverflow.com/questions/14764330/ghostscript-tiffg3-gray-black | 22:50.14 |
| this seems like such a common problem, but i only find people asking the question, and no one answering it | 22:50.34 |
| essentially, i would like no dithering on my tiff output | 22:51.24 |
Robin_Watts | zeniko: Sure. | 23:52.57 |
| Forward 1 day (to 2014/01/03)>>> | |