| <<<Back 1 day (to 2013/11/26) | 2013/11/27 |
kens | Wonderful email from customer #396, anyone fancy a trip to sunny Switzerland ? | 08:41.20 |
chrisl | Oh, yeh, I can see everyone jumping at the chance to stick their head in the noose with shouty-agro-customer....... | 08:52.58 |
tor5_ | morning paulgardiner | 10:11.26 |
paulgardiner | hi tor5 | 10:11.36 |
tor5 | I see you've found some fixer for ios7 :) | 10:11.48 |
| fixes, even | 10:11.54 |
| automaticallyAdjustsScrollViewInsets looks like it could be the bug I've seen | 10:12.12 |
| shame about losing the textured backgruond | 10:12.30 |
paulgardiner | yeah. That's the one. | 10:12.30 |
| Maybe we can define our own texture | 10:12.45 |
tor5 | and are you sure about the pbxproj changes? | 10:12.47 |
paulgardiner | Not sure about the pbxproj changes. Xcode added them. I took them out, and it added them again. | 10:13.24 |
tor5 | could reuse the darkdenim3 background that we use for the android app | 10:13.37 |
| the pbxproj changes seem pointless, but will surely break it for anyone not compiling for ios7 | 10:14.01 |
paulgardiner | tor5: yeah, that would be good | 10:14.04 |
| tor5: without those changes, it warns that signing needs to be enabled for iOS 7, then asks if I want to enable signing. I say yes and it puts those changes in. | 10:15.11 |
| I should try it on Xcode 4 to see if it still works | 10:15.29 |
tor5 | paulgardiner: I'll leave the choice up to you. I'd hate losing ios6 support though. | 10:15.37 |
| but that's probably what apple wants us to do, so *shrug* | 10:15.50 |
paulgardiner | Actually, I need to make the automaticallyAdjustsScrollViewInsets call conditional on 7 | 10:16.12 |
tor5 | paulgardiner: does it not exist on ios 6 and earlier? | 10:16.48 |
paulgardiner | I believe not. Building with Xcode 4 will tell me | 10:17.22 |
| I'm now looking at wiring up JavaScriptCore which is built in with 7 | 10:17.59 |
tor5 | paulgardiner: ah! well, carry on then :) | 10:18.24 |
| maybe it'll be possible to extract javascriptcore from the usual webkit, I think it's supposed to be C only (not C++) and open source licensed. | 10:18.52 |
| (or we can pay sebras to work on that quad-wheel thing until it passes all our tests) | 10:19.17 |
| s/pay/pester/ ;) | 10:19.28 |
sebras | tor5: hello? | 10:30.18 |
| tor5: ah, yes. that would be a challenge. espescially since I don't particularly like JS itself. but it annoys me that we need to embed huuuge js engines for four lines of js code. | 10:31.24 |
Robin_Watts | I think there may be a C only javascript engine in Picsels stuff. No idea how good it is. | 10:35.33 |
| tor7, paulgardiner: 2 commits on robin/master (ignore the last 3) | 10:36.28 |
paulgardiner | Is javascriptcore not a good option? | 10:37.14 |
| My comment above wasn't intended to point to a problem. | 10:37.36 |
| Being built in with iOS 7 seemed a good thing to me. Saves us bloating our app | 10:38.09 |
sebras | tor5: occupied with work. | 11:01.17 |
tor5 | paulgardiner: yes, using the built in one is better than shipping v8. best would be one platform independent code base that has a tiny JS interpreter, though. | 11:03.12 |
paulgardiner | javascriptcore has a nice small API. I've not looked to see whether the source is huge and hairy. | 11:04.05 |
Robin_Watts | chrisl: Did you speak to shelly last night about those bugs? | 11:19.37 |
| paulgardiner, tor8: Another fix on robin/master - still ignore the last 3. | 11:32.52 |
paulgardiner | Robin_Watts: 5 down looks fine. 4 down looks like it should work and is clever, but makes me wonder whether better would be to fix error handling in the rendering pipeline. | 11:42.00 |
Robin_Watts | paulgardiner: Yes, I'd love to fix the error handling, but it's not trivial. | 11:42.35 |
| This at least is safe. | 11:42.38 |
| The problem is that if we throw an error during a 'push', do all callers then catch that and don't do the equivalent 'pop' ? | 11:43.28 |
| Or do we assume that all callers DO the pop? | 11:43.44 |
| What is most natural will depend on what the caller is doing - for instance the display list will most probably want to do the pop, because it's just playing out calls from a list. | 11:44.37 |
paulgardiner | Yeah. Fair enough. Does sound nasty. And, in any case, the current fix can always be removed if we fix the the pipeline | 11:45.10 |
| One thought: could pop take an argument saying what to pop. | 11:45.29 |
Robin_Watts | Possibly we could label each level in the stack with the type of operation that got it there (was it a begin group or a push clip etc?) | 11:45.52 |
paulgardiner | Then if something tried to pop after not pushing it could silently fail | 11:45.55 |
Robin_Watts | but then how would we handle that when we go wrong? | 11:46.20 |
paulgardiner | Ah. So is isn't unique pointers that are being pushed | 11:46.22 |
Robin_Watts | Suppose we: begin group, push clip, begin group, push clip and that last one fails. | 11:46.51 |
| when we get an end clip, how do we know where we are? Should we ignore it? | 11:47.26 |
paulgardiner | Current fix looks good anyway. | 11:47.30 |
Robin_Watts | Thanks. I'm open to suggestions for better fixes. | 11:47.48 |
paulgardiner | Yes. Nasty. | 11:49.01 |
tor5 | Yes. Very nasty. | 12:04.50 |
| I gather we haven't spotted the ownerpassword bug because almost nobody ever uses owner passwords? | 12:05.24 |
Robin_Watts | tor5: presumably. | 12:05.53 |
| also, it may not even be possible to make such a file with current acrobat? | 12:06.14 |
Robin_Watts | will now attempt to run a bit further than the half mile ha managed yesterday before the migraine struck. | 12:07.14 |
chrisl | Robin_Watts: I did speak to Shelly, he is going to take a look at them. | 12:56.38 |
Maloji | hi | 13:17.33 |
ghostbot | hola | 13:17.33 |
Maloji | can any one help me about thus MUPDF | 13:17.51 |
| error | 13:17.53 |
| I have built it and evrytime i run it..shows error | 13:18.22 |
| ExceptionInInitializer error | 13:19.06 |
henrys | Robin_Watts: any word about the office stuff? and BTW I hope you are better today | 14:03.48 |
Robin_Watts | henrys: feeling much better. | 14:03.59 |
| I tried to contact him yesterday and he was out of the office for a vacation day. | 14:04.11 |
| will try again now. | 14:04.13 |
henrys | Robin_Watts: is it natural light or monitor light as well that triggers it i.e. should you take some time off? | 14:04.58 |
Robin_Watts | henrys: It's unclear what triggers it. | 14:05.27 |
| I think it's to do with bright lights flashing at me, like sun through trees or headlights etc. | 14:05.59 |
| but even that is a wild guess. | 14:06.03 |
henrys | it seems like I can't escape the brits - work, running club, bike club they just keep popping up - like californians or something. | 15:31.40 |
Robin_Watts | henrys: We get everywhere. | 15:33.36 |
chrisl | Well, we did used to own everywhere....... | 15:34.47 |
henrys | and obviously there are far too many of you for that island you live on. | 15:35.33 |
chrisl | Or, if you read the Daily Mail news paper, we're being forced off it by marauding hordes or East Europeans | 15:36.21 |
henrys | oh yeah those crazy bastards that work 8 hours a day like mexicans in the US. | 15:39.06 |
Robin_Watts | Another of the valgrind issues boils down to a jbig2dec one. Will park that one then. | 15:49.14 |
henrys | chrisl: so I should report that bug to urw. Do we know that the UFST has cyrillic coverage. | 15:49.23 |
| ? | 15:49.24 |
| chrisl: and wanted to ask about shelly because we should get these fixed since we've had a complaint from a potential customer. | 15:50.48 |
chrisl | henrys: I haven't actually reached that point yet, I was only checking whether the fonts were consistent with the information from URW+ themselves. | 15:51.13 |
| I told Shelly there a bit of urgency on the bugs, and to make sure he kept the bug threads up to date - especially if he ran out of time, or reached an impasse | 15:51.59 |
| henrys: in the glyph information I extracted from cust 532's product, there are a bunch of glyph names that look "odd", and aren't listed in the UFST metrics files, so I'm trying to fathom if they have extra glyphs in their FCOs or if they have some crazy mapping going on..... | 15:55.53 |
henrys | chrisl: do we know they are using FCO's for PDF and Postscript - perhaps they just licensed some type 1' from micro type and have their own scaler for postscript. | 16:01.56 |
chrisl | henrys: they are definitely using FCOs for our PDF, yes. They are *not* using Type 1's in their PS. | 16:03.10 |
henrys | chrisl: okay | 16:03.27 |
chrisl | I am 90 per cent certain they are using the MT fonts for their PS | 16:04.18 |
kens | chrisl when are you heading out to Maui ? | 16:16.58 |
chrisl | kens: Next Tuesday | 16:17.18 |
kens | Ah ok | 16:17.22 |
| I noticed Michael was heading out today (unless I misread his comment on IRC) | 16:17.40 |
chrisl | kens: I didn't want to be away more than a week | 16:19.54 |
kens | Makes sense | 16:20.06 |
chrisl | <sigh> that's another merge commit...... | 16:20.57 |
kens | the one from till ? | 16:21.14 |
chrisl | Yes | 16:21.38 |
kens | Has anyone told him we don't do that ? | 16:22.03 |
chrisl | At least twice | 16:22.17 |
kens | Oh great | 16:22.22 |
chrisl | I'm tempted to suggest we "park" his account | 16:23.37 |
henrys | chrisl: just pull the commit and tell him to rebase. | 16:26.56 |
chrisl | henrys: that will cause problems for anyone who has pulled since the commit was pushed | 16:27.39 |
kens | Me for one | 16:27.47 |
henrys | chrisl: I must have missed it. | 16:28.06 |
kens | I may have done a commit since then as well | 16:28.11 |
chrisl | henrys: it the current "top" commit in the central repo | 16:28.29 |
henrys | I wonder if we could disallow merges somehow without a force or something .... | 16:28.50 |
chrisl | kens: you haven't pushed a commit..... | 16:29.03 |
kens | No, I see I did mine first :-) | 16:29.15 |
chrisl | paging tor5! Any idea if we could disallow merge commits in the ghostpdl repo? | 16:29.40 |
| henrys: so, do we want the merge commit zapped on casper? | 16:31.35 |
henrys | chrisl: sure | 16:32.14 |
| chrisl: that was the policy we agreed to - I don't personally agree but that's another issue. | 16:32.48 |
chrisl | henrys: IIRC, the policy was because of the way the cluster tracks the history - it needs the history to be linear | 16:33.29 |
henrys | I don't remember the cluster part of the argument, but let's not worry about it pull the commit and let's move on. | 16:34.39 |
Robin_Watts | chrisl: merges won't upset the cluster too much. | 16:34.46 |
| it just won't test all along the branch that was merged. | 16:35.13 |
chrisl | kens: you'll want to stash what you're working on, and (assuming you have no local only commits) do: git reset --hard HEAD^^ and then git pull --rebase | 16:35.40 |
kens | I'm currently clean, wa slooking at a problem I passed over to Micahel | 16:36.03 |
henrys | chrisl: we all agreed to the policy and fair warning was given. | 16:36.04 |
chrisl | Robin_Watts: in that case, I don't remember the core of the argument - nor do I care to try.... | 16:36.19 |
| tkamppeter__: ping | 16:36.55 |
henrys | chrisl: wow have you seen this guitarist bonamassa? just incredible | 16:37.07 |
chrisl | henrys: I've been wittering on about him for ages - seen him live at the Hammersmith Apollo last year | 16:37.45 |
henrys | freakishly talented | 16:38.19 |
chrisl | I've got five or six of his albums. He's probably my favourite "current" performer | 16:38.28 |
| henrys: when he was 12 years old, he was the support act for a BB King concert | 16:40.01 |
marcosw | chrisl: you removed the last Till commit but left 87a7fd. If it's all good now I'll fix the cluster archvie. | 16:40.50 |
Robin_Watts | henrys: just spoke to admins. Legal ownership situation is all cleared up now. | 16:41.10 |
chrisl | marcosw: yes, I only removed the merge commit, not the "valid" - hmm, did we lose kens's commit, then? | 16:41.41 |
kens | I'll check | 16:42.00 |
chrisl | No it *seems* to be there..... | 16:42.29 |
kens | Both mine were before till's | 16:42.30 |
| last one from me was c8cfa99 | 16:42.51 |
chrisl | Yes, just a moment's confusion | 16:43.15 |
Robin_Watts | There was a company in Malta as well as one in the UK, for tax avoidance purposes. Both are in liquidation. both now being liquidated by the same people, and there is a clear legal statement that the IP is licencable safely. | 16:43.29 |
kens | That sounds like as good as it's going to get | 16:44.09 |
Robin_Watts | And I've just been given the name of a technical person to talk to figure out exactly what we would be interested in. | 16:44.30 |
tkamppeter | chrisl, still there? | 16:49.11 |
chrisl | tkamppeter: yes. About the commits earlier - we don't allow merge commits on ghostpdl | 16:49.42 |
henrys | anyone know if I can buy window 8.1 on the windows store for running on vmware fusion or should I go buy a dvd? | 16:49.59 |
tkamppeter | chrisl, I was preparing a commit and pushed it and there seemed to be another commit having gone in the way and I have simply pulled and then it did this merge. | 16:50.40 |
chrisl | tkamppeter: you should "git pull --rebase" to get your tree up to date and ready to push | 16:51.19 |
| tkamppeter: there is a way to make "pull" default to rebase instead of merge, but I can't remember what it is right now...... | 16:52.09 |
henrys | never mind I see how I can buy windows 8.1 "shutters" no pun intended | 16:52.11 |
tkamppeter | chrisl, sorry for producing the noise, did not know about this. | 16:52.25 |
chrisl | tkamppeter: these things happen..... | 16:52.52 |
tkamppeter | chrisl, but my changes are still in the repo? | 16:53.12 |
chrisl | tkamppeter: you can see the current history here: http://git.ghostscript.com/?p=ghostpdl.git;a=summary | 16:53.45 |
tkamppeter | chrisl, thanks, they are, seems to be all OK. | 16:55.10 |
chrisl | Except something does seem to have gone missing, I think from kens somehow.... :-( | 16:55.44 |
kens | Hmm, I will look, I can always regenerate it, fortunately the hard ones are (I hope!) safe | 16:56.11 |
chrisl | It's changes in pdf_draw.ps | 16:56.34 |
kens | Yes, and its gone | 16:56.41 |
| Odd. | 16:56.48 |
Robin_Watts | Hold on, let me have a look at the history. | 16:56.53 |
kens | Yes definitely vanished | 16:57.56 |
tkamppeter | chrisl, if I compare the history with the e-mail notifications I got, "pdfwrite - ignore annotation /AP where the appearance is not a stream" from kens is missing. | 16:58.08 |
chrisl | tkamppeter: yep, I see the same thing. | 16:58.29 |
tkamppeter | chrisl, kens, so the merging seems to be broken, unreliable, perhaps even a bug in GIT. | 16:59.02 |
chrisl | tkamppeter: no, merge works, we just don't allow commits with two parents. | 16:59.25 |
| kens, Robin_Watts: I have Ken's commit back here, if you want me to push it | 16:59.48 |
Robin_Watts | chrisl: If you can push to put the repo back into proper shape, then go for it. | 17:00.08 |
chrisl | kens2: your commit should be back | 17:00.55 |
kens2 | I'll pull and check, thanks chrisl | 17:01.11 |
chrisl | Robin_Watts: gotta love reflog ;-) | 17:01.15 |
Robin_Watts | chrisl: Alas, I never pulled during the affected time, so I don't have the refs to log :) | 17:01.48 |
kens2 | Yes, that looks good thanks | 17:02.14 |
chrisl | Robin_Watts: I pulled it before I zapped the offend commit, for largely that reason | 17:02.21 |
| Okay, excitement over, back to farting about with fonts...... | 17:02.54 |
| tkamppeter: okay, so you can tell git pull to default to rebase for the master branch by doing "git config branch.master.rebase true" | 17:05.44 |
| Robin_Watts: your turn to deal with the web site security fetishists...... | 17:27.24 |
kens2 | Well, that's Raed's latest dumb file polished off, but he won't like the answer; 'its fixed but you need to upgrade to current code'. I believe its to do with the colour management in pdfwrite, and he needs *all* the fixes I've made and I certainly am not going to try and pick them out one by one. | 17:30.51 |
| So I'm off now, goodnight all | 17:30.57 |
henrys | chrisl: 691213 is only about the urw afpl 136 right? Not GPL fonts | 17:50.54 |
chrisl | henrys: yes, the 136 set - erm, are they AFPL? I thought they were commercial only...... | 17:52.03 |
henrys | chrisl: oh I thought it was the same arrangement as the pcl fonts. I'll check the contract with miles | 17:55.39 |
chrisl | henrys: tbh I just assumed they were commercial only because they're not downloadable anywhere "public" | 17:56.24 |
Robin_Watts | Who wrote the sampled function code in mup | 18:03.17 |
| df | 18:03.21 |
henrys | chrisl: mail sent to urw | 18:05.49 |
chrisl | henrys: cool, thanks. I'd like to get that bug put to bed, and it will mean I can include those fonts in my comparisons | 18:06.37 |
Robin_Watts | Hours of staring at code, and it turns out to be a missing "-1". sheesh | 18:17.26 |
| paulgardiner, sebras: 1 new commit on robin/master (4th one down) | 18:27.19 |
| Hi zeniko | 19:53.44 |
zeniko | Hi Robin | 19:53.59 |
| In case you've got some minutes to spare, there are some commits on zeniko/master which should be ready for merging | 19:55.37 |
Robin_Watts | Sure. let me have a look. | 19:56.05 |
zeniko | From "fix memory leak..." until "fix regression" (6th to 4th down) | 19:56.05 |
| Your own commit looks good (I'd just have expected whitespace around the operator for consistency) | 19:57.48 |
Robin_Watts | font memory leak thing looks good. | 19:58.54 |
| The fontpath thing looks OK to me, but it would be good to have a comment in there saying "This is kept for use for downstream consumers, such as SumatraPDF". | 20:00.25 |
| Otherwise I can see us removing it again when we forget why it's there :) | 20:00.40 |
zeniko | I've updated the commit with such a comment in font.h | 20:04.15 |
Robin_Watts | Looking at the render glyph scissor rectangle change. | 20:05.40 |
zeniko | As for the third commit, that issue was discovered in https://code.google.com/p/sumatrapdf/issues/detail?id=2466 (includes a document for reproducing at saner zoom rates) | 20:06.19 |
Robin_Watts | So the problem is that the subpixel movement of the glyph can cause it to clip against the edges of the clip rectangle. | 20:06.24 |
| At high resolutions, we can move the glyph by +/- 1/2 pixel | 20:07.39 |
| So I kinda see the logic here, it's a question of where the scissor comes from. | 20:08.10 |
| If the scissor is from (say) a group or a clip path, then we can't move it. | 20:09.03 |
zeniko | Aren't the glyphs supposed to be moved into the scissor using the ctm (and when glyphs are cached, they're supposed to be in a rectangle starting at 0,0)? | 20:09.58 |
Robin_Watts | subpix_ctm maps the glyph into the rectangle starting at 0.0 | 20:13.00 |
| ctm maps the rendered glyph onto the final output. | 20:13.22 |
zeniko | Right, and fz_render_t3_glyph is passed subpix_ctm and scissor instead of subpix_ctm and subpix_scissor (or ctm and scissor) | 20:14.14 |
Robin_Watts | Oh. this only affects what gets cached, not what gets painted on the screen. | 20:15.36 |
| In that case the scissor can move safely. | 20:15.51 |
| So yes, I think you're right. | 20:15.55 |
| OK. I've got to go to dinner now, but I'll get those pushed to the main repo. | 20:16.12 |
| Thanks! | 20:16.14 |
zeniko | Thanks and enjoy your meal! | 20:16.31 |
ray_laptop | sojic_: You seem to be dropping off and rejoining ghostscript frequently. Can you investigate your network problem and/or leave #ghostscript until you need to be here. My IRC client puts those messages out every time and i get a pair of these every 5 minutes | 21:49.44 |
| marcosw: I noticed that last night's saved-pages-test generated a *LOT* (37,724) md5 sum differences, but when I run a regression test with extras=--saved-pages-test I only get 1,164 non-pdfwrite differences, 473 pdfwrite diffs and 245 ps2write diffs. Do you have any idea why the differences ? | 22:24.08 |
| marcosw: and when I look at a few, I don't get any differences | 22:24.33 |
tkamppeter | chrisl_away, thanks, have applied the command now. | 23:12.01 |
henrys | I guess there is some kind of bug in ray_laptop's expense report spreadsheet joann found an addition error in my report and she said it happens all the time. | 23:22.51 |
| yes mine adds k9 - k28 and my report begins on k7 so it misses the first 2 entries | 23:24.58 |
| I'm a natural at these spreadsheet bugs ;-) | 23:27.42 |
| Forward 1 day (to 2013/11/28)>>> | |