| | <<<Back 1 day (to 2014/07/14) | 2014/07/15 |
| SpNg | Iâm trying to rasterize an EPS at various sizes. Currently I use -dEPSfitpage -g1000x1000 or some dimension for g and then trim the image. Is there a better way to do this? | 00:17.12 |
| mattchz | marcosw> please would it be possible to change that alias from testdrive@ -> testflight@ ? | 09:18.22 |
| jogux | mattchz: is platform/ios/do-build.sh missing a #!/bin/bash ? | 09:25.14 |
| | oh, wait, no, that's not a new script | 09:25.31 |
| | it's fine | 09:25.35 |
| mattchz | yeah, the other script had #!/bin/sh | 09:25.43 |
| | that one was correct already. | 09:25.53 |
| jogux | nods | 09:25.58 |
| mattchz | I might move the scripts out of platform/ios in a follow up, actually. | 09:28.35 |
| jogux | mattchz: mupdf changes all look fine, although I believe the addition of 'bash' to the xcode pbxproj etc isn't necessary (though neither does it do any harm I guess). | 10:28.42 |
| mattchz | ta. | 10:29.04 |
| | itâs to avoid issues with permissions being wrong on files. | 10:29.15 |
| jogux | ah right | 10:30.05 |
| Robin_Watts | ooh... Is that mail to support the freeloader finally trying to pay something? | 12:33.04 |
| chrisl | It looks that way - but reading it, you can bet they won't be happy with the quote! | 12:33.57 |
| | It probably also means they've got a stack of issues they want looked at, and once they are complete, it'll be "goodnight vienna"..... | 12:35.35 |
| Robin_Watts | possibly someone should mail miles gently reminding him of the history... | 12:37.01 |
| | They can do the "X hours a year support, after which we change Y per hour". | 12:37.31 |
| mattchz | jogux/robin_watts: please would you mind looking at the latest commits on mu-pdf private please? | 12:38.08 |
| | I canât remember how far back you looked, but I think itâs the most recent 3. | 12:38.18 |
| | thereâs now a top-level script to handle the entire build process. | 12:38.28 |
| chrisl | Robin_Watts: I'll drop a mail to Miles and add Scott in..... | 12:38.32 |
| mattchz | the one thing it doesnât do is do the tagging before uploading. | 12:38.40 |
| | I can change this if necessary, although Iâm in two minds. | 12:38.55 |
| | (prefer to change it as a separate commit though, as itâs vaguely non-trivial) | 12:39.17 |
| | (tagging is done at the end atm) | 12:39.42 |
| Robin_Watts | The optional upload step should be the final thing, IMHO. | 12:40.03 |
| | I'm strongly of the opinion that we should never upload without everything having been pushed into golden. | 12:43.37 |
| | but tor8 is the supreme arbiter here... | 12:43.48 |
| | Other commits look OK. | 12:44.04 |
| | though I wonder if git-check-untracked.sh shouldn't be a git config thing. | 12:44.33 |
| | git config -g alias.check-untracked=!check-untracked.sh or something like that. | 12:45.16 |
| mattchz | oops, forgot to switch a path back after testing, iâve fixed that and repushed. | 12:46.07 |
| | robin_watts: in that case, we should probably push too, as part of the script, which Iâm a little nervous about. | 12:46.46 |
| | robin_watts: I didnât know about git-config aliases. | 12:47.07 |
| | that might be a plan, although the only reason it is actually a separate script is that I canât use a shell function for it, as it gets passed into git submodules foreach. | 12:47.44 |
| mattchz | lunches. brb. | 12:48.11 |
| | robin_watts/jogux: so are you ok for me to push those changes then? | 13:45.50 |
| | Iâll look at the upload final step thing separately. | 13:46.04 |
| Robin_Watts | tor8: You about? | 13:46.22 |
| | mattchz: Unless tor8 is here, and able to look now, sure. | 13:46.45 |
| mattchz | itâs all in the private repo anyway. | 13:47.02 |
| Robin_Watts | yeah. | 13:47.45 |
| tor8 | Robin_Watts: mattchz: sorry, my IRC window was hidden... | 13:47.51 |
| mattchz | at some point Iâll do the actual release this was all intended for ;) | 13:48.00 |
| tor8 | how do I access the private repo? | 13:48.01 |
| mattchz | tor8: at the moment just git clone ghostscript.com:/home/matt/private-repos/mupdf-private.git | 13:48.43 |
| Robin_Watts | On casper: /home/matt/private-repos/mupdf-private.git | 13:48.45 |
| mattchz | (I will push it to /home/git-priv/⦠something tho) | 13:49.01 |
| Robin_Watts | The private repo has mupdf.git as a submodule. It seemed the cleanest way. | 13:49.12 |
| mattchz | a bit fiddly, sadly, but seem to work. | 13:49.38 |
| tor8 | Robin_Watts: yes, I think a private repo with the public one as a submodule is the right choice, even though it is fiddly | 13:50.07 |
| | having private/public repos and automatic scripts that filter out the private files on publicly accessible gits is even more fiddly | 13:50.37 |
| mattchz | keeps meaning to look at googleâs ârepoâ tool that they use for android. | 13:51.21 |
| tor8 | mattchz: another option would be to have the entire platform/ios directory as a private git that you just check out in your private workspace, but then you don't get the benefits of having the public and private repositories in sync | 13:52.55 |
| mattchz | nod. | 13:57.37 |
| | thereâs no real ânice way" | 13:57.42 |
| tor8 | mattchz: not having to keep secrets is the "nice way" :) | 13:58.05 |
| mattchz | :) | 13:58.16 |
| | itâs possible that we can distribute the sdk, and perhaps pass the api keys through the environment or something. | 13:58.34 |
| | but, Crashltics havenât got back to me. | 13:58.45 |
| tor8 | mattchz: if it's just a matter of where to keep the Crashlytics.framework, I'd work on the main repo and only check in a placeholder | 13:58.50 |
| | and a blank key | 13:59.05 |
| | then .gitignore the key file :) | 13:59.26 |
| | I don't like the idea of checking in binaries in the main repo, and even less so proprietary binaries... | 14:00.39 |
| mattchz | Could perhaps have done that. Iâm tempted to just go with this though if everyone is happy because it works and might be useful for other stuff in the future, potentially. | 14:01.02 |
| | and it took me ages :-) | 14:01.08 |
| tor8 | it doesn't affect the main repo, so you (and any successors who end up with the task of building the ios binaries) are the only ones to be affected | 14:01.54 |
| mattchz | that was the main goal tbh, not to affect the main repo. | 14:02.27 |
| tor8 | (yeah, sorry, I wasn't paying attention when you discussed this before) | 14:02.35 |
| mattchz | no probs | 14:02.46 |
| | I guess thereâs no âniceâ solution, but this on balance seems one of the best. | 14:03.11 |
| tor8 | but this solution would have been my other alternative (to placeholder the framework and a blank key file protected against accidental checkins with .gitignore) | 14:03.27 |
| henrys_ | chrisl: were you using a debug build for the xpswrite timings? Thereâs a huge difference, lots of checking code. | 14:04.27 |
| mattchz | nods. | 14:04.31 |
| | arguably we could just .gitignore the framework too. | 14:04.38 |
| | it would leave âredâ entries in the Xcode project view, but probably would be harmless. | 14:04.57 |
| tor8 | I might also have suggested using .bash as the suffix for shell scripts that use bash specific features, but I'm as guilty of that as anyone | 14:05.34 |
| chrisl | henrys_: I can't honestly remember - but it was the same build for both xpswrite, pdfwrite and ps2write | 14:05.58 |
| tor8 | ... and prefix git commits with "ios:" or "android:" or "x11:" or "win32:" if they're specific to a platform, but I'm not consistently doing that either :) | 14:06.28 |
| mattchz | heh | 14:06.50 |
| tor8 | makes it easier to skip over the git commits that you're not interested in :) | 14:07.02 |
| | when looking through what's changed | 14:07.12 |
| mattchz | nod. Iâll try and do that in the future. | 14:07.15 |
| | (in the main repo at least; the private repo pretty much everything is ios specific) | 14:07.32 |
| tor8 | yeah. we're not doing it consistently, but I will try to remember to do it myself | 14:07.46 |
| tor8 | nods. | 14:08.03 |
| henrys_ | chrisl: apparently pdfwrite and ps2write donât have very good robustness checking in the debug code, Iâll report a bug to ken ;-) | 14:09.23 |
| | paulgardiner: about the fringe dead people are not allowed to come back to the show unannounced, that really got us upset. | 14:10.38 |
| | last night | 14:11.01 |
| paulgardiner | henrys_: who was that then? It's been a while. | 14:12.02 |
| henrys_ | http://scifi.stackexchange.com/questions/7213/why-did-charlie-francis-appear-in-the-episode-11-of-season-2-of-fringe | 14:12.16 |
| paulgardiner | Ah. I do remember being confused by one or two episodes. Was that the one with the credits changed to have a seventies feel? | 14:14.38 |
| henrys_ | paulgardiner: I havenât noticed that | 14:15.12 |
| Robin_Watts | Yeah, that was a hangover from season 1. | 14:15.43 |
| | You have to imagine that one being shown 'out of order'. | 14:15.57 |
| | But hold onto your hats, because he's in later episodes too :) | 14:16.28 |
| paulgardiner | You shouldn't have more problems like that. Dead people coming back becomes far more common place and acceptable in later seasons. :-) | 14:17.15 |
| henrys_ | Robin_Watts, paulgardiner I can deal with that kind of âcoming back' | 14:18.03 |
| mattchz | robin_watts: did you recommend Banshee? I seem to have that now I have Now TV. | 14:18.13 |
| Robin_Watts | mattchz: Yes. We're just watching Season 2 now. | 14:18.25 |
| mattchz | good? | 14:18.55 |
| Robin_Watts | Yes. | 14:19.08 |
| | It's from Alan Ball (Six Feet Under, True Blood (while it was good) etc) | 14:19.46 |
| mattchz | thereâs just too much stuff to watch. Currently watching the new 24 and Orange is the New Black S2. And I noticed all the Sopranos are now on Now TV. | 14:19.48 |
| | robin_watts: cool | 14:19.57 |
| paulgardiner | Oh that has to be worth a go then | 14:20.04 |
| Robin_Watts | Got the last 24 episode tonight, and started OITNB S2 last night. | 14:20.13 |
| mattchz | oh, is the last one tonight? I wondered when it was. | 14:20.31 |
| | Any idea if that Silicon Valley thing is any good? | 14:21.12 |
| Robin_Watts | Supposed to be, yes. | 14:21.22 |
| mattchz | ah, itâs HBO, probably good then. | 14:21.45 |
| | tor8: robin_watts: stuff is pushed into a private repo on casper. Iâll update the wiki. | 14:25.23 |
| | Iâll probably delete most of what I wrote yesterday, now the script does all the magic for you. | 14:26.20 |
| Robin_Watts | mattchz: Or leave that there as as subsection of "what the script does for you" ? | 14:27.40 |
| mattchz | could do. script does a bit more possibly. | 14:28.18 |
| henrys_ | not much for a gs and mupdf meeting. kens and mvrhel_laptop are out and ray isnât about. Anybody have anything? | 14:30.04 |
| | chrisl and ray have bugs to review. | 14:30.33 |
| chrisl | henrys_: you mentioned trio last night..... I've been thinking maybe we should go with the approach Robin took for mupdf instead... | 14:30.48 |
| Robin_Watts | nothing from me. Just a reminder that I'm away from next saturday (26th) for 2 and a half weeks. | 14:31.19 |
| chrisl | henrys_: I posted my thoughts on Bug694205 | 14:31.46 |
| henrys_ | chrisl: I was surprised it figured highly in the profile. Is there a way to build without it so I can see how it compares to std | 14:31.56 |
| chrisl | henrys_: not directly, you can hack the source easily enough. But with the distros wanting it shared, and Tim Waugh found some problems, I'm considering alternatives | 14:33.06 |
| henrys_ | chrisl: ah okay so weâll let shelly test that and go from there | 14:33.11 |
| | tor8: anything for the meeting? | 14:34.04 |
| chrisl | henrys_: obviously trio is a bit slower than the "stock" implementation, because glibc and co have a load of special (often assembly) code for major platforms. | 14:34.48 |
| henrys_ | chrisl: I must have missed Timâs issue. Is there a bug for that? | 14:34.59 |
| rayjj | morning, all. Logs lag (at least I don't know what chrisl and henrys are discussing) -- /me waits... | 14:35.05 |
| chrisl | henrys_: yes, there is a bug - it's about a rounding issue..... | 14:35.35 |
| Robin_Watts | yeah, it's a lot easier to do printf formatting of FP numbers fast if you can access the FP representation directly. | 14:35.41 |
| chrisl | Robin_Watts: on the other hand, doing the post process replacement of the decimal takes time, too :-( | 14:36.20 |
| henrys_ | chrisl: my performance complaint is really unsubstantated yet so donât let the muddy things yet | 14:36.21 |
| marcosw | rayjj: Did you see the email last week for you with the subject "Follow-up on fts-09-0920.pdf"? | 14:36.57 |
| chrisl | henrys_: your comment is just another factor in my thinking, not the progenitor of the thought | 14:37.26 |
| rayjj | marcosw: yes, that's what I'm working on. I was going to give Len an update yesterday PM, but had laptop problems. | 14:37.35 |
| henrys_ | rayjj: I responded to your xpswrite bug report(s) | 14:38.14 |
| marcosw | rayjj: great, I did try the file with the stock Ghostscript and didn't see any issues, so I presume it's specific to their version/modifications. | 14:38.17 |
| chrisl | henrys_: Bug695235 | 14:38.19 |
| rayjj | marcosw: et al., in case anyone needs peeved, I can only fire it up for limited times. AC problems that won't be fixed until Friday | 14:38.48 |
| | marcosw: yes, I'm able to reproduce it, but tracing down the 'invalidrestore' is painful -- what in the mess of pasta is leaving something on the op stack ... | 14:39.51 |
| | chrisl: have you come up with any techniques for tracking invalidrestore ? I use PDFDEBUG coupled with -ZAI but that's a LOT to plow through which is what's taking the time | 14:41.09 |
| marcosw | rayjj: I was going to ask about peeved. I assumed hardware issues since it's been down for ~4 weeks. | 14:41.37 |
| chrisl | rayjj: I tend to use PDFDEBUG to narrow down the area, and then use old skool stack and pstack | 14:42.01 |
| | rayjj: do you still see gs-devel? | 14:43.04 |
| rayjj | chrisl: OK. It's an AP dict and the 'restore' doesn't happen until after the print page | 14:43.24 |
| | chrisl: yes, I see gs-devel | 14:43.46 |
| henrys_ | so we have a skype discussion at the hour all are invited. | 14:44.12 |
| chrisl | rayjj: did you see the question from Dwight Kelly? I wondered if you had any insight | 14:44.13 |
| | rayjj: I can't remember that area of the pdf interpreter very well - but if you want me to take a look at it, I can certainly do so | 14:44.51 |
| rayjj | well, maybe I don't see gs-devel. There are some in my inbox, but only from back in March. | 14:45.40 |
| chrisl | rayjj: here's the question: http://ghostscript.com/pipermail/gs-devel/2014-July/009669.html | 14:45.59 |
| | and my reply: http://ghostscript.com/pipermail/gs-devel/2014-July/009670.html | 14:46.17 |
| rayjj | chrisl: I'll continue to struggle with it, but if I have any info or questions by the end of the day, I might ask. | 14:47.34 |
| chrisl | rayjj: the other thing is judicious use of .setdebug can cut down the information overload considerably | 14:48.40 |
| rayjj | chrisl: on genht, we have toolbin/halftone/gen_ordered.c but those functions are written in C | 14:49.28 |
| | chrisl: I did put .setdebug off/on around the PDFDEBUG printing. Thanks, but I knew that 'trick' | 14:50.17 |
| chrisl | rayjj: genht: I figured that was the case, I just wanted to check I hadn't missed some arcane piece of gs "magic" | 14:51.13 |
| rayjj | chrisl: well, you didn't mention gen_ordered.c -- I think that might be of more use to the submitter. Do you want me to follow up ? | 14:52.05 |
| | chrisl: in the meantime, I'll look to see why I don't get gs-devel | 14:52.28 |
| tor8 | henrys_: did you send the font metrics and glyph lists off to URW? | 14:53.22 |
| henrys_ | tor8: yes I did | 14:53.38 |
| chrisl | rayjj: Dwight specifically mentioned PS spot functions, so I assumed he has custom ones he wants threshold arrays from. It's not worth spending time on, it was just the Ken and I know Dwight of old, so I wanted to at least reply | 14:53.56 |
| henrys_ | tor8: heâll send a status soon when he gets back to Germany⦠vacationing in the states | 14:54.24 |
| tor8 | henrys_: yeah, I doubt we'll hear anything until august given that July is the great european vacation month. | 14:55.00 |
| chrisl | rayjj: oh, and on gs-devel: I regularly get "too bouncies, go here ... to reactivate subscription" - despite the fact there is no evidence of mail bouncing | 14:55.14 |
| rayjj | chrisl: I'll reply, because we also have 'threshold_from_order' in the code that lets someone that want to use GS and is not afraid of programming | 14:55.17 |
| chrisl | rayjj: okay, thanks | 14:55.42 |
| henrys_ | tor8: could be july and august with a world cup win | 14:56.05 |
| rayjj | chrisl: you load the screen using GS (from the PS source) and call 'threshold_from_oder' to get it into a threshold array. That lets you have a PS spot function. | 14:56.29 |
| | chrisl: reply sent to Dwight via gs-devel. If he worked with you two, then I'm sure he's bright enough to (ab)use ghostscript for what he wants | 15:36.46 |
| chrisl | rayjj: he was actually a Jaws customer - but one of the better ones to deal with (usually!) | 15:37.42 |
| rayjj | chrisl: Oh, wait -- I think Scott and I visited him somewhere in Canada ? | 15:38.33 |
| chrisl | No, he's US..... | 15:39.07 |
| | Alpharetta, GA | 15:39.25 |
| rayjj | chrisl: the name rings a bell, but I've never been there. Maybe I met him at a show | 15:39.42 |
| | it would have been several years ago | 15:39.59 |
| chrisl | Very likely - he's been around for quite some time | 15:40.05 |
| | henrys_: so, there does seem to be a bit of a hit with trio vs "native" sprintf (but note, without the locale "compensation") - but it's pretty small compared to the difference between xpswrite and pdfwrite: http://pastebin.com/rX5A5qb1 | 15:43.10 |
| henrys_ | chrisl: thanks, there is really no way xpswrite is going to be anything like pdfwrite with no fonts. If you force pdfwrite to outlines they should be similar | 15:45.44 |
| | rayjj:donât know if you saw my comments about your xpswrite bug, but Iâm a bit confused, see the bug | 15:47.40 |
| chrisl | henrys_: I don't think there is any way to run pdfwrite like that because of how it handles fonts. I'm not surprised xpswrite is slower (for various reasons), but I am quite surprised by how much slower..... | 15:48.04 |
| henrys | let me compare with pxlcolor | 15:49.36 |
| chrisl | I get about 5 seconds for pxlcolor | 15:50.50 |
| henrys | chrisl: you did set the cache to 0? | 15:51.23 |
| chrisl | Yes | 15:51.46 |
| henrys | chrisl: Iâll look further xps is very bulky of course | 15:53.02 |
| mattchz | tor8: robin_watts and I were discussing the tagging of builds with this script. | 15:53.51 |
| chrisl | henrys: yeh, I was expecting 4-5 times slower than pdfwrite - but I had no real basis for that expectation. | 15:54.07 |
| mattchz | at the moment, the upload of the build to TestFlight is done /before/ we tag the git trees | 15:54.16 |
| rayjj | henrys: I've updated and rebuilt and am trying the bug. | 15:54.17 |
| mattchz | (we tag both mupdf and mupdf-private with the same tag for the release) | 15:54.26 |
| | I felt it was better not to tag until after the build had uploaded, because if it failed to upload, then we might end up with a tag for a build that doesnât exist. | 15:55.12 |
| | robin_watts thought the opposite - i.e. we shouldnât have any builds uploaded that have no corresponding tags. | 15:55.30 |
| | in practice, the difference is pretty minor; the script tags the trees before it exits, and reminds the user that they should push the tags to the repos manually. | 15:56.20 |
| Robin_Watts | We can always retry the upload. | 15:56.21 |
| | but I know how lazy programmers are. | 15:56.34 |
| mattchz | I was vaguely against the script pushing to the golden repos, because that seems too risky. | 15:56.46 |
| | in which case the difference is possibly moot. | 15:57.06 |
| Robin_Watts | mattchz: What I'm hearing there is you arguing against using a script at all :) | 15:57.12 |
| mattchz | not really; all changes are local. | 15:57.32 |
| chrisl | henrys: but based on the other problems, and the performance deficit, I think I'll look at removing trio and going with Robin's solution for mupdf | 15:57.38 |
| mattchz | the user can then inspect them if they like, check that they are happy, then push. | 15:57.46 |
| | once somethingâs been pushed, itâs a PITA to undo. | 15:57.57 |
| | because someone might have checked it out. | 15:58.20 |
| | which is why I wasnât wanting to push from the script. But Iâm happy to change it, if everyone thinks that is preferable. | 15:58.51 |
| | if we go down that route, effectively the upload failing would be a soft failure, and the user would then have to upload the build manually using the web interface. | 15:59.34 |
| | but they canât just re-run the script to do it, because the tags etc. have already been set and pushed. | 16:00.13 |
| rayjj | henrys: I retested and updated the bug. If you want separate bugs, let me know (PLRM-3rd, tiger.eps and colorcir.ps ALL have problems) | 16:05.14 |
| | henrys: Note this is on Windoze 7 with 32-bit build | 16:05.54 |
| | chrisl: Dwight replied privately (not on gs-devel) so I just cc'ed tech. I guess he likes the gen_ordered approach | 16:07.20 |
| | I forgot to mention that was mvrhel_laptop's code | 16:07.48 |
| mattchz | robin_watts: tiny commit in mupdf-private which just adds the upload API key if thatâs ok⦠| 16:07.52 |
| chrisl | rayjj: great, thanks for doing that! | 16:08.21 |
| rayjj | chrisl: np | 16:09.03 |
| | off for coffee. bbiab | 16:09.09 |
| Robin_Watts | mattchz: lgtm. | 16:09.44 |
| mattchz | ta. | 16:09.49 |
| | arse. we already have a 1.5-ios tag from my previous release attempt. I suggest from now one using tags like 1.5-rc1 for testflight builds. | 16:10.40 |
| | and leaving 1.5-ios for the final thing. | 16:10.53 |
| | is there any sane we can delete the existing tag without breaking it for people who have already pulled? | 16:11.24 |
| jogux | well you probably want to do 1.5-rc<something> now right :-) | 16:12.16 |
| mattchz | oh, indeed. | 16:12.36 |
| jogux | I suggest just using ios-1.5-appstore or something for the final build this time :) | 16:12.37 |
| chrisl | I don't remember there being problems deleting tags - I've messed up gs releases/candidates a couple of times, and no one complained when I deleted them | 16:12.54 |
| mattchz | http://nathanhoad.net/how-to-delete-a-remote-git-tag | 16:13.00 |
| | Iâm guessing itâs perhaps not dangerous to delete a tag like it is change a branch history. | 16:13.12 |
| chrisl | That's my feeling - but I'd try to check with tor8 first..... | 16:13.44 |
| jogux | yeah, I don't think it's a massive issue, this only issue is if you've not pulled your tag will still exist and point at the wrong place | 16:14.00 |
| mattchz | chrisl: cool,good to know, thanks! | 16:14.13 |
| | jogux: true. | 16:14.19 |
| | in that case I could delete 1.5-ios, and use -appstore this time. | 16:14.53 |
| | but weâll see what tor8 says :-D | 16:15.08 |
| Robin_Watts | mattchz: Deleting a git tag is fine, as long as we don't have a build somewhere that refers to it. | 16:16.30 |
| mattchz | actually, there is an issue iirc that itâll just get reinstated the next time someone else pushes. | 16:17.07 |
| chrisl | tags shouldn't be pushed automatically in that case | 16:18.23 |
| | In fact, don't you always have to explicitly specify --tags to push tags..... | 16:20.51 |
| mattchz | chrisl: oh? | 16:40.35 |
| chrisl | Huh? | 16:41.17 |
| mattchz | chrisl: was just wondering when tags are pushed otherwise... | 16:41.41 |
| chrisl | Maybe they push automatically if you've created them, but I seems to remember having to specify --tags - that *might* be after deleting a local tag, though | 16:42.42 |
| mattchz | ah right. | 16:42.54 |
| | this link claims you have to do voodoo to actually delete a remote tag: http://nathanhoad.net/how-to-delete-a-remote-git-tag | 16:43.10 |
| chrisl | Hmm, maybe it's when I've deleted and made new local flag, then. It's something I try to avoid doing too often! | 16:44.31 |
| mattchz | yeah, best avoided, I guess :) | 16:45.42 |
| | Iâve updated the page: http://twiki.ghostscript.com/do/view/MuPDF/IosReleases | 16:45.55 |
| | itâs really hard to describe git stuff in english :) | 16:46.12 |
| | new RC build for mupdf uploaded to testflight | 16:52.08 |
| rayjj | mattchz: is that android ? | 16:52.29 |
| mattchz | rayjj: ios | 16:52.40 |
| rayjj | mattchz: OK, thanks | 16:52.48 |
| mattchz | all being well, itâll get uploaded to Apple soon. | 16:53.01 |
| henrys | exhausts all posibilities before using windows egads. | 17:36.11 |
| | rayjj:can you attach your tiger.xps output to the bug when you get a chance. | 18:15.21 |
| | ? | 18:15.23 |
| | for the logs | 18:15.30 |
| rayjj | henrys: I attached the tiger.xps | 18:29.47 |
| henrys | rayjj: same weird numbers as the plrm page 1. How come Michael doesnât see this? I certainly donât get it on linux or mac so it must be specific to windows | 18:41.08 |
| | nvm Iâll look at Iâm sure itâs something Iâve fouled up | 18:42.08 |
| mvrhel_laptop | Robin_Watts: you around? | 18:44.32 |
| rayjj | henrys: did you try 32-bit linux ? | 18:44.54 |
| mvrhel_laptop | what bug/issue are you looking at rayjj and henrys? | 18:45.23 |
| rayjj | the xpswrite bug 695367 | 18:45.41 |
| | mvrhel_laptop: ^^^ | 18:46.38 |
| | mvrhel_laptop: as I understand it, you use xpswrite (with gs) to print from gsview. Is that correct ? | 18:47.29 |
| mvrhel_laptop | yes | 18:47.56 |
| rayjj | mvrhel_laptop: BTW, is there a 'master' repo for gsview ? And what toolchain do I need to build it ? | 18:48.04 |
| mvrhel_laptop | rayjj: it is in a branch on my mupdf repos and you would need VS2013 | 18:48.47 |
| | need to go get kids lunch now. bbiab | 18:49.12 |
| rayjj | mvrhel_laptop: there's a bunch of "small" enhancements that would probably (at some point, after learning curve) be easier to fix than to open bugs | 18:49.16 |
| | mvrhel_laptop: but I don't need to run Win 8, do I ? | 18:49.54 |
| rayjj | hopes not | 18:50.00 |
| mvrhel_laptop | yes. I think so, but I am not sure. also the code is in C# with xaml UI | 18:50.33 |
| rayjj | mvrhel_laptop: you think i *DO* need Win 8 ? | 18:51.02 |
| mvrhel_laptop | rayjj: see if you can install VS2013 | 18:51.08 |
| | I dont know | 18:51.10 |
| rayjj | mvrhel_laptop: OK. I'll give it a whirl | 18:51.49 |
| SpNg | whatâs the best way to output an EPS to specific bitmap size without needing to trim it. Currently I use -dEPSfitpage g1000x1000 and then trim⦠| 18:59.14 |
| Robin_Watts | mvrhel_laptop: Sorry, yes,I'm around. Talking to you on skype. | 19:09.02 |
| rayjj | SpNg: if you use EPSCrop with -r___ it will produce an image that has the width and height, but there is no way (AFAIK) to have gs scale to the largest dimension and trim the other dimension | 19:21.45 |
| SpNg | rayjj: ok, thatâs what Iâm seeing too. Iâm just having problems trimming images after they are created. Sometimes the color white is trimmed when it shouldnât be. | 19:24.09 |
| rayjj | SpNg: you can use -sDEVICE=bbox -dEPSCrop to show you the BoundingBox (or just grep for BoundingBox), then compute the value for -r | 19:24.13 |
| SpNg | rayjj: thatâs an interesting idea. | 19:24.38 |
| rayjj | SpNg: BTW, there is no guarantee that the %BoundingBox (that EPSCrop uses) is correct. It depends on how good the creator of the EPS was | 19:32.54 |
| | I think I found who left the crumbs on the opstack (cust 532's issue) | 21:13.20 |
| tor8 | mattchz: Robin_Watts: (for the logs) deleting tags is fine; as long as you don't reuse the same tag name later and someone has pulled the tag in between | 21:43.22 |
| henrys | rayjj: weird Iâm betting on a trio bug - one of those sprintf is giving bogus data back but I donât see anything wrong with it. | 22:41.53 |
| mvrhel_laptop | rayjj you there? | 23:33.53 |
| | I have a small list of things that I am going to fix up on gsview. not sure if you had been playing with it or not. if you have some things you thought needed fixed let me know | 23:34.29 |
| | I may spend a few minutes fixing my things | 23:34.40 |
| Robin_Watts | mvrhel_laptop: Did you get the testing working to your satisfaction? | 23:37.32 |
| mvrhel_laptop | Robin_Watts: yes I think I would like you or Joseph to review my commit. I will then move on to see if I can fix up some of the remaining issues that I see | 23:39.51 |
| | my repos shows up now in projects | 23:40.33 |
| Robin_Watts | eh? | 23:41.07 |
| | mvrhel_laptop: We can't have that. | 23:41.33 |
| mvrhel_laptop | ah | 23:41.53 |
| | oops | 23:41.57 |
| Robin_Watts | I've moved it :) | 23:42.12 |
| mvrhel_laptop | ok. so where do I find it then... | 23:42.50 |
| Robin_Watts | It's now in private-repos | 23:42.54 |
| mvrhel_laptop | and how do I see it with the web browser | 23:43.18 |
| Robin_Watts | You can't. | 23:43.23 |
| mvrhel_laptop | like http://git.ghostscript.com/ | 23:43.29 |
| | oh | 23:43.30 |
| | ok | 23:43.32 |
| Robin_Watts | Currently we have no web git interface for the private repos, which is a shame. | 23:43.53 |
| mvrhel_laptop | that was why joseph said private-repos | 23:44.01 |
| | and I just stuck it in repos | 23:44.08 |
| | ok | 23:44.10 |
| | makes sense to me now | 23:44.15 |
| Robin_Watts | You're not the first person to have done it :) | 23:44.24 |
| mvrhel_laptop | I need to change things here though with respect to that | 23:44.38 |
| Robin_Watts | yeah, you need to change the remote. | 23:44.48 |
| | I have been doing: git fetch pete then gitk pete/master to do reviews. | 23:45.23 |
| | That's what I mainly use the web view for. | 23:45.57 |
| mvrhel_laptop | ok | 23:46.05 |
| Robin_Watts | I'm going to lose a couple of hours tomorrow. It's my dads 70th birthday, so we're going out for lunch. | 23:58.57 |
| mvrhel_laptop | nice. have a fun time | 23:59.15 |
| | Forward 1 day (to 2014/07/16)>>> | |
IRC Logs