| | <<<Back 1 day (to 2014/01/12) | 2014/01/13 |
| paulgardiner | tor8: two iOS commits on paul/master. Possibly this is ready for a release now. | 13:55.03 |
| tor8 | paulgardiner: not having an ios device handy to test, how does the animation work? | 14:01.08 |
| | does the onTap trigger only on release? | 14:01.21 |
| | paulgardiner: how do you like dancing around apple's framework oddities and bugs, like in the last patch? | 14:02.23 |
| | it drove me nuts... | 14:02.26 |
| paulgardiner | Oh. Handn't considered that, but it does seem to. Just see the one animation | 14:02.36 |
| | tor8: can't say I exactly enjoy dancing around the oddities, but it seems no worse than other systems I've had to suffer. Android, though, seems to somehow keep things more upfront. | 14:04.37 |
| tor8 | paulgardiner: android at least lets you look at the source | 14:05.49 |
| paulgardiner | With Android, you seem to get less for free and have to combine various parts to get behaviour (like e.g. polling their inertial scroll class and using it's result to move the page), but the advantage is things aren't hidden so much in the standard view classes | 14:06.44 |
| | Come to think of it, I'm using the gesture handled, aren't I. So it would definitely be only on the release | 14:11.17 |
| | Monitoring raw UI events, I'd have had to be more careful. | 14:11.48 |
| | tor8: look okay? | 14:29.55 |
| kens | chrisl Bug #694947 I wonder if they might really be using a level 1 PostScript printer ? O.O | 14:34.20 |
| chrisl | kens: that would be... amazing! | 14:34.56 |
| kens | I can't really see how else setpagedevice could be throwing an error. pswrite does have osme old bacward compatible code for calling setpage instead. | 14:35.32 |
| chrisl | You can always just ask | 14:36.44 |
| kens | I sot of have, in that I've asked for the error and noted that ps2write isn't going to work on a LL1 printer. | 14:37.13 |
| | So the LL2 spec is 23 years old now, even assuming people were still produing LL1 printers for 5 years after the spec was published, it would have to be an 18 year old printer..... | 14:38.08 |
| chrisl | Oh, I hadn't seen that yet..... | 14:38.16 |
| kens | Well it took me long enough to get them to describe the actual problem instead of 'not working'.... | 14:38.42 |
| chrisl | Yeh, doesn't seem to be the most helpful of people :-( | 14:39.05 |
| kens | Well I'm not going to be too helpful to him if it turns out he's expecting it to work an a level 1 printer.... | 14:39.41 |
| tor8 | paulgardiner: so there's an animation when you release the finger but no indication that the finger is down? | 15:01.40 |
| | paulgardiner: that doesn't match my recollection of the behaviour of stock button, but I may well be wrong | 15:02.14 |
| paulgardiner | tor8: well there is sort of: you can tell your finger is down because you can't see the button with your finger being in the way :-) | 15:02.52 |
| Wooziebe | hi everybody | 15:03.01 |
| | can somebody help me? | 15:03.10 |
| tor8 | paulgardiner: if you have baby fingers or one of those capacitive touch pens? ;) | 15:03.49 |
| Robin_Watts | tor8, paulgardiner: 2 commits on robin/master | 15:03.56 |
| Wooziebe | i'm running 7pdf printer and i always recieving the following error: gs:interpreter revision (910) does not match gs_init.ps (853) | 15:04.15 |
| | what is wrong here? | 15:04.27 |
| Robin_Watts | tor8: paulgardiner spends hours convincing himself that the light goes off in the fridge when he closes it. | 15:04.39 |
| paulgardiner | tor8: Just checked the iPad and you're right, but I don't know if we should worry too much. You do tend to tap them and then you can't tell the difference | 15:04.47 |
| Robin_Watts | Wooziebe: Sounds to me like the copy of ghostscript that is being called is v9.10, but it's being fed data that assumes t's 8.53 | 15:05.17 |
| paulgardiner | Robin_Watts: Not hours! It took only about 5 min to take all the shelves out and climb in. | 15:05.32 |
| Wooziebe | problem is that i can't find gs_init.ps | 15:05.58 |
| tor8 | paulgardiner: well, as long as there is some visual feedback I'll take your word for it that it's good :) | 15:06.22 |
| | paulgardiner: the commits look fine | 15:06.27 |
| paulgardiner | tor8: ta | 15:06.36 |
| chrisl | Wooziebe: on recent ghostscript releases the initialisation files are built into the executable | 15:06.39 |
| Robin_Watts | tor8: paulgardiner or my commits ? | 15:07.26 |
| Wooziebe | ow like that so it's actually 7-pdf that feds GS wrong? | 15:07.32 |
| chrisl | Wooziebe: if could be - why not ask the 7-pdf developers? | 15:07.53 |
| Wooziebe | because they don't reply :-s | 15:08.05 |
| | the bastards | 15:08.08 |
| paulgardiner | tor8: what's involved in a release? I've never done one before. And is there a good way to distribute a prelease test version to some people that have iOS devices? | 15:08.16 |
| ray_laptop | Wooziebe: what OS ? | 15:09.08 |
| Wooziebe | wins 7 - 64 bit | 15:09.15 |
| chrisl | Wooziebe: well the first thing I would do is make sure you only have one version of Ghostscript install, and remove *any* evidence of the previous version (i.e. files that the uninstaller might have missed). | 15:09.33 |
| tor8 | tor8: I'm pretty sure everything has changed since last | 15:10.01 |
| Wooziebe | even cleaned the registry and all other files butt wil check again | 15:10.09 |
| ray_laptop | Wooziebe: When you install gs it sets the path to the init files into the registry. Also gs 9+ uses the init files in the executable | 15:10.13 |
| | Wooziebe: so you shouldn't have that problem unless you have a -I on the gs command line | 15:10.40 |
| Robin_Watts | grabs a late lunch. bbs. | 15:10.40 |
| tor8 | paulgardiner: you used to be able to create a package file you could send to people to upload using itunes, but only if their device ID:s have been compiled into the signing profile | 15:10.45 |
| | paulgardiner: I think you use the Xcode organizer window to create and upload to the iTunes marketplace thing | 15:11.17 |
| ray_laptop | but if gs is being invoked by some 3rd party pdfwriter 'port monitor' you don't control the command line | 15:11.24 |
| paulgardiner | tor8: did you renew your account? I can't remember what we decided to do. | 15:12.05 |
| Wooziebe | just a second ray_laptop ; deleting all installations of 7PDf and GS | 15:12.24 |
| paulgardiner | Robin_Watts: the two commits look fine to me | 15:12.43 |
| tor8 | Robin_Watts: the bottom two LGTM | 15:12.46 |
| Wooziebe | cleaning environment vars and forgotten folders | 15:12.50 |
| ray_laptop | Wooziebe: I recommend getting a new (better) pdfwriter -- we usually recommend primopdf, since those folks actually abide by the licensing terms | 15:12.59 |
| tor8 | paulgardiner: yes, it's been renewed since december sometime | 15:13.03 |
| ray_laptop | Wooziebe: I am not familiar with 7pdf | 15:13.16 |
| tor8 | or maybe even november | 15:13.20 |
| paulgardiner | tor8: is there a charge for a release or is it included as part of the annual fee? | 15:13.41 |
| tor8 | paulgardiner: that's all included | 15:13.56 |
| | so no, no more money needs to change hands for a release | 15:14.07 |
| kens | ray_laptop : Wooziebe GS_LIB environment variable will cause this error | 15:14.14 |
| ray_laptop | I _really_ hope we get an Artifex pdf 'virtual printer' soon, so we can tell people to 'just use ours' | 15:14.51 |
| Wooziebe | kens: i've set the GS_LIB environment variable still no go | 15:15.28 |
| ray_laptop | kens: was that you or henrys that was going to work on that ? | 15:15.36 |
| Wooziebe | but i will try installing the primopdf | 15:15.40 |
| | get right back at folks | 15:15.51 |
| kens | ray_laptop : I have code that will insert an XPS file into the Windows piepline. | 15:15.58 |
| ray_laptop | Wooziebe: if you have gs 9.10, you don't need any environment variable. The fonrts andinit files are in the executable | 15:16.11 |
| kens | But it still needs an input->XPS | 15:16.14 |
| | Also, that doesn't install a Port Monitor for printing from any application | 15:16.30 |
| | Last I heard Michael was the candidate for that one but I might easily be mistaken | 15:16.50 |
| ray_laptop | kens: right. The port monitor was the issue I thought | 15:16.50 |
| kens | I don't recall what is the status with the putative port monitor | 15:17.07 |
| ray_laptop | just as long as it's not me ;-) | 15:17.15 |
| kens | It is all different in Windows 8 anyway. | 15:17.19 |
| | SO if we come up with something for Wndows _> 7 we will need something else for Windows 8 I think. | 15:17.41 |
| ray_laptop | kens: right. I think mvrhel was tapped for Win 8 printing (to pdf). | 15:17.45 |
| kens | ANd then there is TIFKAM | 15:17.51 |
| henrys | ray_laptop:a port monitor is a problem 8.1 has permission issues. It has to be specially installed and I don't think most users are going to be willing to do it. | 15:17.55 |
| ray_laptop | henrys: a printer driver can be installed to 'print to FILE, so in the XPS printing pipeline we need a printer driver that prints PDF to a file from the XPS, right ? | 15:20.06 |
| henrys | ray_laptop: I'm talking about redmon like redirection - http://answers.microsoft.com/en-us/windows/forum/windows8_1-hardware/windows-81-and-using-printer-redirection-port-rpt1/f2c406f0-7470-472b-801e-4f3857f29fc6 | 15:22.01 |
| Wooziebe | okay, PrimoPDF did the job, thx guys | 15:23.09 |
| | keep up the good work ;-) | 15:23.46 |
| henrys | ray_laptop, kens:If we are going to PDF in the end I would much rather stay away from the XPS stuff altogether or let MS do XPS->PS then we do PS->PDF. | 15:33.42 |
| kens | Sorry wasn't paying attention | 15:35.33 |
| | I'm not sure what it takes to do PS output in WIndows 8 | 15:35.50 |
| | Windows 7 and below its a port monitor, but they differe across versions of Windows | 15:36.07 |
| | And are all a pain to write. | 15:36.14 |
| ray_laptop | henrys: the disadvantage with that approach is that any transparency is flattened, and gradients as well | 15:36.20 |
| kens | ray_laptop : that's only a problem if you aren't planning on printing | 15:36.38 |
| | But yes, for PDF output its a limitation | 15:36.48 |
| | We can (with many caveats) do XPS->PDF right now though | 15:37.20 |
| | SO a 'Port Monitor' which accepted XPS as an input would be acceptable | 15:37.36 |
| henrys | kens:yes it is the many caveats that is the concern | 15:37.38 |
| kens | henrys if we want to do this we cna spruce up the XPS interpreter | 15:37.54 |
| henrys | kens:from my read of the spec - XPSdrv will produce PS | 15:38.01 |
| kens | And I cna pay more attention to the open bugs | 15:38.05 |
| henrys | kens:true | 15:38.16 |
| kens | henrys yes, but how do we intercept the PS from Windows 8 ? | 15:38.24 |
| | THat's where the Port monitor comes in in earlier versions of Windows | 15:38.44 |
| marcosw1 | henrys: I've updated <http://bugs.ghostscript.com/show_bug.cgi?id=694835> | 15:40.45 |
| henrys | kens:I'm still studying the spec but it does look like you can tap in after that conversion. Anyway we'll look at that later for now the xps redirection will have to do. | 15:43.00 |
| | marcosw1: and what of the original plot? | 15:44.49 |
| Robin_Watts | tor8, paulgardiner: Thanks. | 15:45.47 |
| henrys | kens: anyway by tomorrow I hope to be read up on this and we can discuss it at the meeting with michael | 15:46.16 |
| kens | henrys oK then you will know more than me, its been a while since I looked at this | 15:46.34 |
| henrys | paulgardiner: the important thing over the next few days is a time estimate for the project. That's from Miles not me, I'd rather you just go at it and not think about the schedule yet but Miles really wants to know something about his investment, which seems reasonable. | 15:54.37 |
| paulgardiner | henrys: time scale for integrating the office agents from Picsel code into MuPDF? | 15:55.47 |
| henrys | paulgardiner: yes, just viewing | 15:56.37 |
| paulgardiner | henrys: okay. I'll see what I can do. | 15:57.33 |
| Robin_Watts | henrys: I've just sent some notes on the picsel code to tech. That should give people the information they need to be able to check it out, build something and try to run it. | 16:03.28 |
| henrys | Robin_Watts: yes I saw it thank you. | 16:03.49 |
| Robin_Watts | Although if people want to actually try the main app, it's going to be far simpler for them just to buy it from the app store. | 16:04.04 |
| paulgardiner | Robin_Watts: I have a historical apk I could upload somewhere. Could be a year old, but not THAT much has changed I think | 16:09.28 |
| Robin_Watts | paulgardiner: Is that a v2 apk? | 16:09.44 |
| | I only have v1 here, so no file explorer/dashboard etc. | 16:09.59 |
| paulgardiner | The installed app on my S2 is called Smart Office 2 | 16:10.40 |
| Robin_Watts | sounds perfect then. | 16:10.47 |
| mvrhel_laptop | morning Robin_Watts . thanks for the write up about the code | 16:20.53 |
| Robin_Watts | mvrhel_laptop: morning. No worries. I'd been meaning to do it all week, but have been staring at fuzzing bugs instead. | 16:21.22 |
| kens | paulgardiner : Robin_Watts : | 16:24.54 |
| | http://stackoverflow.com/questions/21085295/android-mupdf-reader-set-up | 16:24.57 |
| | I'm thinking he hbas no C compiler ? | 16:25.08 |
| Robin_Watts | kens: yeah, looks like a duff installation of the ndk. | 16:26.15 |
| kens | Hmm that didn't occur to me, I'll add it as a comment thanks# | 16:26.33 |
| Robin_Watts | gcc is part of the ndk. | 16:26.47 |
| kens | I figured that :-) | 16:26.54 |
| Robin_Watts | paulgardiner: Cor. It's great seeing smart office 2. It has a load of the mathpath/grid morphing stuff in it that I spent ages coding, but never saw used in anger before I left. | 16:53.29 |
| paulgardiner | Oh where's that? I'd forgotten about some of my stuff: zoom out to 2up 3up animations, and the page selector carousel | 16:54.43 |
| Robin_Watts | If you load a PDF file, it shows you the pages in a mathpath. | 16:55.09 |
| | When you click on a page to go into the document, if you zoom out far enough you go into a grid, and it morphs to fit different numbers of pages to the screen. | 16:55.41 |
| paulgardiner | Oh yeah. The initial page spinner | 16:55.44 |
| | yeah I did the different number of pages thing, but using your grid morph of course | 16:56.29 |
| Robin_Watts | It looks really neat! | 16:56.54 |
| paulgardiner | Isn't the scoll smooth. And the PDF redraw is fast too, although a bit inaccurage | 16:57.38 |
| Robin_Watts | Presumably this is using Ali Baxters opengl acceleration for unidisp clients/lomond. | 16:58.33 |
| paulgardiner | Ah yeah. I'd forgotten the opengl stuff. That would explain it | 16:59.09 |
| | Robin_Watts: of course, it must have a really clever animation engine too. :-) | 17:02.40 |
| Robin_Watts | tor8, paulgardiner: Another commit on robin/master when you get a mo. | 17:09.24 |
| henrys | Robin_Watts: is zeniko now checking in stuff directly to ghostscript? - he does have a ghostpdl user repo. | 17:10.57 |
| Robin_Watts | henrys: I was not aware he was doing that. I thought he was checking jbig2/openjpeg stuff into his copy of gs, and then uploading that for you to review. | 17:11.58 |
| | But all fixes to jbig2/openjpeg now go via gs. | 17:12.22 |
| | but I thought I saw a commit from him the other day... so either you pushed it for him, or he has moved to committing by himself. | 17:12.52 |
| | or my memory is failing again. | 17:13.01 |
| henrys | Robin_Watts: I've reviewed them. | 17:13.22 |
| | Robin_Watts: but that's all I know I guess I'll go ahead and commit them. | 17:13.37 |
| Robin_Watts | That would fit with my understanding of the way we want this to work. | 17:14.20 |
| | marcosw1: ping? | 17:16.05 |
| | Evening zeniko. | 17:20.51 |
| zeniko | Robin_Watts: Evening already? ;-= | 17:21.05 |
| Robin_Watts | zeniko: Feels that way. | 17:21.16 |
| zeniko | Sure does. | 17:21.29 |
| | Another fuzzing fix on zeniko/mupdf (verify that openjpeg actually allocates data) | 17:21.30 |
| Robin_Watts | zeniko: looking now. | 17:21.45 |
| zeniko | Also, if you wouldn't mind, the fz_load_system_font patch has been approved by tor8 and is ready to land | 17:22.01 |
| henrys | zeniko:I've reviewed your changes you asked about did you want me to commit them or did you want to? | 17:22.53 |
| zeniko | The other patches ("omit line join" up to "colorspace lookup strings") don't hurry but would be nice to have for the next release | 17:22.53 |
| Robin_Watts | The allocation one looks good. I will push that and the fz_load_system_font one now. | 17:23.25 |
| zeniko | henrys: I'm not even sure whether I've commit access | 17:23.28 |
| Robin_Watts | Then I will look at the others. | 17:23.34 |
| zeniko | henrys: I wouldn't want to accidentally break your main repo, so I'd rather let you do the honors | 17:24.50 |
| paulgardiner | Robin_Watts: Fix for 694890 LGTM | 17:25.16 |
| Robin_Watts | paulgardiner: Thanks. | 17:25.36 |
| henrys | zeniko you have an account but I'm not sure if you are in the right group. | 17:27.33 |
| Robin_Watts | zeniko: yeah. We'd never break the repo. Oh no. | 17:28.07 |
| henrys | paulgardiner: have you committed to the gs repo - I don't see you in the group gs-priv, odd. | 17:28.40 |
| Robin_Watts | zeniko: Does your openjpeg commit correspond to a given bug? | 17:28.48 |
| zeniko | Robin_Watts: bug 694888 looked the closest | 17:29.30 |
| paulgardiner | henrys: I can't remember. I've worked on gs only once. Someone may have pushed it for me | 17:29.32 |
| Robin_Watts | zeniko: Thanks. | 17:29.47 |
| henrys | zeniko: I think you should be able to commit now, if not let me know. | 17:30.50 |
| | paulgardiner: I added you | 17:30.59 |
| zeniko | henrys: The entire shebang (all 13 patches)? | 17:31.40 |
| Robin_Watts | zeniko: MuPDF patches still need to be reviewed. | 17:32.42 |
| kens2 | has had enough today. Goodnight all | 17:32.50 |
| Robin_Watts | (all MuPDF commits have to be reviewed by someone other than the original author. It's something we've been trying for a while, and it seems to cut down on the number of silly mistakes) | 17:33.29 |
| zeniko | Robin_Watts: no worries, I still prefer you to do the checkins after review | 17:34.01 |
| Robin_Watts | gs commits don't *need* to be reviewed, but it's still a good idea. | 17:34.04 |
| | sure. | 17:34.05 |
| | (although, I can't comment on jbig/openjpeg ones - henrys may like to be in the loop for those) | 17:34.36 |
| zeniko | I'm still quite used to Mozilla's policy where every commit requires at least one review (and where you could let others do the checkins for you) | 17:35.14 |
| henrys | okay so I'll commit for the http://git.ghostscript.com/?p=user/zeniko/ghostpdl.git;a=summary? Have they been regression tested against gs? | 17:42.04 |
| | zeniko: ^^^ | 17:42.12 |
| zeniko | henrys: my latest clusterrun of the ghostpdl code shows just a few oddities (bmpcmp doesn't show any differences at all) | 17:44.42 |
| Robin_Watts | tor8, paulgardiner, zeniko, sebras: So, just looking at: http://git.ghostscript.com/?p=user/zeniko/mupdf.git;a=commitdiff;h=678aa6495e792b7576dd7007eaf617ab71003acb | 17:49.59 |
| | which has caused me to look back at the stroking code to try to understand it. | 17:50.19 |
| | Why do we hold sn and bn separately? | 17:50.26 |
| | As far as I can tell, they are assigned together, and increment in lockstep. | 17:50.40 |
| | if (sn == 2) seg[0,1] are the beginning/end of the last segment added to the subpath | 17:53.12 |
| | if (sn == 1) then only seg[0] is valid so far. | 17:53.27 |
| | if (sn == 0) nothing has been added so far. | 17:53.57 |
| | if (bn == 2) beg[0, 1] are the beginning/end of the entire subpath. | 17:54.17 |
| | if (bn == 1) then only beg[0] is valid so far. | 17:54.32 |
| | if (bn == 0) nothing has been added so far. | 17:54.42 |
| | oh, no. if (bn == 2) beg[0,1] are the beginning/end of the first segment in the subpath. That's more useful. | 17:57.50 |
| | paulgardiner, tor8: 2 reviews on robin/master now. | 18:19.17 |
| mvrhel_laptop | brb | 18:46.57 |
| Robin_Watts | henrys: How come changes to -zI can produce 99 differences? | 19:31.12 |
| henrys | Robin_Watts: I'll look I pushed it locally | 19:48.56 |
| | Robin_Watts: are you getting regression results from commits by email? I don't seem to be. | 19:52.56 |
| Robin_Watts | henrys: yes. | 19:53.06 |
| | but in the meantime: http://ghostscript.com/regression/cgi-bin/clustermonitor.cgi?report=2c5cd8c29271d8735349353538c4a353f3aa563b&project=ghostpdl | 19:53.37 |
| henrys | Robin_Watts: there's nothing in that commit that isn't #define DEBUG code. | 19:55.57 |
| | weird | 19:56.03 |
| Robin_Watts | So your user push of this didn't show the same differences? | 19:58.27 |
| henrys | Robin_Watts: no but I've had that change locally for a while, maybe something changed but all the code is DEBUG. I'll look for differences. | 20:01.08 |
| Robin_Watts | Does seem very strange. | 20:01.26 |
| | I was going to suggest reverting it and doing a local push to see if the same files change back, but experience tells me that won't work. | 20:01.52 |
| henrys | Robin_Watts: doing diffs locally not seeing anything. | 20:16.12 |
| | Robin_Watts: doing a local push of the commit before the -ZI changes, but we look pretty busy now. | 20:22.14 |
| marcosw1 | thinks he added the new cluster nodes just in time. | 20:27.59 |
| henrys | lunching | 20:28.48 |
| | well at least the regression emails came in. | 20:53.11 |
| Robin_Watts | marcosw1: I'm confused by something you said. | 20:56.01 |
| | You said that each one of the uber servers corresponded to 4 nodes ? | 20:56.16 |
| | Is peeved not one of them? How come it's just 1 node? | 20:57.10 |
| henrys | marcosw1: I have no explanation for this, I've push an earlier version - no diffs and looked locally: http://www.ghostscript.com/regression/cgi-bin/clustermonitor.cgi?report=2c5cd8c29271d8735349353538c4a353f3aa563b&project=ghostpdl | 20:59.58 |
| kens | Robin_Watts : ping ? | 21:21.00 |
| | OK Robin_Watts (for the logs) I'm not sure if you see gs-devel so I've CC'ed you on a reply about LCMS and memory. Given that I've opnly loosely been following your work with Marti, it might be good if you could review what I've said. I cna forward the original mail if you don't have it. | 21:25.22 |
| ray_laptop | Robin_Watts: I don't know about the 'uber' servers, but peeved is a dual Xeon (each 4 core), but is logically the same as 1 8 core machine -- RAM is shared, as are the rest of the MB resources. | 22:12.16 |
| | Robin_Watts: AFAIK, each "node", no matter how many cores, is on IP address. If you run 'top' on a node, you should see all jobs on that node (and not more) | 22:13.54 |
| mvrhel_laptop | bbiaw | 22:18.24 |
| | Forward 1 day (to 2014/01/14)>>> | |
IRC Logs