| <<<Back 1 day (to 2014/04/21) | 2014/04/22 |
sebras | Robin_Watts_: well, right now we basically poll state->content_length. | 00:00.38 |
| Robin_Watts_: regardless if its value has changed. | 00:00.46 |
| Robin_Watts_: if you were to wait then we'd need semaphores or something. | 00:01.17 |
Robin_Watts_ | semaphores etc is fine, it's not in the core. | 00:01.54 |
sebras | Robin_Watts_: ok, maybe that is the next step. the first one must be to hand errors though. | 00:02.31 |
| which we do. now. | 00:02.36 |
Robin_Watts_ | gotta love customers. | 09:40.32 |
kens | THe ARabic one that just came in ? | 09:40.39 |
Robin_Watts_ | "Do you want A or B?". "A". Spend day doing A. "I've done A" "Oh, we want B!" | 09:41.04 |
kens | Yeah I thought that | 09:41.15 |
pedro__ | Robin_Watts: what do they want now? | 09:44.36 |
kens | External font instead of built into library | 09:47.10 |
Robin_Watts_ | pedro__: I need to find out whether there is an alien included in the lib we build for them or not. | 09:48.28 |
| If so, I need to check where it looks for fonts. | 09:48.45 |
| I can't see one using ar. | 09:49.03 |
| You said there was font enumeration blah in the alien? | 09:49.28 |
pedro__ | it's set up through a property in the aline/app (PicselConfig_fontDir or something along those lines) | 09:49.54 |
| then read by the font engine on startup | 09:50.04 |
pedro__ | looks | 09:50.19 |
Robin_Watts_ | Thier build line uses --iapt --<customer-name> | 09:54.41 |
pedro__ | ah, funky | 09:55.07 |
Robin_Watts_ | iapt = alien reference applications ? | 09:55.25 |
pedro__ | yup | 09:55.30 |
Robin_Watts_ | So would such a build include an alien in the lib ? | 09:55.51 |
pedro__ | will need to check the current ue2 builds - iapt builds were generally the picselwin32 lib plus a 'nice' alien layer suitable for customers | 09:57.23 |
| ie source they'd build & use as a base for their apps | 09:57.40 |
Robin_Watts_ | yeah, we provide them 3 builds. | 09:58.01 |
jogux_mac | robin : you'd need to check everywhere to see :-) | 09:58.31 |
Robin_Watts_ | a gtk linux one that comes with a zipfile of stuff. | 09:58.50 |
| and 2 lib only builds. | 09:58.59 |
| jogux_mac: I'm having trouble seeing where "everywhere" is. | 09:59.10 |
jogux_mac | but I think the fact that it's gtk gives it away - I don't believe there's any code to do a library-with-an-alien for gtk. | 09:59.41 |
| ios/android/bada/wince/win8 are the ones that support that I believe. | 10:00.16 |
Robin_Watts_ | jogux_mac: There is a linuxgtk-alien in alien-reference-applications | 10:00.20 |
jogux_mac | robin: the main place would be where the --customer-name is handled in tgvbuild.py, that's the most likely place anything would be configured if it's not explicitly on the command line. | 10:00.46 |
| robin : yep - but I don't believe there's any code to build the gtk alien /into/ the library. | 10:01.14 |
Robin_Watts_ | OK, so we can assume that they are building their own alien stuff. | 10:01.51 |
| Certainly if I ar the libs and then grep the filenames that come out 'alien' doesn't appear anywhere. | 10:02.16 |
jogux_mac | sounds like a reasonable assumption based on current evidence. | 10:02.34 |
Robin_Watts_ | So, if we assume that they are handling the alien stuff based on our example code... where would that example code live? | 10:03.21 |
| In alien_reference_applications ? | 10:03.35 |
jogux_mac | for a --iapt build, yep | 10:03.46 |
Robin_Watts_ | jogux_mac, pedro__ : A few small commits on robin/master then. | 10:16.39 |
pedro__ | cool | 10:17.01 |
| Robin_Watts: looks good - and the SIL licence is already in the epage/licenses directory | 10:31.28 |
Robin_Watts_ | pedro__: Ta. | 10:47.14 |
| I'm gonna have a run, then think about how best to do builds for the customer. | 10:48.07 |
| I may need to pull the fontmap changes onto the release branch. | 10:48.17 |
paulgardiner_lap | Hi tor8. I accidently pushed 3 commits directly to golden yesterday (I have one computer with the remotes named differently from the others. Doh!). They were ready to go, but unreviewed of course. Could you give them a look over? | 11:00.06 |
tor8 | paulgardiner_lap: airprint, bitmap based printing and share icons? | 11:03.52 |
| paulgardiner_lap: there's some messed up tab/space indenting in the first commit (the airprint one) | 11:05.13 |
| and the second one... | 11:05.50 |
paulgardiner_lap | tor8: okay. I have a follow up for that in any case, so I can sort out tabbing within that. | 11:07.13 |
tor8 | paulgardiner_lap: actually, there's been quite a few whitespace errors slipped in over time in the ios code | 11:07.43 |
| I blame Xcode :) | 11:07.55 |
paulgardiner_lap | Ah good. I thoroughly agree. | 11:08.16 |
tor8 | let's just do a separate whitespace cleanup commit | 11:08.23 |
paulgardiner_lap | Okay no probs | 11:09.21 |
tor8 | paulgardiner_lap: there is a git config trick you can do to clean up tab/space mixups automatically when you commit | 11:09.21 |
| argh, ^W didn't do what I expected... | 11:09.56 |
paulgardiner_lap | I think I may have that git config but not for .m files I'd guess | 11:10.27 |
tor8 | paulgardiner_lap: if you run scripts/gitsetup.sh it'll set up the configuration and hooks | 11:10.30 |
| .m files are listed as "tab" files in the .gitattributes | 11:10.49 |
paulgardiner_lap | okay. I'll give it a go, although I'm pretty sure I ran that before... I guess it may have changed since I ran it | 11:11.39 |
tor8 | paulgardiner_lap: anyway, the commits look fine (but the share icons are not used atm?) | 11:16.16 |
paulgardiner_lap | tor8: Also, AFAICT, there is no way to determine the dpi or resolution of the printer, At the moment I'm using points as pixels which is not going to be acceptable, I'd imagine. I'm just looking at uping by a factor of 4 (hence 288 dpi). Any thoughts? | 11:56.26 |
tor8 | hm, I'd pick 300dpi as that's a pretty common (low-end) printer resolution | 11:56.59 |
| 600dpi would probably be better, but the size bloat of the images might be too much to handle | 11:57.21 |
| paulgardiner_lap: there's also a whitespace fix commit on tor/master | 11:58.31 |
paulgardiner_lap | ta | 11:59.17 |
jogux_mac | paul : assuming this is mupdf/airprint... can't get you just get the resolution of the current graphics context? | 12:04.01 |
paulgardiner_lap | jogux_mac: didn't see how. I tried looking at the user-space to device-space mapping, but that was 1-1 | 12:06.27 |
| tor8: commit LGTM | 12:06.35 |
jogux_mac | paul > is it possible that what you have is right then, or have you seen evidence of poor output quality? | 12:07.13 |
paulgardiner_lap | jogux_mac: I've used only the printer emulator, so I guess it's possible that I'd get a more sensible value with a real printer. | 12:08.26 |
tor8 | it might be that when the destination printer is "Print to PDF" the device space may be the same as the user space | 12:08.34 |
paulgardiner_lap | I'm selecting colour laser or something like that | 12:09.01 |
jogux_mac | paul : hmm. dunno. I can check on a real printer tonight if you want. I suspect the emulator is spot on though. | 12:09.52 |
paulgardiner_lap | jogux_mac: that would be useful thanks. As you say, I'd expect the emulator to be spot on | 12:11.18 |
| I can make dpi a variable of my routine for now. | 12:11.50 |
tor8 | paulgardiner_lap: sorry to saddle you with two new bugs (really, they're just asking why isn't 1.4 up on the app stores yet) | 12:33.40 |
| I figured you would want to handle the iTunes app store; maybe Robin should have the android one instead? | 12:34.00 |
paulgardiner_lap | yeah, sounds like a plan. iOS should be good to go once I have the print quality sorted. We have a share icon now and I've fixed the problem of not being able to edit files incoming from other apps | 12:35.30 |
Robin_Watts_ | I can do the android app build(s) if required. | 12:35.44 |
tor8 | Robin_Watts_: please :) | 12:37.40 |
jogux_mac | paul : oo, awesome. | 12:38.05 |
Robin_Watts_ | jogux_mac: What does the -language switch affect? | 12:40.40 |
| Is it purely which fonts are included? | 12:40.50 |
jogux_mac | robin : iirc, it /doesn't/ affect fonts | 12:40.58 |
pedro__ | unicode filters iirc | 12:40.59 |
Robin_Watts_ | Or does it bring in other things too (cmaps maybe?) | 12:41.00 |
jogux_mac | but does affect encoding tables | 12:41.02 |
| and perhaps cmaps | 12:41.07 |
| ie. ja-jp might cause eucjp -> utf filters to be built in. maybe. | 12:41.29 |
Robin_Watts_ | Pondering whether to add ar-ar to the customers builds. Probably. Won't hurt. | 12:41.51 |
jogux_mac | or maybe it does affect fonts. hm. | 12:42.00 |
| for tgv, I'm not sure it affects the library. | 12:42.15 |
| as everything is meant to be triggered via the alien calling Picselsomething_register-language_jajp | 12:42.34 |
| or something like that :) | 12:42.37 |
Robin_Watts_ | New commit on robin/smart-office-2-1-28-release-branch then. Just pulling the fontmap changes from master. | 12:48.04 |
jogux_mac | [but the code in the alien is often written to pay attention to the defines that -language= sets] | 12:52.39 |
henrys | horror - I've run out of coffee on meeting day | 12:58.59 |
kens | Disaster, but you still have time for a quick dash to the nearest store | 13:00.55 |
henrys | bbiab | 13:04.12 |
paulgardiner_lap | tor8: It's working at 300dpi now, and I'm managing to obtain a big enough buffer to render a whole page in one go, at least under the emulator. Do you think I need to restrict the buffer size and render strips? And if so do they need little overlaps in case there's inaccuracy where strips might meet off of a pixel boundary? | 13:10.17 |
| tor8: I'm tempted to stick the current version up on testflight see how it pans out. | 13:11.34 |
tor8 | good question, but I suspect we can get away with a single page at 300 dpi on most devices | 13:12.45 |
jogux_mac | how much memory is that? | 13:12.53 |
kens | depends on the media size | 13:13.31 |
tor8 | jogux: about 32MB (4 bytes per pixel) | 13:14.08 |
kens | 24 Mb for 8bpc RGB letter I think | 13:14.10 |
| or yes 32 Mb for CMYK | 13:14.21 |
tor8 | kens: we always render with alpha in mupdf | 13:14.31 |
jogux_mac | ouch! I don't think you'll reliably get away with that on a 4th gen ipod touch. | 13:14.32 |
| (and I'm guessing that figure is even higher if pdf has an A3 page?) | 13:14.46 |
kens | yes, double it | 13:14.54 |
| But will people really print A3 form such a limited device ? | 13:15.13 |
jogux_mac | I'd say aim for 4-8M strips. | 13:15.30 |
| kens : do you even have to ask that? :-) | 13:16.07 |
kens | Yeah I know..... | 13:16.20 |
| SOme of our customers print *way* bigger than that..... | 13:16.37 |
tor8 | someone will always try do the stupidest thing imagineable | 13:16.40 |
jogux_mac | is it reasonable for them to is a slightly different question, but I'd argue we should at least gracefully fail, and it's probably as easy to do strips than gracefully handful the fail :) | 13:16.58 |
paulgardiner_lap | jogux_mac: yeah, I was thinking 4M. | 13:17.03 |
jogux_mac | even on a 2nd gen ipad I think you'd struggle to get 24-32M + scratch space + ui memory etc. | 13:17.26 |
kens | thinks VM error is all peopl elike that deserve..... | 13:17.43 |
tor8 | rendering in strips is pretty easy, but managing that without the output getting cracks where they abut might be troublesome. a bit of overlap as paul suggested would actually be a pretty neat solution. | 13:17.55 |
paulgardiner_lap | we do fail gracefully, where "gracefully" means "no one dies" | 13:18.03 |
jogux_mac | kens : on iOS the app will just exit with no explanation if the OS decides it's unhappy with your memory usage :-S | 13:18.07 |
tor8 | I suspect that what the ios printing pipeline does is create a PDF from all the draw calls, and then print that | 13:18.17 |
kens | crappy Apple 'operating system'..... | 13:18.26 |
tor8 | jogux_mac: I thought that was more common for if your app was in the background? | 13:18.57 |
kens | resolves to stay further way from such things | 13:19.08 |
paulgardiner_lap | Seeing as I'm unable to detect the actual resolution, I bet the joins between strips will be visible | 13:19.26 |
jogux_mac | tor8 : it'll usually jetisson background apps first, but if the foreground app triggers it's "greedy app" trigger that will go. | 13:19.28 |
paulgardiner_lap | ... but we can play "it's your printer mate" card | 13:19.50 |
tor8 | linux does the same thing if it's unhappy with your memory usage... hard kill if it discovers you actually use the memory it overcommitted to you :) | 13:19.51 |
| paulgardiner_lap: well, an overlap should be easy to do, and if you make it reasonably sized nobody will be the wiser and we won't have to play the card :) | 13:20.37 |
paulgardiner_lap | tor8: yeah it does create a pdf | 13:20.42 |
jogux_mac | paul> I may have missed something, but I thought your overlap trick should cover that. | 13:20.43 |
tor8 | I wish more people generating pdfs with abutting image tiles would do that overlap thing | 13:20.53 |
| I think we've had to turn off antialiasing on images altogether just to not see the image seams everywhere :( | 13:21.28 |
paulgardiner_lap | jogux_mac: yeah I guess the results from giving a little overlap should be acceptable | 13:21.57 |
henrys | paulgardiner_lap, Robin_Watts_ can sot stuff be discussed here or do we need Skype? | 13:32.43 |
paulgardiner_lap | henrys: either works for me. I think we went to Skype just for customer specifics | 13:36.14 |
henrys | paulgardiner_lap, Robin_Watts_ let's do SOT state of affairs after the gs/mupdf here then. | 13:38.11 |
Robin_Watts_ | jogux_mac, paulgardiner_lap, pedro: Did anyone nod at robin/smartoffice-2-1-28-release-branch ? | 13:48.32 |
paulgardiner_lap | Robin_Watts_: I'll take a look now | 13:49.07 |
pedro__ | sorry Robin - got stuck figuring why the Good SQL DB has suddenly stopped authenticating on this win server - will take a look | 13:51.52 |
| looks fine - do we assume they have the font available natively or are you going to bring across he romfont changes too? | 13:57.58 |
Robin_Watts_ | They are going to have it natively. | 13:58.10 |
pedro__ | cool | 13:58.16 |
paulgardiner_lap | Robin_Watts_: so although it's an alternative requested by a particular customer it's good for all? | 13:58.34 |
Robin_Watts_ | After asking for it to be built in the lib, and me spending all yesterday sorting that out, they then changed their mind this morning. | 13:58.34 |
pedro__ | :( | 13:58.48 |
Robin_Watts_ | paulgardiner_lap: Extra entries in the fontmap never hurt us. | 13:58.52 |
| It just means we recognise that font as being an arabic one. | 13:59.14 |
paulgardiner_lap | okay good | 13:59.29 |
kens | meeting in 20 minutes ? | 14:08.46 |
Robin_Watts_ | yup. | 14:08.53 |
kens | Good, coffee time then..... | 14:09.03 |
ray_laptop | looks like I need to reboot. system running slowly. it's been at least a week, so Windows is due anyway. :-( | 14:11.29 |
henrys | My plan was to follow the agenda ⦠looking at support bugs this looks alarming if real http://bugs.ghostscript.com/show_bug.cgi?id=695173 , Robin_Watts_ ? | 14:31.11 |
kens | I was leaving that for chrisl I couldn't easily put together a test on Linux | 14:31.59 |
chrisl | I saw it, but I've been rather busy | 14:32.27 |
henrys | kens:in the gs section I made eps2write "done" but it seems like some other stuff could be checked off? | 14:32.49 |
kens | I know, I did have a quick bash at it, but my Linux skills aren't up to it.... | 14:32.57 |
henrys | chrisl: can I assign it to marcos to reproduce? | 14:33.05 |
kens | henrys, let me look | 14:33.08 |
Robin_Watts_ | henrys: I haven't looked at that. I will try to look this afternoon. | 14:33.10 |
chrisl | henrys: it's assigned to support anyway | 14:33.23 |
kens | henrys you can clearly close the PDF/VT thing. By the way, do we have a central repository for docs ? I have two ISO specs for that one. | 14:34.08 |
| teh colour management continues..... | 14:34.20 |
| Not sure where we want to go with gsprint, I've sort of thought that was with Michael | 14:34.39 |
henrys | support used to get fielded and dispatched quickly now I think we have so many of bugs it's become a place where things fester. | 14:34.55 |
mvrhel_laptop | sorry I am late | 14:35.02 |
henrys | kens:hold on! | 14:35.17 |
kens | :-D | 14:35.32 |
henrys | sprint chrisl and kens: can we do all this stuff with XPSPrint() http://pages.cs.wisc.edu/~ghost/gsview/gsprint.htm | 14:35.51 |
kens | henrys I'mnot sure all of it is relevant, by using XPS we are levering much more off the Windows engine | 14:36.42 |
henrys | we want to keep compatibility with that list of options I think. | 14:36.43 |
| yes a few things are irrelevant now. | 14:37.02 |
kens | We certainly don't do 2-up, nor convert to mono | 14:37.09 |
marcosw | late, sorry. | 14:37.25 |
henrys | kens: you didn't see that anywhere in the XPS api? | 14:37.50 |
mvrhel_laptop | can you specify that stuff in the print ticket? I cant remember | 14:37.53 |
| henrys: I will look at what features I have in gsview printing. it will be the same | 14:38.14 |
kens | henrys, I don't think that would be in the XPS API | 14:38.14 |
mvrhel_laptop | I think it is in the print ticket set up isnt it kens? | 14:38.30 |
henrys | kens:maybe not - but a lot of those other features must be there. | 14:38.54 |
kens | If we want n-up we need to convert 2 pages into a larger one, or scale them down to the current size, neither are really printer options IMO, but we could probably do somethign with it, depending what the input is (PS & PDF are easy) | 14:39.02 |
ray_laptop | interesting that the only N-up setting was twoup | 14:39.09 |
kens | Indeed | 14:39.17 |
henrys | kens:anyway we can all have a look at this this week. I need to fix up the xps too. | 14:40.09 |
ray_laptop | many windows printer drivers support N-up with full GUI options to show the layout and allow for defining margins between the small pages | 14:40.12 |
| so I'm not sure what value just twoup is | 14:40.27 |
mvrhel_laptop | henrys: yes that would be great | 14:40.29 |
henrys | ray_laptop: that's why I thought the XPS pipeline would have something. | 14:40.39 |
kens | henrys it would be the printer end, you would configure your pritner that way | 14:40.56 |
| I've yet to make head or tail of the print ticket stuff which is where thsi would all live | 14:41.17 |
| Ideally you owuld embed the print ticket in the XPS file and then you don't have to worry about it when printing | 14:41.47 |
henrys | pdf-vt struck down. We don't have a centralized repo for documents. | 14:42.02 |
kens | Which would make it your problem of course ;-) | 14:42.03 |
Robin_Watts_ | henrys: /home/specs on casper? | 14:42.22 |
kens | OK would anyone *like* a copy of the PDF/VT spec ? I'd like to back i up somewhere off-site | 14:42.29 |
chrisl | We could just add a directory in svn-private | 14:42.37 |
ray_laptop | chrisl: that makes sense | 14:42.53 |
henrys | chrisl: great | 14:43.03 |
chrisl | kens: send them to me, I'll put them in svn | 14:43.13 |
kens | Thanks chrisl | 14:43.24 |
henrys | kens:shall I just leave CM or do you want me to tick off sub items? | 14:43.49 |
ray_laptop | I have a bunch of specs that I could put there. Some are old, but they may not be available readily on the net | 14:44.21 |
kens | henrys that ons is not completed | 14:44.32 |
henrys | skipping down to /tmp files | 14:44.39 |
| paulgardiner_lap: what were we going to do? my guppy memory strikes again. | 14:44.58 |
kens | It looks OK as is | 14:44.59 |
henrys | ray_laptop, chrisl watching the performance discussion - could we have caught any of this in regression testing. This is really painful for a customer and if we can avoid it, it'll pay off | 14:46.51 |
paulgardiner_lap | sorry got carried away with this airprint thing. just catching up with the meetng | 14:47.34 |
| gs temp files? | 14:47.57 |
henrys | ray_laptop: saved pages with psdcmyk? I assume you are working on that as time permits? Should it be an enhancement? | 14:47.59 |
| paulgardiner_lap: yes? | 14:48.03 |
| s/?// | 14:48.09 |
paulgardiner_lap | I think what I had in mind will work, but while I was trying to implement it I found what looked like a shortcut, but that didn't work | 14:48.42 |
chrisl | henrys: it's hard to see how - IIRC, neither Ray nor I have actually been able to reproduce the performance deficit in the more recent code | 14:48.45 |
paulgardiner_lap | I think going back to plan A will work | 14:48.52 |
| henrys: my plan was to look at tmp files again on the flight to the next meeting if we can wait until then | 14:50.42 |
henrys | was ray_laptop going to try out the raspberry pi AirPrint stuff? I don't have a name for that agenda item. | 14:50.52 |
| paulgardiner_lap: oh that's right. | 14:51.06 |
| paulgardiner_lap: I've added a comment so I won't forget again | 14:51.55 |
paulgardiner_lap | henrys: you also mentioned atexit which would be vastly simpler, but AFAIK at exit doesn't get called if you SIGKILL a program (but maybe I'm wrong) | 14:52.41 |
henrys | chrisl: the MAXPATH thing kind of bothers me. Are you sure malloc() would be too painful? | 14:52.45 |
chrisl | henrys: that's done, ages ago | 14:52.56 |
henrys | chrisl: oh ages? before the last meeting? | 14:53.31 |
chrisl | henrys: no after: http://git.ghostscript.com/?p=user/chrisl/ghostpdl.git;a=commitdiff;h=fe0b8fcf | 14:53.43 |
Robin_Watts_ | paulgardiner_lap: Yeah, we abandoned atexit specifically because it can't be relied on. | 14:54.02 |
henrys | chrisl: super | 14:54.09 |
chrisl | henrys: I moved as much as I sensibly could into allocated memory | 14:54.26 |
paulgardiner_lap | Robin_Watts_: Okay good | 14:54.53 |
henrys | chrisl: yeah I'm looking - I do read the commits I must have missed or forgotten this one. | 14:55.09 |
| marcosw1: there are some customer notifications needed in bugzilla. | 14:55.48 |
| for fixed bugs | 14:55.53 |
| on to the mupdf section: I continue to ask miles about the domain each time I talk to him, that's all I can do. | 14:56.44 |
| we are trying to get financial backing for the URW fonts. | 14:57.10 |
| tor8:anything on the mupdf list you want to discuss? | 14:57.47 |
Robin_Watts_ | "should all mupdf files have copyright headers? #robin" - No. So you can close that item :) | 14:58.18 |
henrys | Robin_Watts_: done | 14:58.52 |
| 2 more minutes to this meeting, anything else? | 14:59.04 |
tor8 | henrys: not really. I've asked Robin to do the android release, and Paul looks like he'll be able to do the iOS release any day now | 14:59.48 |
| and I'm working my way through the mupdf bug list | 15:00.07 |
Robin_Watts_ | Just to mention VisualGDB to windows users that occasionally have to debug on linux and would prefer to be able to do so from within visual studio. | 15:00.14 |
paulgardiner_lap | henrys: on MuPDF, there's just one tiny bit left to do on the iOS app and that should be ready to release. Interacts with other apps correctly now. Just a little more work in Airprint to improve dpi without using huge buffers | 15:00.15 |
chrisl | Robin_Watts_: do they do "site" licensing? | 15:00.39 |
Robin_Watts_ | For > 10 licenses. | 15:00.46 |
chrisl | Hrm, overkill, then :-( | 15:00.59 |
Robin_Watts_ | yeah. The linux version is quite reasonably priced. | 15:01.13 |
henrys | I'm going to add the Good integration to the agenda list. | 15:01.55 |
Robin_Watts_ | Though I suspect the "quick debug" feature is one that people would like, so that means you need the "custom" version. | 15:02.23 |
chrisl | Robin_Watts_: oh, and the jom thingy - could you open a bug, or send me an e-mail with what you did and found summarised. I'll look at it when I have a bit of time | 15:03.12 |
Robin_Watts_ | chrisl: Sure. | 15:03.21 |
chrisl | Robin_Watts_: assuming you're already snowed under...... | 15:03.38 |
Robin_Watts_ | chrisl: getting there :) | 15:04.01 |
henrys | Robin_Watts_, paulgardiner_lap : should we discuss SOT status? | 15:04.33 |
paulgardiner_lap | sure | 15:04.59 |
pedro__ | henrys: we have a windows 2012 server up and running in a VM with Good control server and proxy. Progress slower than (I'd) expected due to numerous windows user authentication issues, sql server bugs/auth access issues and problems partly due to figuring out the Good docs. | 15:05.09 |
Robin_Watts_ | henrys: The 'arabic' issue, looks to be solved by including a relavent font (or at least a font mapping). | 15:05.59 |
pedro__ | henrys: we have a windows 2012 server up and running in a VM with Good control server and proxy. Progress slower than (I'd) expected due to numerous windows user authentication issues, sql server bugs/auth access issues and problems partly due to figuring out the Good docs.i've been working throguh the integration chnages for the new Good SDKs (they've deprecated the secure doc transfer from... | 15:06.01 |
| ...the APIs used in SOT) | 15:06.02 |
Robin_Watts_ | I am doing new builds for the almost-customer now. | 15:06.09 |
paulgardiner_lap | pedro__: to be honest, I expected that to be hell | 15:06.36 |
pedro__ | its partly learning curve, part hell ;) | 15:06.48 |
Robin_Watts_ | Miles contacted me about another (related) potential customer earlier in the week, and I've investigated their problem, and got back to them that that is solved. | 15:06.55 |
paulgardiner_lap | henrys: the rebranding of splash screens and other app-included docs I believe | 15:07.07 |
| insert "is done" somewhere in there | 15:07.28 |
jogux_mac | paul : that means that there's more work to do on the ios good than we expected/hoped :( | 15:08.53 |
ray_laptop | sorry. phone call I had to take. | 15:09.26 |
paulgardiner_lap | jogux_mac: yeah, just saying that doesn't surprise me at all | 15:09.35 |
marcosw | are we still on track for a SOT release the end of this week? | 15:10.32 |
jogux_mac | henrys : ATS wise, I now have the disk from marcos with everything on it - I've copied the VM images onto my machine at home so will be looking at getting them booted and fixing the various issues they have, and /then/ I can start actually trying to make ATS do something (and then the fun of figuring out how to make it work with git etc) | 15:10.41 |
henrys | pedro__: miles is eager to have a good release, so if anything can be done in parallel while you work on the windows server that would be good to hand off | 15:11.16 |
paulgardiner_lap | What do we feel must be completed before we can release? | 15:12.38 |
henrys | Robin_Watts_: I think he's been copying me in on all the customer stuff. I saw that one | 15:12.43 |
pedro__ | sure - we basically have 3 parallel chunks - the server/framework bits, the android app and the ios app - both ios and android need updated - haven't tackled anything on ios app yet so that may be worth handing off | 15:13.36 |
Robin_Watts_ | henrys: yeah. I don't think I'm supposed to actually do new builds for them. | 15:14.01 |
jogux_mac | pauigardiner : at least ATS testing I guess, overnight da tests might be nice (but probably don't have time) | 15:14.36 |
henrys | pedro__: I had the impression jogux_mac was going to look at iOS? | 15:14.47 |
Robin_Watts_ | Nothing should distract jogux from ATS, IMHO. | 15:15.03 |
| (except when I have questions, obviously :) ) | 15:15.18 |
jogux_mac | henrys : it was on my list, but I've been on holiday, and it's now a bigger task :-( | 15:15.22 |
paulgardiner_lap | jogux_mac: Yes ATS is one. The rebranding was another. The layout improvements that we have so far. Maybe that's it. | 15:15.45 |
henrys | paulgardiner_lap: we do need ATS testing before we release, so maybe you take iOS? does that make sense and Robin_Watts_ look at android? | 15:16.18 |
paulgardiner_lap | Yeah I can take iOS | 15:16.32 |
jogux_mac | we need to bash the app about a bit as an enduser on ios7, as this'll be the first ios7 sdk release. | 15:16.39 |
paulgardiner_lap | jogux_mac: you had it building at least though, didn't you? | 15:17.11 |
jogux_mac | paul: yes. it failed ;-) | 15:17.18 |
marcosw | miles is eager for me to do a comparison of new vs. old SOT showing the layout improvements. Can I do that with the current git code or are the further improvements coming before the release? | 15:17.22 |
henrys | paulgardiner_lap: I didn't think we had other code changes to do a release for "Good" Have you changed all the logos and picsel-like stuff? | 15:17.31 |
jogux_mac | henrys : good have changed their iOS SDK API, so we have code changes to do | 15:18.00 |
paulgardiner_lap | henrys: yeah, I believe I've caught them all including th good-specific stuff | 15:18.10 |
Robin_Watts_ | marcosw: Any comparison you do now will be disappointing. Miles needs to let us get over the start up hump and actually do some development work that's not driven by our tails being on fire before he looks for such things. | 15:18.24 |
marcosw | Robin_Watts_: you haven't worked with marketing before have you :-) | 15:18.50 |
henrys | jogux_mac: yeah I meant other than "Good" changes, i.e. we agreed not to hold up the release for any bugs in bugzilla, right? | 15:18.56 |
jogux_mac | henrys : right, yes (sorry) | 15:19.08 |
Robin_Watts_ | I am utterly out of the loop w.r.t. good, so I'm not sure what's required for the android work. | 15:19.14 |
paulgardiner_lap | henrys: although there is another issue related to branding: the app has some documents online and it currently kicks off a browser to the picsel site | 15:19.47 |
henrys | marcosw1: I'll talk to Miles today and he won't bug you about that. | 15:19.49 |
jogux_mac | robin : I suggest me/Pete give you/Paul a quick brain dump post meeting. | 15:19.53 |
Robin_Watts_ | marcosw: Yes, but I find it works best if you treat them like naughty dogs. Scold them often, but don't expect them to change their ways. | 15:19.58 |
pedro__ | there are various changes needed; the android port is 'in development' - aka we're figuring out what exactly is implemented and what needs to be completed. Add to that the fact the SDK has changed | 15:20.24 |
paulgardiner_lap | henrys: it also has some sort of "sign up" feature which I don't know much about. That might be hosted also on picsel. | 15:20.35 |
jogux_mac | paulgardiner : signup feature is gone if you build the appstore release :-) | 15:20.59 |
marcosw | henrys: you don't have to do that, I think i've explained it to him. | 15:21.02 |
paulgardiner_lap | jogux_mac: excellent | 15:21.14 |
marcosw | henrys: I've sent out all the customer notificaitons forclosed bugs. | 15:21.19 |
jogux_mac | paul : or add -define=REGMAN_DISABLE to your build command | 15:21.25 |
henrys | marcosw: okay about miles - thanks | 15:21.32 |
Robin_Watts_ | henrys: I have a couple of gs customer bugs, it seems. Where does that sir prioritywise? | 15:22.04 |
| s/sir/sit/ | 15:22.10 |
henrys | Robin_Watts_: let me look | 15:22.52 |
Robin_Watts_ | resolves one as wontfix. | 15:23.55 |
ray_laptop | Robin_Watts_: I see 2 (P2) bugs for you -- bug 695153 and bug 692351. Did you close a different one ? | 15:26.24 |
henrys | Robin_Watts_: is that other one for mvrhel_laptop | 15:26.26 |
| ? | 15:26.27 |
| Robin_Watts_: I'd rather get some momentum on the good stuff. If they complain we'll change things | 15:27.28 |
| Robin_Watts_: it does seem mvrhel_laptop should have a look at that one. | 15:27.52 |
Robin_Watts_ | 692351 is not a customer bug. | 15:28.02 |
| I closed a different one. | 15:28.11 |
mvrhel_laptop | on 695153? | 15:28.27 |
ray_laptop | then why is bug 692351 P2 I wonder ? | 15:28.40 |
henrys | mvrhel_laptop: yes | 15:28.42 |
Robin_Watts_ | 695153 is to do with interpolated images. | 15:28.58 |
mvrhel_laptop | I can take a look at that in the next day or two | 15:29.03 |
Robin_Watts_ | Specifically it's to do with the code I added to only decode the required patch of an image not quite making the patch small enough. | 15:29.33 |
mvrhel_laptop | ok. | 15:30.04 |
| so you already know where the issue is. | 15:30.54 |
Robin_Watts_ | ish | 15:31.04 |
| So, if I'm understanding the state of play with Good: 1) We have an ios build that should work and doesn't because of changes in the SDK. | 15:31.33 |
| 2) We don't have a Good server working yet. | 15:32.11 |
henrys | Robin_Watts_: okay but mvrhel_laptop can look at it while you help out with "good" right? Nobody else can really work on "Good" just yet. | 15:32.11 |
jogux_mac | robin : it also fails because there's a build option that has vanished since the previous good build. | 15:32.19 |
| (but that's presumably easy to trace back; I've just not had time to delve yet) | 15:32.35 |
Robin_Watts_ | 3) We don't have an android build for good at all currently. | 15:32.40 |
mvrhel_laptop | Robin_Watts_: so I have VS2010 installed now and I think I just have a path issue with python, as me dropping the offending command during build into the python console window runs just fine | 15:33.05 |
henrys | kens:did you want marcosw to reproduce that api bug or are you still going to give it a go? | 15:33.05 |
jogux_mac | henrys : are you / miles making progress on the veracode contract etc? | 15:33.08 |
pedro__ | we have a good server which is running here in a VM (albeit it has just fallen over when I moved it to another machine) | 15:33.10 |
jogux_mac | mvrhel : did you post the log somewhere? | 15:33.36 |
Robin_Watts_ | jogux_mac: Is that a customer name? Don't mention customers by name in here. | 15:33.53 |
henrys | marcosw: there was another bug fix but the customer had his email on the bug, you don't do those right? | 15:33.55 |
Robin_Watts_ | mvrhel_laptop: OK. | 15:34.17 |
jogux_mac | robin_watts : veracode app testing is a prerequiste for submitting the app to good | 15:34.18 |
Robin_Watts_ | oh, right, that thing. | 15:34.27 |
pedro__ | robin: the android build has a bunch of the functionality implemented but now needs updated to use the new SDK | 15:34.39 |
jogux_mac | there seems to be a special veracode build option. | 15:34.41 |
marcosw | henrys: yes, I assume if the customer entered the bug themselves they are paying attention to the bugzilla emails and have already been notified. | 15:34.43 |
henrys | marcosw: fair enough | 15:34.57 |
pedro__ | (it has secure FS and code to integrate with the GoodForEnterprise apps to handle attachment viewing/editing) | 15:35.06 |
Robin_Watts_ | mvrhel_laptop: What command is going wrong? | 15:35.06 |
mvrhel_laptop | 1>EXEC : error : Failed to execute command python C:\vrhel\Artifex_laptop\git_sot\sot\epage/foundation/libraries/io/fs/mkromfss2.py C:\vrhel\Artifex_laptop\git_sot\sot\epage C:\vrhel\Artifex_laptop\git_sot\sot\genroot\windows\x86\vs2010\all_das\debug_puts_enable\excel_editor\pdf_export\search_enable\bin\romfss2-compiler.exe C:\vrhel\Artifex_laptop\git_sot\sot\genroot\windows\x86\vs2010\api/romfss2 | 15:35.42 |
| -blob.h C:\vrhel\Artifex_laptop\git_sot\sot\genroot\windows\x86\vs2010\api/romfss2-tmp libraries/font/library libraries/image/resource | 15:35.43 |
Robin_Watts_ | mvrhel_laptop: Right, I feared it might be that one. | 15:36.11 |
| Any chance I can get a remote desktop login to your machine to debug that? | 15:36.29 |
mvrhel_laptop | sure we can do that if that is what you would like to do | 15:36.48 |
henrys | jogux_mac: I've asked miles to handle that separately with the Good business rep and he said he would - are we waiting for something specific? | 15:36.55 |
jogux_mac | I suspect there may be another error message hiding somewhere in the build log | 15:37.04 |
Robin_Watts_ | mvrhel_laptop: It's a script I wrote, and I've looked at it, and I can't see why it should be failing. | 15:37.12 |
jogux_mac | henrys : no; it's certainly not blocking us currently, just checking it is progressing. | 15:37.24 |
mvrhel_laptop | how is it only on my machine though | 15:37.25 |
Robin_Watts_ | mvrhel_laptop: I have no idea. It works fine on mine. | 15:37.43 |
mvrhel_laptop | oh ok | 15:37.49 |
henrys | Robin_Watts_: before I forget did we verify mvrhel_laptop can build okay with ProjectDir so all msvc's will work? Can we update the readme? | 15:38.47 |
| ProjectDir works for me. | 15:39.04 |
mvrhel_laptop | let me try | 15:39.53 |
henrys | what does ray_laptop use? | 15:41.02 |
Robin_Watts_ | VS2008 :) | 15:41.11 |
mvrhel_laptop | oh you have not commited this change yet | 15:41.14 |
Robin_Watts_ | hence he doesn't have a problem. | 15:41.19 |
| mvrhel_laptop: The change is that in the Properties -> Debugging thing, rather than having a working directory of .., have a working directory of $(ProjectDir)\.. | 15:42.01 |
mvrhel_laptop | let me try that | 15:42.14 |
| I am sure it will be fine | 15:43.00 |
Robin_Watts_ | henrys: I'll update the solution and the ReadMe. | 15:49.08 |
mvrhel_laptop | Robin_Watts_: when do you want to try the remote desktop? | 15:49.53 |
Robin_Watts_ | mvrhel_laptop: Whenever is convenient for you. | 15:50.09 |
mvrhel_laptop | now is as good as any | 15:50.18 |
Robin_Watts_ | Let's discuss this on skype? | 15:51.30 |
mvrhel_laptop | yes. I already started :) | 15:51.47 |
chrisl | ray_laptop: I expect you read my e-mail, but I had all kinds of issues with the new sim from Len..... | 15:57.39 |
ray_laptop | chrisl: it wouldn't build for me | 16:00.09 |
| chrisl: but I did notice the .obj files in the root | 16:00.27 |
| chrisl: luckily there aren't many of the 'aux' files | 16:01.04 |
chrisl | ray_laptop: Adding the AUXDIR definition got it to build for me | 16:01.11 |
| ray_laptop: I'm wondering how it can run successfully on the target - given it accessing freed memory..... | 16:03.27 |
kens2 | Night folks | 16:13.18 |
Robin_Watts_ | Jogu, pete, anyone else. | 16:35.35 |
| I'm hitting a very strange problem on mvrhel's build. | 16:35.48 |
| My python script ends with a sys.exit(0) | 16:36.07 |
| and yet the python script that's calling it, sees the return value as 1. | 16:36.22 |
| which is causing it to think that the build is failing. Anyone ever seen anything like that ? | 16:36.46 |
jogux_mac | woot. it's called from a .postlink right? | 16:36.50 |
| ah, no, prebuild (of course, d'oh) | 16:38.15 |
Robin_Watts_ | ok, I've found the problem. | 16:43.15 |
| print "retval"+str(retval) | 16:43.25 |
| gives retval0 | 16:43.32 |
| but sys.exit(retval) does not then return 0. | 16:43.50 |
| If I use: if retval==0:\n sys.exit(0)\nelse:\n sys.exit(retval) | 16:44.23 |
| I get what I want. | 16:44.27 |
jogux_mac | ah, fun. is this os.spawnve being a PITA again? | 16:44.37 |
Robin_Watts_ | I do not understand why, but I don't think I care :) | 16:44.37 |
| No, changing spawnve to use subprocess made no difference. | 16:44.55 |
| Oh, well... maybe. | 16:45.51 |
| The line in question is: retval = callSystem(....) | 16:46.15 |
jogux_mac | wel, sys.exit() is documented to return 1 if passed a non-number object. and os.spawnve appears to be documented to return an integer. so... who knows. | 16:46.21 |
Robin_Watts_ | I'm passing it a long. | 16:46.46 |
jogux_mac | you would presume a long is numeric. | 16:47.01 |
ray_laptop | chrisl: not sure I need to, but I should probably get so I can build the release from Len (in case he comes in with a problem). What start keyword did you use ? | 16:47.05 |
Robin_Watts_ | or at least, when I did print "retval="+retval it complained that retval was a long, not a string, so I had to use str(retval) | 16:47.23 |
ray_laptop | chrisl: (since you got it to build) | 16:47.32 |
Robin_Watts_ | Anyway, builds are working for mvrhel now I think (just waiting to see if it completes). | 16:47.49 |
| I'll do a commit for that in a mo. | 16:48.02 |
chrisl | ray_laptop: I had to use "gemini_prn_es" - that's the only solution file I could see | 16:48.30 |
jogux_mac | robin : I can't claim to understand :( | 16:48.46 |
jogux_mac | does wonder why it differs for mvrhel though. | 16:49.07 |
pedro__ | different python implementations? | 16:49.25 |
mvrhel | that was very strange | 16:49.34 |
Robin_Watts_ | mvrhel: It's not that you're running a 64bit python or something? | 16:50.43 |
jogux_mac | is anyone messing with the iOS certificates etc? | 16:50.45 |
Robin_Watts_ | Where did you get python? activestate? | 16:50.55 |
mvrhel | I am running 64 bit | 16:50.57 |
jogux_mac | miles just got another 'your certificate has been revoked' message from apple :-S | 16:51.06 |
ray_laptop | chrisl: OK. that's what I tried. I get a bunch of errors: fatal error C1083: Cannot open source file: '..\..\..\gs\ufst\rts\rac\Bitio.c': No such file or directory | 16:51.11 |
mvrhel | but I don't know *anything* about pyton | 16:51.12 |
| I got it from active state | 16:51.17 |
| like written in the twiki | 16:51.29 |
Robin_Watts_ | mvrhel: I know that the number of spaces matter on indents, but all the rest of my python programming comes from copy/paste/stack overflow | 16:51.45 |
pedro__ | they're probably down to me from sunday/yesterday trying to get a good ios build going | 16:51.49 |
chrisl | ray_laptop: that's weird, that's the wrong directory..... should be "gs2" I think | 16:51.52 |
ray_laptop | in python, spaces/tabs determine blocks. | 16:52.09 |
Robin_Watts_ | mvrhel: So, the chrome remote desktop thing seems to work well. | 16:52.10 |
ray_laptop | If you need, I know python pretty well | 16:52.20 |
Robin_Watts_ | The only bad thing is that the keyboard mapping is strange. | 16:52.24 |
mvrhel | yes. I will see if I can get this set up with my father | 16:52.27 |
ray_laptop | chrisl: right | 16:52.29 |
Robin_Watts_ | My typing is not *quite* that bad usually :) | 16:52.53 |
chrisl | Robin_Watts_: when you have a minute, see if you find this acceptable: http://git.ghostscript.com/?p=user/chrisl/ghostpdl.git;a=commitdiff;h=fb11fa25 | 16:53.00 |
Robin_Watts_ | mvrhel: Let me know if the build runs through OK. | 16:53.04 |
ray_laptop | STRONGLY prefers python to perl | 16:53.05 |
mvrhel | Robin_Watts_: ok will do | 16:53.15 |
| thanks | 16:53.19 |
Robin_Watts_ | chrisl: Looks great. | 16:53.38 |
chrisl | Robin_Watts_: so just push? | 16:53.46 |
Robin_Watts_ | sure. | 16:53.58 |
chrisl | Thanks | 16:54.08 |
Robin_Watts_ | Thank you1 | 16:54.38 |
chrisl | NP, As kens said, it was probably easier for me to build the example, so..... | 16:55.07 |
mvrhel_laptop | Robin_Watts_: success | 16:56.00 |
| so how do I run this thing now... | 16:56.06 |
Robin_Watts_ | mvrhel_laptop: Woo Hoo! | 16:56.18 |
| So, towards the end of the log, hopefully you got a binary name? | 16:56.42 |
| actually, scratch that. | 16:57.12 |
ray_laptop | chrisl: strange because I see: set SIM_GS_UFST_DIR=%SIMDIR%\..\gs2\ufst in setsim.bat | 16:57.29 |
Robin_Watts_ | The Properties -> NMake -> Output field should be set, I hope? | 16:57.32 |
pedro__ | scripts/showbuild.py should help | 16:57.33 |
chrisl | ray_laptop: did you run a previous simulator in the same command prompt window? | 16:58.03 |
Robin_Watts_ | mvrhel: To: ..\genroot\windows\x86\vs2010\debug\nonet\romfss2\heap\all_das\debug_puts_enable\excel_editor\pdf_encrypt\pdf_encrypt_aes\pdf_export\search_enable\en-gb\bin\da-test.exe | 16:58.18 |
| or something like that ? | 16:58.26 |
mvrhel_laptop | yes | 16:58.28 |
Robin_Watts_ | So in the debug tab... | 16:58.38 |
chrisl | ray_laptop: I always get caught out because setsim picks up values from the environment in preference to setting them itself..... | 16:58.43 |
Robin_Watts_ | Command: $(Target) | 16:58.44 |
mvrhel_laptop | that was in the build output and in the Nmake Output field | 16:58.45 |
ray_laptop | chrisl: I just unzipped a fresh copy | 16:58.48 |
Robin_Watts_ | Command Arguments: -hq -save savedir ../../MyTests/catalog.pdf out.txt pdf | 16:58.59 |
chrisl | ray_laptop: but did you start a *new* command prompt? | 16:59.05 |
mvrhel_laptop | yes command is target | 16:59.09 |
Robin_Watts_ | Working Directory: $(SolutionDir)..\epage | 16:59.12 |
mvrhel_laptop | working dir is correct | 16:59.28 |
| command args are -pdfexport file://localhost/out.pdf ../../MyTests/prop.pdf out.txt pdf | 16:59.33 |
Robin_Watts_ | You'll want to edit ../../MyTests/catalog.pdf to point to a valid input file. | 16:59.52 |
| relative to the "epage" directory. | 17:00.02 |
mvrhel_laptop | ok | 17:00.11 |
| let me try a file | 17:00.15 |
chrisl | Robin_Watts_: there might be an easy solution to the dependency problem in jom - see the end of this file: https://qt.gitorious.org/qt-labs/jom/raw/README | 17:00.21 |
Robin_Watts_ | chrisl: .SYNC ? | 17:00.43 |
chrisl | Yeh | 17:00.47 |
Robin_Watts_ | will .SYNC not break nmake ? | 17:00.55 |
chrisl | In nmake we'll define it as a empty target | 17:01.13 |
Robin_Watts_ | chrisl: Cool. I'll get you an email for what I've got so far. | 17:01.49 |
ray_laptop | chrisl: I run it using a "shortcut" | 17:02.18 |
chrisl | Robin_Watts_: no rush - I know you're busy. I just stumbled across that README and thought "Oooh" | 17:02.25 |
Robin_Watts_ | indeed! | 17:02.32 |
ray_laptop | chrisl: so it always starts a new command prompt | 17:02.34 |
chrisl | ray_laptop: okay, scratch that idea...... I'm tempted to say chuck it back at Len and say "can we have a working sim, please"..... | 17:03.12 |
Robin_Watts_ | mvrhel: The line with -hq in should run and save some bitmaps into epage/savedir | 17:03.16 |
| The line with -pdfexport should produce you an epage/out.pdf | 17:03.34 |
chrisl | ray_laptop: tbh, I have these kinds of problems every time I get a sim update, which is why I dread it so much :-( | 17:03.44 |
ray_laptop | chrisl: Ahh. I found it. A typo in my shortcut. | 17:04.02 |
| chrisl: yeah, me too | 17:04.13 |
Robin_Watts_ | The 'test-shell' project is pretty much the simplest build we can do that does document loading/rendering/pdf export | 17:04.14 |
| as such it's going to be one of the easiest ways to try hacking^H^H^H^H^H^H^Hworking on the internals of smart office. | 17:04.49 |
| If you want a proper app, then you'll need to build one of the other targets. | 17:05.23 |
chrisl | ray_laptop: but even once it builds, it just crashed.... | 17:05.54 |
ray_laptop | chrisl: HURRAY. Len just confirmed the performance improvements with your latest patches :-) | 17:15.43 |
Robin_Watts_ | ray_laptop: So we're up to the required speed now ? | 17:16.00 |
chrisl | ray_laptop: I just saw the mail - YAY!! | 17:16.06 |
Robin_Watts_ | Congratulations, to both of you. | 17:16.18 |
ray_laptop | Robin_Watts_: we actually do better than the 871 times | 17:16.24 |
| Robin_Watts_: however, as I mentioned to Len, the font cache improvements would improve 871 as well. | 17:16.59 |
| and since we are only 2 seconds faster with 9.06 compared to 8.71 ... | 17:17.34 |
Robin_Watts_ | Don't give him an excuse to stay with an accelerated 8.71 :) | 17:18.00 |
ray_laptop | Robin_Watts_: no, they are committed to going forward with 9.x | 17:18.24 |
| but they still have 8.71 products in the mix | 17:18.42 |
| Robin_Watts_: I just don't know if they have any NEW 8.71 products in the pipeline | 17:19.09 |
| we should request a list of products shipping and in development | 17:19.57 |
Robin_Watts_ | so gsvg has been eaten? | 17:22.05 |
chrisl | Robin_Watts_: we killed gsvg as soon as tor8 had made a start on the mupdf version | 17:23.06 |
Robin_Watts_ | will remove the project from the solution then. | 17:23.45 |
chrisl | Oh, I thought I did that - sorry...... | 17:24.00 |
ray_laptop | chrisl: Len mentioned that he isn't concerned about the 'instability' of this version of the simulator (he just called). It was just one he threw together for us to work with. He'll give us another "cleaner" version in the future, once the dust settles | 17:30.16 |
chrisl | ray_laptop: I assumed it was something of the sort - still, a bit frustrating..... | 17:31.39 |
ray_laptop | chrisl: agreed. | 17:32.03 |
chrisl | Right, I'm off to probably lose a squash match - 'nite all | 17:33.09 |
ray_laptop | chrisl: I've given up trying to get them to fix the 'wheel.exe' viewer -- It doesn't let me see the whole page | 17:33.20 |
| chrisl_away: have a good time | 17:33.29 |
Guest69874 | Hello, I dropped by to say thank you for the MUPDF 1.4 release. However the rendering speed compared to previous 1.3 release on identical files has a ~15~20% loss. I have tested large batches of files, with hundreds of originals with smallformat and largeformat (drawings) content. Maybe for a future release someone is interested in investigating what caused the rendering performance loss. | 17:37.06 |
Robin_Watts_ | Guest69874: We'd certainly be interested in knowing why it has got slower. | 17:38.20 |
| Whether it's something specific to your machine/setup or something more general. | 17:38.38 |
| What platform are you on? | 17:38.45 |
Guest69874 | windows 32bit | 17:38.54 |
ray_laptop | Robin_Watts_: you beat me to it | 17:38.56 |
| Guest69874: can you send any files which are particularly slower ? | 17:39.21 |
Robin_Watts_ | Guest69874: Are you building your own binaries, or using the canned ones ? | 17:39.22 |
Guest69874 | I am calling 'mudraw' from the command line, using the original supplied exe (binary). | 17:40.04 |
Robin_Watts_ | Guest69874: Usually I do the builds, but this time tor did them. I wonder if there is a compiler difference. | 17:40.41 |
| Guest69874: Just give me a mo. | 17:41.08 |
ray_laptop | Robin_Watts_: well, it should be easy to compare the pre-built to what you have | 17:41.24 |
Robin_Watts_ | ray_laptop: I'm just doing a build, and I'm going to upload it for Guest69874 to try. | 17:42.00 |
ray_laptop | Robin_Watts_: oh, that makes sense, too | 17:42.14 |
| Robin_Watts_: do you build with 2005 or something else ? | 17:43.28 |
Robin_Watts_ | Guest69874: http://ghostscript.com/~robin/mudraw.exe | 17:44.12 |
| ray_laptop: VS2005. | 17:44.17 |
Guest69874 | testing now, one moment please, I have a short & long test batch | 17:44.48 |
ray_laptop | Robin_Watts_: do you build with 2005 or something else ? | 17:46.21 |
Guest69874 | my corporate AV is blocking exe downloads, are you able to zip or rar that please ? I cannot disable the AV on this station | 17:47.59 |
Robin_Watts_ | ray_laptop: VS2005. | 17:48.04 |
| Guest69874: http://ghostscript.com/~robin/mudraw.not_an_exe_honest | 17:48.48 |
ray_laptop | Robin_Watts_: does that _really_ fool browsers that restrict downloads ? | 17:50.04 |
sebras | tor8: Robin_Watts_: two silly patches over at sebras/master and this one for curl http://pastebin.com/wCgMpJmF | 17:50.24 |
Robin_Watts_ | we'll see in a minute. | 17:50.26 |
| sebras: Will look as soon as I can. Desperately trying to finish something atm. | 17:50.52 |
sebras | Robin_Watts_: ok. I'll stop nagging. :) | 17:51.14 |
Guest69874 | sorry , Symantec AV is too smart, it deletes the EXE despite the extension change, I cannot download from your link | 17:54.12 |
| can you zip or rar please | 17:54.44 |
Robin_Watts_ | Guest69874: http://ghostscript.com/~robin/mudraw.zip | 17:55.13 |
ray_laptop | Guest69874: I'm surprised it lets .zip through unless it has a password. gmail looks inside .zips and won't let us send them as attachments if there is an exe in it | 17:56.28 |
Robin_Watts_ | ray_laptop: shh! It might hear you. | 17:56.47 |
ray_laptop | I usually just use the secret password "secret" | 17:57.32 |
| grrr... the fast thresholding image code is going to take a lot of work to get cleaned up. The methods it uses to fill the contone buffers with scaled image data are totally wrong :-( | 17:59.26 |
| I have one case working, however. Now it's just "make work" | 17:59.49 |
mvrhel_laptop | ray_laptop: is this something that I did wrong? | 18:03.57 |
Hakan_AG | sorry, back after testing...my connect dropped. the last 'mudraw.exe' that was supplied is even slower then the original 1.4 release :( | 18:04.37 |
| mudraw 1.3 = 5 secs - mudraw 1.4 (experiemental) = 12 sec | 18:05.46 |
| mudraw 1.4 (original release) 7 sec | 18:05.56 |
mvrhel_laptop | Robin_Watts_: you still here? | 18:06.19 |
Hakan_AG | 62 pages at 8.5x11, 36dpi render to PNG on 2 threads | 18:06.20 |
Robin_Watts_ | What are the binary sizes for 1.3 and 1.4 please? | 18:06.20 |
| mvrhel: I am. | 18:06.32 |
Hakan_AG | 1.3 = 8,503,296 bytes | 18:06.50 |
| 1.4 official = 8,589,312 | 18:07.19 |
mvrhel_laptop | so what is the syntax supposed to be if I want to render a ppt file. I am guessing that -pdfexport is setting up to convert to pdf? | 18:07.21 |
Robin_Watts_ | You want to render a ppt to bitmaps? | 18:07.45 |
mvrhel_laptop | Robin_Watts_: yes | 18:07.52 |
| I am confused by the out.txt and "pdf" that you had on the command line | 18:08.15 |
| at the end | 18:08.21 |
Robin_Watts_ | Command Arguments: -hq -save savedir ../../MyTests/input.ppt out.txt ppt | 18:08.23 |
| -hq says "do hq renders" (i.e don't just layout etc) | 18:08.45 |
mvrhel_laptop | hq? | 18:08.58 |
Robin_Watts_ | -save blah says "save the output into blah" | 18:08.58 |
mvrhel_laptop | high quality? | 18:09.04 |
Robin_Watts_ | High Quality as opposed to Fuzzy. | 18:09.08 |
| Fuzzy being a scaled up thumbnail. | 18:09.24 |
mvrhel_laptop | ok | 18:09.28 |
Robin_Watts_ | Next is the input file. | 18:09.34 |
mvrhel_laptop | ok | 18:09.41 |
Robin_Watts_ | Then a file for various output stuff to be dumped too (typically includes a dump of the laid out objects) | 18:10.12 |
mvrhel_laptop | ok | 18:10.19 |
Robin_Watts_ | Then the format to use. | 18:10.20 |
mvrhel_laptop | ah ok | 18:10.23 |
| and -pdfexport does the layout into pdf | 18:10.51 |
Robin_Watts_ | Yes. | 18:10.56 |
ray_laptop | mvrhel: sorry - I was working on it. But, yes. it probably is. The dda logic was picking the wrong pixels, and sometimes the wrong number of pixels was generated. | 18:10.58 |
Robin_Watts_ | -pdfexport is followed by the file to export to. | 18:11.14 |
mvrhel_laptop | ok thanks Robin_Watts_ | 18:11.30 |
| ray_laptop: I am surprised to hear that that part is wrong. I did quite a bit of testing with it | 18:11.59 |
Robin_Watts_ | Hakan_AG: OK, so I'm at a loss to immediately say why. | 18:17.23 |
Hakan_AG | RobinWatts : I have to leave, if you require real world test files to test when you have time, pls ask me at hakan.usakli@astleygilbert.com. thank you. version 1.4 is definately somewhat slower. | 18:17.58 |
Robin_Watts_ | Hakan_AG: Could you find the file that's most obviously slowed down, and open a bug at bugs.ghostscript.com and attach it there please? | 18:18.16 |
| That way it won't get forgotten. | 18:18.21 |
| At the moment, I am busy with other things, but I would like to fix this. | 18:18.34 |
Hakan_AG | Understood, I will do that tomorrow and try to find a good test file. | 18:18.46 |
Robin_Watts_ | Thanks. | 18:18.50 |
mvrhel_laptop | Robin_Watts_: ok that worked thank you | 18:19.30 |
Robin_Watts_ | mvrhel_laptop: Phew. | 18:19.39 |
mvrhel_laptop | so I am up and running now | 18:19.41 |
Robin_Watts_ | mvrhel_laptop: That's great news. | 18:19.51 |
mvrhel_laptop | I might poke a bit at one of these rendering issues on the artifex power point to see if I can see anything obvious | 18:20.21 |
| just to get a taste of what this code is like.. | 18:21.01 |
Robin_Watts_ | mvrhel_laptop: Go for it. | 18:22.07 |
| sebras: Your 2 mupdf commits look fine. | 18:23.07 |
| The curl one... is the patch reversed? | 18:23.52 |
ray_laptop | mvrhel: It's not very visible. I'm comparing the 4-bit to the 8-bit,visually, and out of 1500 files I reviewed, I found about 20 "objectionable" differences. Those are the ones I'm looking into. | 18:24.34 |
Robin_Watts_ | sebras: No, I get it, sorry. | 18:25.30 |
mvrhel_laptop | ray_laptop: so the 8 bit and the 4 bit both look bad? | 18:25.50 |
| oh never mind | 18:25.59 |
| I see what you mean | 18:26.08 |
| the 1 bit and the 4 bit will both look bad | 18:26.19 |
| I suspect | 18:26.24 |
ray_laptop | mvrhel: it's almost impossible to see with 1-bit halftoned since the halftoning "hides" the differences a lot | 18:27.40 |
mvrhel_laptop | ok that would explain why I did not see it then | 18:27.54 |
Robin_Watts_ | sebras: Pushed. Thanks! | 18:30.30 |
| henrys: Want to review this for me? http://git.ghostscript.com/?p=user/robin/ghostpdl.git;a=commitdiff;h=a33e6e4084bc77a8c1f94500576fada62e6e2077 | 18:32.09 |
ray_laptop | Robin_Watts_: the UFST_ROOT is really set to C:\ufst ??? I know this is how it used to be, but shouldn't it be somewhere else ? | 18:40.55 |
Robin_Watts_ | ray_laptop: That's all explained in the ReadMe. | 18:41.57 |
| People using the ufst are expected to edit that to suit their setup. | 18:42.14 |
| It's not one of the 'standard' projects. | 18:42.28 |
sebras | Robin_Watts_: thanks for reviewing. :) | 18:42.40 |
Robin_Watts_ | sebras: No problem. Thanks for fixing my code :) | 18:42.53 |
sebras | Robin_Watts_: I don't think error handling in this case warrants a semaphore and that kind of locking overhead. | 18:48.51 |
| especially since content_length is not protected either. | 18:49.05 |
mvrhel_laptop | and that is probably enough SOT for today. head already hurts | 19:13.25 |
| lunch time | 19:13.29 |
Robin_Watts_ | mvrhel_laptop: hehe. | 19:13.51 |
mvrhel_laptop | Robin_Watts_: perhaps you can give me a minor tour of things one day... | 19:13.55 |
Robin_Watts_ | mvrhel_laptop: I can do my best. | 19:14.12 |
mvrhel_laptop | ok that would be great. maybe tomorrow for a tad bit | 19:14.46 |
Robin_Watts_ | I know the rendering side of it all quite well. The layout side less so - Jogu and Pedro have more experience in that area. | 19:15.02 |
| but I'll gladly share my confusion. | 19:15.16 |
mvrhel_laptop | ok . I am looking at some rendering issues in how the code is handling images in the ppt file. I have a single page with a single image. I had hoped it would be easy to understand but I fear I am getting lost in the edr stuff | 19:15.57 |
| I also probably need to understand a bit about the format of the files too though. something that I don't know at all | 19:16.45 |
| is there any documentation anyplace? | 19:17.19 |
| Robin_Watts_: ^^ | 19:17.21 |
| I mean even documentation at the level that we have for ghostscript would be helpful | 19:17.44 |
Robin_Watts_ | mvrhel_laptop: Look at the ghostscript regression dashboard. | 19:18.26 |
| There is an 'epage-docs' link in the top right section. | 19:18.41 |
mvrhel_laptop | aha | 19:19.00 |
Robin_Watts_ | In the edr directory there is an "edr-training-course.ppt" | 19:19.09 |
| I'm going to spend some time looking at that I think :) | 19:19.19 |
| mvrhel_laptop: By the way, when I mentioned VisualGDB earlier, I was thinking that you'd be interested in it. | 19:19.48 |
mvrhel_laptop | Robin_Watts_: oh this will be helpful | 19:20.09 |
| hmm may want a local copy of these docs. very laggy for some reason | 19:21.29 |
| browsing through them is bringing firefox to its knees for some reason | 19:22.30 |
| ok off to eat. bbiaw | 19:22.52 |
henrys | Robin_Watts_: I'm good with the patch | 19:51.36 |
jogux | mvrhel_laptop : the da testshell dumps out a textual representation of the edr and display list to the output file - understanding that is a good place to start figuring out if the problem is the DA not putting the right thing into the edr, or layout losing it / putting it in the wrong place. | 20:05.22 |
| [the windows application, if build with -debug, also has some similar functionality] | 20:05.45 |
| emobix has some training courses that cover much of the basics; it might be worth us presenting those to you sometime (subject to Miles/Henry agreeing). To some extent they may duplicate what you can find in epage-docs. | 20:07.05 |
mvrhel_laptop | jogux: great thanks for the suggestions | 21:44.45 |
Robin_Watts_ | jogux, pedro, paulgardiner: (For the logs) 2 commits on robin/master. The python workaround for mkromfss2.py that michael needed and an updated solution. | 23:07.46 |
| henrys: Thanks. | 23:07.52 |
| jogux, pedro: AFAICT the AndroidManifest.xml for the SOT doesn't have the debuggable flag set. which is a bit of a pain. | 23:08.53 |
| I might look at updating it so that its set for debug builds at least. | 23:09.18 |
| Forward 1 day (to 2014/04/23)>>> | |