| <<<Back 1 day (to 2015/10/19) | 20151020 |
Jaikumar | Is there a way to use ghostscript without installing it? I'm using Magick.NET to convert .PS to .PNG | 08:01.43 |
kens | Same answer as yesterday | 08:01.53 |
tor8 | Robin_Watts: is there an e-mail list subscription web page for the artifex mailing lists? | 10:12.38 |
Robin_Watts | tor8: Which artifex mailing lists? | 10:13.04 |
tor8 | I keep getting duplicate mails for some bug reports, and gmail makes them a pain to read and filter | 10:13.09 |
Robin_Watts | The gs- ones have a webcontrol. | 10:13.19 |
tor8 | and I'm not really interested in the so-commits list | 10:13.22 |
Robin_Watts | You'll need to talk to marcos. | 10:13.34 |
tor8 | it looks like bugzilla comes through two mailing lists, gs-bugs@ghostscript.com and mudpf-bugs@artifex.com | 10:13.53 |
Robin_Watts | Personally, I find it REALLY annoying that I get 3 copies of every frigging bug email. | 10:13.56 |
kens | You get 2 copies if you are assigned to the bug and get gs-bugs | 10:14.14 |
tor8 | I tried deleting the duplicates from one, but that takes down the whole thread :( | 10:14.20 |
kens | I don't know how you get 3 | 10:14.21 |
tor8 | kens: I get three as well, one from gs-bugs, one from mupdf-bugs, and one because I'm on the bug | 10:14.43 |
Robin_Watts | All new mupdf bugs are assigned to mupdf-bugs@artifex.com, and then when they get reassigned out mupdf-bugs remains CC'd. so gets copies. | 10:15.06 |
tor8 | I'd be more than happy to have just he one from gs-bugs and then flag it as important based on my name in a filter | 10:15.08 |
kens | Oh 3 for MuPDF. I guess MuPDF bugs shouldn't be triggering gs-bugs | 10:15.09 |
tor8 | kens: I think it's because of the default assigned email address is mupdf-bugs@artifex.com | 10:15.45 |
| so I'd like to unsubscribe from that list | 10:16.24 |
| Robin_Watts: you can uncheck so you don't get mails from when you're on the bug though, in the email preferences in bugzilla | 10:16.47 |
kens | Hmm I didn't know that | 10:17.39 |
chrisl | I'd have to wonder what the point of mupdf-bugs is if tor8 wasn't subscribed to it..... | 10:37.11 |
Robin_Watts | chrisl: I have to wonder what the point of having bugzilla mail us all multiple times about every bloody bug is. All it means is that I'm much more likely to ignore a genuine bug as I skip loads of posts on the grounds like they look like ones I've read already. | 11:07.08 |
chrisl | Robin_Watts: well, I filter the mails, so I don't see it as quite the problem | 11:18.29 |
| Although, I'll admit I'll happily forego the SOT mailing lists because, from point of view, they're just wasting space.... | 11:21.58 |
henrys | well we can kick around subscriptions/mailing lists at the meeting, but like chris I filter them so it doesn't bother me much. | 12:50.58 |
kens | hah a henrys | 12:51.29 |
| Henry, your suggestion of reset works perfectly for PCL | 12:51.42 |
| But not at all for PXL | 12:51.49 |
| PXL has hte PJL handling in BeginSeesion, whch gets called for exevery page | 12:52.36 |
| Or at least every page in the test file I'm using | 12:52.51 |
henrys | I thought we agreed to always run the pcl startup to set your stuff in motion and not fool with XL at all | 12:52.52 |
kens | That doesn't work if the input file is PXL | 12:53.08 |
| Because the PCL startup stuff doesn't get called | 12:53.33 |
| Or to be more precise, do_reset doesn't | 12:53.53 |
henrys | I was going to change that for your stuff and a couple other reasons | 12:53.55 |
kens | Ah OK | 12:54.00 |
| SO if I just leave it out of PXL it'll all magically work later :-) | 12:54.18 |
| I'll do that then | 12:54.29 |
henrys | yes in the new system PCL/XL are going to look like 1 language. Not so much for your stuff but it will let me clean up passthrough mode. Passthrough really threw a wrench in the design we had. Very stupid of HP to do that... | 12:56.37 |
kens | THat's absolutely fine, suits me down to the ground | 12:56.59 |
| And yeah passthrough is a total kludge | 12:57.10 |
| I'm assumign some big customer demanded it | 12:57.18 |
henrys | yeah HP engineers aren't stupid, I can imagine the dropped jaws when that was required. | 12:58.38 |
kens | I can imagine what was said during the meetings :-) | 12:59.02 |
henrys | I'm worried my commit mistake last night will break bisecting. I suspect that commit failed because it ran out of time, we could remove the commit and all resync. | 13:10.50 |
| meh it should be okay, it won't slow down any single job too much. | 13:16.49 |
Robin_Watts | henrys; https://www.indiegogo.com/projects/revolights-eclipse-connected-bike-lighting-system/x/1895095#/ | 13:18.45 |
henrys | oh that is nice... | 13:19.56 |
Robin_Watts | henrys: They might do a bulk price if you do all your bikes :) | 13:20.55 |
henrys | I love the turn signals | 13:21.53 |
| road riders who care about performance probably won't touch that though, great for commuters, anything added to the wheel is not going to fly for serious riders | 13:24.43 |
| ah you're putting that on my $3000.00 wheel? not... | 13:25.39 |
Robin_Watts | henrys: I would have thought that if you were adding weight anywhere, the wheel rim was the place for it. | 13:31.50 |
rayjj | Robin_Watts: it's still weight you have to haul up hills | 13:46.28 |
| like the back packers that cut 2/3 of the handle off their toothbrush | 13:47.31 |
Robin_Watts | rayjj: You're thinking of prisoners. | 13:51.41 |
kens | Henrys I've made a bunch of changes on my repository to handle the pdfmarks according to a reset, removed code no longer required and so on. | 13:56.47 |
| http://git.ghostscript.com/?p=user/ken/ghostpdl.git;a=shortlog;h=refs/heads/dynamic_PJL_allocs | 13:57.01 |
| Top 4 commits there, you can ignore the first one though | 13:57.16 |
kens | thinks I should probably rename my branch...... | 13:59.36 |
henrys | Robin_Watts, rayjj: we love light wheels here's a simple explanation: https://van.physics.illinois.edu/qa/listing.php?id=7559 | 14:00.19 |
| and there going to foul up the aerodynamics of the wheel with that thing. | 14:00.39 |
| s/there/they're | 14:01.49 |
Robin_Watts | henrys: I am dubious about his explaination. | 14:04.44 |
| explanation, even. | 14:04.55 |
| The wheels are flywheels, so clearly adding weight on the rims makes them harder to spin up. | 14:05.59 |
| They also make it more resistant to spin down of course. | 14:06.20 |
| I guess it only makes a difference when you start accelerating/braking. For constant speed travel in a straight line, it should make no odds. | 14:07.43 |
henrys | Robin_Watts: if you are competitive or group riding here you are accelerating all the time. | 14:09.55 |
| that's why I said it doesn't matter to a commuter ... | 14:10.46 |
Robin_Watts | fair enough. | 14:11.24 |
chrisl | Hmm, even to a commuter - staying with traffic etc, means a lot of accelerating a braking | 14:11.25 |
| in the city | 14:11.31 |
Robin_Watts | weight of that set of lights vs weight of the sandwiches in your backpack... for a commuter it's probably uninportant. | 14:11.59 |
| (unless you are a strava fiend like Jogu) | 14:12.09 |
henrys | everybody here is on strava I'm one of the few holdouts... just use garmin | 14:17.13 |
tor8 | Robin_Watts: having all the weight at the rims is going to add a fair amount of extra momentum to overcome when accelerating/braking | 14:18.07 |
| 0.6kg extra going up a hill, not really bothered :) | 14:18.35 |
paulgardiner | Interesting. Had no idea that the wheels counted twice. Neat. | 14:18.47 |
tor8 | 0.6kg of mass that needs to be spun up and down all the time, more bothered. | 14:18.48 |
Robin_Watts | tor8: Yeah. For racing bikes, I can see that. For commuter bikes, I think you're overestimating the weight of the lights. | 14:18.51 |
tor8 | Robin_Watts: depends on how many red lights you have on the commute :) | 14:19.12 |
| the new version has the batteries on the rim as well... | 14:19.35 |
Robin_Watts | And 'not being dead' at the end of the commute counts for something. | 14:20.02 |
henrys | yeah that death will really slow you down much more than the heavy wheels | 14:20.45 |
Robin_Watts | cycled in Oxford for 6 years. The sheer weight of numbers of bikes there means that cars are more attuned to them, and even so car/bike interactions are very common. | 14:21.29 |
| (Because both car drivers and cyclists are fallible/assholes). | 14:22.07 |
chrisl | Even more so now, it seems..... | 14:22.39 |
Robin_Watts | Cyclists in other cities are going to be even more vulnerable. I wouldn't fancy it. | 14:22.47 |
tor8 | you need to take a cue from holland, and put bike lanes/roads everywhere | 14:23.54 |
Robin_Watts | oh, meeting. | 14:26.36 |
| I'd forgotten. | 14:26.42 |
fredross-perry | morning | 14:26.45 |
Robin_Watts | fetches drink. | 14:26.53 |
henrys | paulgardiner: you might actually enjoy this http://www.amazon.com/Bicycling-Science-David-Gordon-Wilson/dp/0262731541 ... there are lots of surprising results and it's an interesting read history wise | 14:27.20 |
| tor8, mvrhel_laptop: so what do we know about openxps? What are we missing in gh and mupdf? | 14:29.54 |
| tor8, mvrhel_laptop : I assume you guys read the customer request? | 14:30.23 |
kens | I thought MuPDF worked | 14:30.30 |
Robin_Watts | My understanding was that mupdf supported openxps. | 14:30.32 |
tor8 | henrys: just a couple of extra strings to use when looking for the "root" path name in the zip file | 14:30.40 |
mvrhel_laptop | I did. henrys, I think mupdf already does oxps | 14:30.43 |
| that should be easy to do | 14:30.54 |
Robin_Watts | tor8: Ray made that change, and it still didn't work. | 14:31.04 |
mvrhel_laptop | I can take a look today if you want | 14:31.23 |
kens | maybe that's not strictly an OXPS problem as a fault with GS's XPS handling | 14:31.31 |
henrys | mvrhel_laptop: okay | 14:31.39 |
| tor8: are we sure that's all there is to it? | 14:31.47 |
kens | Well nobody has spent much tmie on it yet | 14:32.12 |
tor8 | http://git.ghostscript.com/?p=mupdf.git;a=commitdiff;h=b6519f17bf729617279e1feb78b08ce4fd56cd5b | 14:32.13 |
| that's the commit for mupdf | 14:32.17 |
mvrhel_laptop | tor8: what about jpegxr? | 14:32.18 |
tor8 | mvrhel_laptop: we don't do jpegxr in mupdf | 14:32.55 |
chrisl | tor8: with that change added to gxps, you just get another error further on | 14:32.57 |
tor8 | there's a patch on sebras branch to add it, but the jpegxr code is a bit nightmarish and I worry about the security implications | 14:33.11 |
| chrisl: ah. | 14:33.19 |
mvrhel_laptop | ok. I doubt there is much oxps content with jpegxr | 14:33.24 |
kens | chrisl I wonder if that's not strictly an OXPS problem so much as a general problem with the GS XPS implementation | 14:33.26 |
| I mean MuPDF might have fixed that as a separate issue | 14:33.47 |
chrisl | kens: could be, I don't know enough about XPS to guess | 14:33.48 |
kens | No me neither, that's what I meant when I said nobody had spent much ime on the problem yet | 14:34.04 |
chrisl | I did day there seemed to have been more divergence between the two that I thought was ideal | 14:34.26 |
| s/day/say | 14:34.35 |
kens | Indeed, that was (obliquely) what I was referring to | 14:34.39 |
henrys | mvrhel_laptop: is there a document for the change from MS? | 14:35.05 |
mvrhel_laptop | henrys: I think I saw something like that at one point. I will see if I can dig it up | 14:35.26 |
henrys | mvrhel_laptop: it would be good to be certain before we respond we have it working | 14:36.03 |
paulgardiner | henrys: oh right. thanks | 14:36.07 |
mvrhel_laptop | henrys: yes I agree. I will find out | 14:36.42 |
paulgardiner | henrys: any idea whether that reads okay on a Kindle (diagrams and all)? | 14:36.53 |
henrys | so we want marcosw to tweek the mailing lists? | 14:37.22 |
| paulgardiner: yeah not sure I have the book. | 14:38.30 |
| so submit mailling list requests to marcos and I and on of us will sort it out. Drop me from blah blah.. etc. | 14:39.49 |
| s/on/one | 14:39.55 |
| tor8, Robin_Watts mupdf release? | 14:40.15 |
| any bugs on the report or aging list that we should talk about? | 14:41.07 |
Robin_Watts | Nothing stopping the mupdf release as far as I know (except possibly simon91's bug report from yesterday?) | 14:41.31 |
| bug report + patches, that is. | 14:41.45 |
| Tor was talking about doing an rc yesterday just before those came in. | 14:42.12 |
tor8 | the new bug report from yesterday regarding digital signatures is the only thing that's a potential hold up now. would paulgardiner be available to fix that issue? | 14:43.41 |
henrys | tor8: is that a regression? | 14:44.12 |
simon91 | tor8: have a fix ready | 14:44.18 |
| tor8: this moment. will upload soon. only 4 lines. | 14:45.08 |
| :) | 14:45.26 |
Robin_Watts | simon91: Thanks! | 14:45.38 |
henrys | simon91: great - hey nice to have you here on IRC! | 14:45.53 |
paulgardiner | I haven't seen the bug yet. Got that number to hand? | 14:45.57 |
simon91 | henrys: yes was a regression. worked in 1.7 | 14:46.04 |
henrys | I guess paulgardiner can review it. | 14:46.40 |
| I'm fine finishing up early unless folks have issues to raise? | 14:46.56 |
simon91 | paulgardiner: report was http://bugs.ghostscript.com/show_bug.cgi?id=696284 | 14:47.44 |
| paulgardiner: Iemme amend the commit message and i will attach the patch to this report. | 14:48.23 |
| paulgardiner: though when i ran it through valgrind i spotted several little mem leaks in pdf-pkcs7.c. like a BIO not getting freed and sutch | 14:49.47 |
henrys | simon91: actually I though you were a different simon, we have another contributor of that name, but nonetheless welcome and thanks for working on mupdf. | 14:50.02 |
Robin_Watts | simon91: FWIW, you can check for leaks without using valgrind; if you do "make build=memento", that will build you a version that uses our Memento lib to check for memory overwrites/leaks etc. | 14:51.33 |
rayjj | mvrhel: Any word from Michael C (company M) ? | 14:51.46 |
mvrhel_laptop | rayjj: so Michael C is at a different company | 14:52.09 |
| I did sent MQ an email yesterday. He may still be out of the country | 14:52.23 |
rayjj | mvrhel: sorry I meant MQ | 14:52.25 |
mvrhel_laptop | s/sent/send/ | 14:52.36 |
| It may be another week before he gets back I think | 14:52.52 |
rayjj | mvrhel: I'm a bit concerned that he might just play with building it himself, get some results and go forth without or help/involvement/knowledge_of_results | 14:53.39 |
paulgardiner | simon91: doesn't surprise me at all. pdf-pkcs7.c, if I remember correctly, isn't using openssl's outer API. It's a bit of their source code altered to do what we need. | 14:54.28 |
rayjj | mvrhel: so the procurement of boards and updating the agreement, etc., are all going through MQ ? | 14:54.32 |
henrys | skype meeting in 5 minutes | 14:54.43 |
rayjj | mvrhel: this project comes to mind every time I get an email from company M's extranet with an update of some "ALERT" (usually sales unk) | 14:55.51 |
simon91 | Robin_Watts: I know. But with memento I don't spot stuff like http://bugs.ghostscript.com/show_bug.cgi?id=696259 | 14:55.52 |
mvrhel_laptop | rayjj: MQ is going to get the boards etc. yes | 14:59.14 |
rayjj | mvrhel: OK. Thanks. | 14:59.34 |
mvrhel_laptop | rayjj: My plan is to go over and see what he has done once he gets back | 14:59.35 |
Robin_Watts | simon91: Absolutely. valgrind is brilliant for stuff like that. | 14:59.54 |
| but for leaks, I frequently find memento more use than valgrind, as it is easier (or at least I find it easier) to use interactively with a debugger to track things down. | 15:00.37 |
simon91 | Robin_Watts: yeah and you don't have to clutter your sources with memonto_label | 15:00.47 |
Robin_Watts | But then, I am not an impartial observer :) | 15:00.47 |
| You don't *have* to clutter your sources with Memento_label. | 15:01.09 |
simon91 | memento_label, even | 15:01.09 |
henrys | valgrind can be used interactively with a debugger. | 15:01.22 |
mvrhel_laptop | bbiab | 15:01.31 |
Robin_Watts | henrys: Not as usefully as Memento. | 15:01.39 |
rayjj | kens: BTW, I looked at your comment on bug 696257. I agree that the correct way to process this is to have the mask and image processed in bands, post-clist. | 15:02.04 |
Robin_Watts | With Memento I can rerun and stop on particular allocations etc and step through BEFORE problems occur. Can't do that with valgrind. | 15:02.17 |
kens | rayjj there might be a better way | 15:02.28 |
Robin_Watts | (I am not dissing valgrind here. It's a massively useful, really clever tool) | 15:02.44 |
rayjj | kens: I'll add some thoughts on the approaches (so I don't forget). | 15:02.44 |
kens | chrisl had some thoughts also | 15:02.51 |
chrisl | Ugh, that means trying to remember..... | 15:03.14 |
kens | The one thing for certain is that requiring a full aimge buffer isn't going to be feasible | 15:03.22 |
| I should say a full image buffer at output resolution..... | 15:04.17 |
simon91 | Robin_Watts: henrys: Currently I'm running all of the test of my pdfout tool under valgrind. This way I found several mupdf bugs. Though really slow. without valgrind all my tests run in 0.5 sec. with valgrind it takes over 1 minute. | 15:05.10 |
| Robin_Watts: henrys: this won't scale for long, so I'm now refactoring these tests into bigger processes. | 15:07.07 |
| Robin_Watts: henrys: tor8: Think that in the long run mupdf will also need a test suite with good code coverage. | 15:09.41 |
Robin_Watts | simon91: We have a test cluster. | 15:10.29 |
simon91 | Robin_Watts: but not in the git repo? | 15:10.49 |
Robin_Watts | Whenever we make a commit (or run a test manually) it runs 18000 jobs through mupdf. | 15:10.51 |
simon91 | Robin_Watts: I see. | 15:11.20 |
Robin_Watts | It looks for rendering changes. | 15:11.25 |
| I think marcosw runs periodic leak checks too. | 15:11.37 |
| The cluster code we use is in a different repo. | 15:11.58 |
| The test suite is split into public and private sections. | 15:12.12 |
simon91 | Robin_Watts: but you don't yet seem to cover the pdf creation and pdf writing API? | 15:12.34 |
Robin_Watts | but the cluster as a whole is private to Artifex (cos there are test files from customers, and commercial test suite companies). | 15:12.38 |
| simon91: Not at all. | 15:12.42 |
| That's all bleeding edge unfinished code. | 15:12.55 |
| You're surfing the breaking wave of progress ;) | 15:13.31 |
simon91 | tor8: Robin_Watts: did you look already at http://bugs.ghostscript.com/show_bug.cgi?id=696260. (the pdf_write_to_file API) | 15:15.01 |
Robin_Watts | and sadly, my attention has been coopted onto another project :( | 15:15.25 |
simon91 | tor8: would be glad to have the FILE* API in this release if that's possible. | 15:16.13 |
henrys | everytime I hear all this noise about test driven development I think about this talk by Rich Hickey, if you have a few minutes it's worth a listen, it's all obvious stuff that all of us sometimes forget about (well at least I do): https://www.youtube.com/watch?v=f84n5oFoZBc | 15:29.28 |
simon91 | paulgardiner: here it is: http://bugs.ghostscript.com/attachment.cgi?id=11994&action=diff | 15:31.07 |
simon91 | needs to do his homework now. CU | 15:33.31 |
paulgardiner | simon91: Thanks | 15:33.44 |
| That looks good to me | 15:37.09 |
simon91 | paulgardiner: it's basically the alternative way of fixing pdf_create_document I suggested in http://bugs.ghostscript.com/show_bug.cgi?id=696251 | 15:39.04 |
henrys | for the off season I'm trying out cross fit - I hope these brutes don't step on my foot or something, look at this girl: http://www.crossfit.com/ paulgardiner is going to like her ;-) ... off to my first class | 16:37.09 |
Robin_Watts | henrys: Helen has a friend who does that. NUTTER. | 16:38.52 |
henrys | I don't think I'll last much more than a month doing it, but I don't do much strength stuff during the tri season so... | 16:39.44 |
paulgardiner | henrys: oh yes, she'd beat me in a fight... maybe a bit too easily! | 18:03.06 |
yitzchok | Hi Ghostscript people. Hope you can help a bit. I have a problem similar to that discussed at https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=522417 . I have an HL-2250DN and printed fine via CUPS for a couple of years. In recent months I did a version update (from the previous Ubuntu LTS to the current LTS) and it seems to me that since then printing has been intermittent. I thought it was connection related but it does seem t | 19:18.15 |
| o depend on the file being printed and I narrowed it down to Ghostscript and found the bug report I mentioned. I don't know for sure what version I was using before the upgrade (which worked) but the new version was 9.10. Yesterday I tried 9.15 and 9.05 and the results were the same. Is there a known current issue with the bit device? Could it be that the problem is caused by a related library which I didn't upgrade or downgrade? | 19:18.16 |
| Would welcome any guidance. Many thanks | 19:18.16 |
Robin_Watts | Well, that must have taken him longer to type than he hang around for an answer. | 19:36.06 |
rayjj | Robin_Watts: maybe he expects an answer "for the logs" (not likely) | 20:40.05 |
tcarmen | Anybody have a second to toss me a clue re: mswinpr2? | 20:51.28 |
| I've been trying to send output to a windows print queue for the last 5 or 6 hours, and it works, but always pops up the "select printer" dialog even when I specify a valid queue. Any ideas? | 20:52.51 |
faLUCE | hello. How can I crop a pdf vectorial image by trimming only top and bottom? | 21:27.32 |
| Forward 1 day (to 2015/10/21)>>> | |