| <<<Back 1 day (to 2012/05/29) | 2012/05/30 |
henrys | marcosw_ for the logs - wanted to let you know henrysx /tmp directory filled up with 200G of banding files and we ran out of disk space last weekend. I cleared out the files. | 01:02.25 |
| sorry ... henrysx6 | 01:02.47 |
kens | henrys (logs again) I imagine the reason the dictionaries in gsdparam.c aren't a problem is because they are 'persistent', but I'll take a look. The dictionary we have in pdfwrite is not persistent and so needs to be copied. | 07:49.06 |
| Also, the majority of those dictionaries are for rarely used PostScritp stuff like InputAttributes | 07:51.46 |
| And indeed a quick test shows that regular PostScript files simply aren't exercising any of those cases. Probably some of the CET files do | 07:53.07 |
| In fact all those cases seem to be tied up with /InputAttributes or /OutputAttributes, and those are very rarely encountered in PostScript. Indeed, InputAttributes are actually read from the device (to describe the possible available media), and shuold never be set by PostScript really. OutputAttributes are the ones which describe requested media and allow matching, but nobody uses these. | 07:57.28 |
tkamppeter | kens, chrisl, after long time I got a bug report again that file XYZ does not print. It is bug 693076. But there is one thing to say about Bugzilla. This question where I got GS from. Missing here is the selection "I am distro packager and forward a bug upstream". | 09:03.05 |
kens | tkamppeter I'll point the 'source' missing thing to Robin | 09:03.35 |
tkamppeter | kens, chrisl, therefore I have chosen "I don't know" to get the bug report accepted. | 09:03.56 |
kens | No problem till | 09:04.12 |
| We know who you are :_) | 09:04.18 |
| Hmm so this is a rendering problem, one for Alex.... | 09:04.38 |
| Till, if I open this file with Acrobat X it looks like the brackets are msising there too | 09:05.44 |
| In fact all the symbols seem to be missing when I open the file with Acrobat | 09:06.22 |
| Same with Acrobat 7 | 09:08.02 |
| Reader 4 is even worse | 09:08.08 |
tkamppeter | kens, the symbols are in there, though, as Poppler is able to display them. | 09:08.55 |
kens | tkamppeter every version of Acrobat or Acrobat reader I have displays this file with the symbols missing | 09:09.06 |
tkamppeter | kens, can one test somehow ehether the page is correct PDF? | 09:09.26 |
kens | We aim for compatibility with Acrobat, not Poppler | 09:09.26 |
| tkamppeter many PDF files are not correct :-( | 09:09.43 |
| But Acrobat generally displays them | 09:09.50 |
tkamppeter | kens, does this mean that this is a bug of Cairo and not of Ghostscript? | 09:09.57 |
kens | tkamppeter I would say so, yes | 09:10.08 |
| If Acroabt will not display the file properly then we would normally regard that as good evidence that the file is broken | 09:10.27 |
| I'll try 2 more consumers | 09:10.45 |
| MuPDF is the same as Acrobat | 09:11.33 |
| Jaws complains that it can't find many of the fonts, even though Acrobat shows them as present. I think the problem may well be incorrect font embedding | 09:12.56 |
chrisl | tkamppeter: that's not a missing option on the bug page - as a maintainer, we expect you to test with *our* release before passing the bug upstream, so you should be selecting "Built from source.... (no shared libraries)" | 09:14.03 |
kens | When I copy and paste the S(t) from the PDF file using Acrobat, the text is pasted correctly | 09:14.12 |
| SO it looks like its the actual font data which is broken | 09:14.32 |
Robin_Watts | tkamppeter: Morning. What chrisl said :) | 09:15.30 |
kens | chrisl do you want to take a quick look at why some of the symbols don't render ? | 09:16.23 |
| I think its more likely to be your field than Alex's given the behaviour of Acrobat | 09:16.46 |
chrisl | kens: not really - 83 pages, and god knows how many fonts, doesn't strike me as a "reasonable" test case | 09:17.25 |
kens | There's only one page isn't there ? | 09:17.41 |
| THe one that Till attached to the bug report is a single page | 09:18.03 |
| I can cut it down with Acrobat | 09:18.14 |
chrisl | Oh, okay, I guess I could look at it - still not really very interested, given we have Acrobat and others on our side...... | 09:18.43 |
kens | I'm inclined to agree, I don't believe this is our bug. | 09:18.59 |
| Here's a new one. In Acrobat X if I highlight text and press 'delete' it scores it through with red instead of deleteing it..... | 09:19.29 |
chrisl | Like an annotation or something? | 09:20.26 |
kens | Yes, exactly like | 09:20.35 |
| I seem to have to select another tool now.... | 09:20.50 |
chrisl | Well, I suppose it would make sense - *if* Acrobat hadn't previously actually deleted text like that! | 09:21.11 |
kens | Yep. And now I'm running up against the 'no system font' and it refuses to delete teh text :( | 09:21.35 |
| I can do it manually of course. | 09:21.41 |
| ANyway, I'm off out for a couple of hours, bye all | 09:22.51 |
chrisl | God, I *hate* cairo's output | 09:33.46 |
| Well, that was a complete waste of my time...... :-( | 10:41.46 |
| tkamppeter: it would have been right neighborly if you'd actually mentioned here or on our bug that the Cairo guys had already fixed the problem | 10:58.04 |
tkamppeter | chrisl, I have done it some minutes ago. | 11:05.23 |
Robin_Watts | I meant to bring up staff meeting dates at the 5pm meeting yesterday. | 11:36.11 |
| Gah. If you carefully build a heap with a object numbers, then renumber those numbers, the odds are strong you won't be left with a heap any more. | 11:42.48 |
henrys | kens:I would think anything that does much of anything with setpagedevice would exercise the dictionaries, but haven't verified. | 13:04.25 |
Robin_Watts | So, what's happening with the next meeting? | 13:07.19 |
kens | henrys, it doesn't seem to | 13:07.29 |
| I set breakpoints in all those locations and ran some files, and the breakpoints were never triggered | 13:07.49 |
Robin_Watts | I've not heard anyone say they can't make kens suggested dates (before the weekend), but then it's not been put to a formal question, and it's not clear if Miles was amenable to it. | 13:08.05 |
kens | Robin_Watts : next meeting is the 9/10 August | 13:08.05 |
| Miles said no | 13:08.15 |
Robin_Watts | Ah, fair enough. | 13:08.20 |
chrisl | August? | 13:09.35 |
kens | September, I can't keep months straight.... | 13:09.54 |
chrisl | Okay - doesn't make much difference for me..... | 13:10.22 |
Robin_Watts | So.. fly out on the 8th, back on the 10th? 16:30 flight on the 10th? | 13:10.40 |
kens | THat's what I was thinking but I haven't looked at flights yet | 13:10.57 |
chrisl | Aw, crap, Virgin Atlantic have changed their web site - what a pain..... | 13:11.57 |
Robin_Watts | Or there is a 7pm flight back. | 13:11.57 |
| 19:05 SFO -> 14:00 LHR | 13:12.12 |
kens | I would have thought 16:30 was OK | 13:12.14 |
Robin_Watts | Me too. | 13:12.18 |
| 61G out 62G back. | 13:34.22 |
kens | Uh, I suppose I'd better go book mine now.... | 13:34.38 |
| Robin_Watts : VS019 10:35 am and VS020 16:30 pm ? | 13:36.40 |
Robin_Watts | yes. | 13:36.46 |
kens | £735.59? | 13:36.48 |
Robin_Watts | yes. | 13:37.31 |
kens | Thanks | 13:37.41 |
Robin_Watts | np. | 13:37.46 |
kens | So 62G out 59G back | 13:43.45 |
| Flight looks quite busy | 13:43.56 |
Robin_Watts | indeed. | 13:44.07 |
kens | Hmm, I see they are increasing the number of points required to retain silver membership | 13:47.29 |
Robin_Watts | kens: really? | 13:48.07 |
kens | site says if your account finishes before Sept 1st its 10 points otehrwise its 15 | 13:48.33 |
| Looks to me like I will have only 12 in February | 13:48.50 |
| When the account ends | 13:48.56 |
Robin_Watts | We get 4 points for a flight to/from SF. | 13:48.58 |
kens | Yes | 13:49.02 |
| I have 4 now, will get 4 for the next trip and 4 for the one after | 13:49.14 |
Robin_Watts | so the fact we've been to miami and london has hurt us. | 13:49.15 |
kens | Not sure about Miami, but London definitely | 13:49.46 |
| Would be 16 for 4 trips | 13:49.56 |
Robin_Watts | didn't fly virgin to miami, did we ? | 13:49.59 |
kens | doesn't remember | 13:50.17 |
| When was that ? | 13:50.28 |
Robin_Watts | last december | 13:50.35 |
kens | Then its before the current period | 13:50.42 |
| Current period (for me) started on 23rd February | 13:50.55 |
| So I have 1 SFO trip and will get 2 more | 13:51.09 |
chrisl | We did fly VA to Miami | 13:51.20 |
Robin_Watts | We did, yeah. | 13:51.26 |
chrisl | TBH, the silver membership isn't that big a deal | 13:51.56 |
kens | But anyway, we need 4 trips on Virgin per year now, previously we only needed 3 | 13:51.58 |
| chrisl, yes I agree | 13:52.06 |
| Just noticed it in passing | 13:52.14 |
chrisl | It does seem strange, I can't see it costing them much having a lot of silver memberships | 13:53.44 |
kens | <shrug> Its a little irritating, but no big deal | 13:54.01 |
| I may earn some more points going on holiday anyway | 13:54.18 |
chrisl | I think we only benefited from it once | 13:54.54 |
kens | As much as that ? :) | 13:55.11 |
chrisl | Yeh, I think there was one time we managed to skip a large-ish boarding queue, but that's all | 13:55.56 |
Robin_Watts | The big benefit of silver membership is that you build miles faster. | 13:56.00 |
kens | I have mor miles than I know what to do with already :) | 13:56.16 |
| But I may get the chance to use some soon | 13:56.26 |
Robin_Watts | (and you get to board sooner, but 5 minutes doesn't make that much odds for a DVT) | 13:56.35 |
chrisl | More miles which they then don't let you spend on useful things like upgrading to premium economy on our flights | 13:57.02 |
Robin_Watts | virgin miles are better than AA miles, which I have never managed to use anywhere despite all my efforts. | 13:59.26 |
chrisl | It's just that I would definitely use them for an upgrade to premium on the return flight, if it was allowed....... | 14:00.43 |
Robin_Watts | yeah, me too. | 14:01.18 |
| The thing is, given they are flying 2 flights a day, I can't see that they can be THAT full on both. | 14:02.07 |
chrisl | I'm surprised they've got two flights a day | 14:03.58 |
Gigs- | I like southwest | 14:32.46 |
Robin_Watts | tor8: ping | 15:40.46 |
| We have pdf_copy_dict and pdf_copy_array, but no generic pdf_copy. | 15:41.11 |
| Any particular reason? | 15:41.18 |
| Morning mvrhel_laptop | 15:48.30 |
mvrhel_laptop | good morning | 15:48.41 |
Robin_Watts | You were going to attach the patch-so-far to that bug? | 15:48.48 |
mvrhel_laptop | hold on | 15:48.55 |
Robin_Watts | (the knockout/stroke one?) | 15:48.58 |
mvrhel_laptop | on phone | 15:49.24 |
Robin_Watts | no worries. | 15:49.29 |
mvrhel_laptop | Robin_Watts: There are some rendering issues that came up when I ran with my patch and had the pdf14_lops off but in a knockout group. This is without your complexity issues | 15:50.10 |
| I was going to look these over to understand what the issues are/were | 15:50.24 |
Robin_Watts | ok. | 15:50.37 |
mvrhel_laptop | otherwise, we have rendering issues that are dependent upon the complexity of thepath | 15:50.50 |
| which will get hard to debug later | 15:51.03 |
| if you want though, I can give you the patch and the list of files that have issues | 15:51.23 |
| and let you work on it | 15:51.26 |
| Looks like Avadhut is happy | 15:52.03 |
Robin_Watts | If it's pdf14 problems, I'll probably make slower headway than you, but I'll try anything once. | 15:52.03 |
| (except morris dancing) | 15:52.27 |
mvrhel_laptop | Robin_Watts: ok I will see if I can attach everything to the bug | 15:54.16 |
| that is I will attach the patch and add in a list of files that have rendering issues with the patch | 15:54.37 |
kens | Pizza time, night all | 16:01.34 |
ray_laptop | oops. missed kens | 16:09.18 |
henrys | mvrhel_laptop:did you report hylafax to scott I didn't see a support message? | 16:11.54 |
ray_laptop | Hylafax has been around for years. I wonder if they just changed from fully GPL open source | 16:13.10 |
mvrhel_laptop | henrys: no I did not. Once I got to looking at their product it is only for Linux systems. Would it not be easy for them to work around GPL if they were just calling gs? | 16:13.35 |
Robin_Watts | ray_laptop: As far as I can tell, hylafax has always existed as a BSD open source project. | 16:14.32 |
mvrhel_laptop | well they have a non-open source product | 16:14.50 |
| also | 16:14.55 |
Robin_Watts | But there is another group selling a hylafax "enterprise edition" that isn't open source. | 16:14.58 |
henrys | I recommend giving everything to scott - who knows? | 16:15.03 |
mvrhel_laptop | right | 16:15.03 |
| ok | 16:15.06 |
| I will do that | 16:15.08 |
Robin_Watts | I suspect that the original hylafax open source version already calls gs (so it probably is at pains to do so keeping within the GPL). | 16:15.51 |
ray_laptop | as long as they don't ship GPL ghostscript "embedded" in their software and just use GPL ghostscript (by calling it) it conforms to the GPL (unfortunately) | 16:15.53 |
mvrhel_laptop | right | 16:16.09 |
Robin_Watts | So unless the enterprise edition has fundamentally changed that, they'll be fine too. | 16:16.44 |
henrys | yes but we can also not fix their bugs if they don't get a support contract. | 16:16.50 |
mvrhel_laptop | that is true | 16:17.01 |
ray_laptop | The concern would be if Ghostscript were actually shipped as part of HylaFax enterprise | 16:20.03 |
henrys | I'll contact scott about it, don't worry about it mvrhel_laptop | 16:21.01 |
ray_laptop | it's hard for me to imagine that they sell some company a $1600 software package and then tell them "now to make it work, go download Ghostscript" | 16:21.10 |
mvrhel_laptop | henrys: ok. thanks | 16:21.19 |
| ray_laptop: yes | 16:21.24 |
| I agree at those prices | 16:21.30 |
ray_laptop | one of us (henrys ?) should download the free trial and see if it has Ghostscript in their package. That would be useful information for Scott and/or Miles | 16:24.22 |
| Even if they just "use" ghostscript (because it is shipped with the linux distros they support) we could add all of their "Customers" to our list as "companies that use Ghostscript" :-) | 16:25.52 |
mvrhel_laptop | good point | 16:27.16 |
ray_laptop | it would be an impressive list for our new web page: http://www.ifax.com/about-us/customers.html | 16:27.17 |
mvrhel_laptop | Robin_Watts: you are the proud owner of 692870 | 16:27.43 |
Robin_Watts | mvrhel_laptop: Thanks. I think. | 16:27.55 |
mvrhel_laptop | that is a nice list | 16:28.41 |
henrys | see the commercial support list of companies here too:http://www.hylafax.org/content/Support | 16:30.34 |
| I imaging this is going to be just like pdf995 though ... | 16:32.07 |
| s/imaging/imagine | 16:32.19 |
ray_laptop | my testing of the pdf14 optimization went pretty well. There were a bunch (69/569) of 'comparefiles' that don't run on their (older) version and none of the files that 'skip' transparency bands show any differences | 16:34.14 |
mvrhel_laptop | great! | 16:35.02 |
ray_laptop | now I am eager to port it over to the HEAD | 16:35.18 |
| but out of the 500 pages that ran only 93 skipped any bands, so I'll want to look at a few of those and make sure I'm not being too conservative. | 16:36.23 |
| on a non-debug build running their code with ppmraw and their banding params and resolution, the "problem" file renders in 0.435 seconds vs. the original time of 0.955 sec | 16:41.11 |
mvrhel_laptop | ok. well on to overprint simulation with RGB | 16:50.33 |
ray_laptop | oh, darn! My changes work fine on a gs build with their code base, but their 'devices' get really funky output. | 17:05.45 |
mvrhel_laptop | ray_laptop: so I am going to add another entry to gx_cm_opmode_t which is already part of color info to indicate when we have an RGB device that wants to simulate overprint | 17:56.47 |
Robin_Watts | mvrhel_laptop: Is this for the psdrgb thing ? | 17:57.18 |
mvrhel_laptop | this value is set to UNKNOWN in gdevbit.c for the DeviceRGBT example | 17:57.20 |
| no | 17:57.24 |
| it is for customer 532 | 17:57.30 |
| oops missed ray_laptop | 17:57.38 |
Robin_Watts | Ok. I'll shut up :) | 17:57.48 |
mvrhel_laptop | I will get to the psdrgb device after I get this stuff done for them | 17:58.15 |
Robin_Watts | Sure. Wasn't meaning to hassle. I'll just be interested to understand what it is you do with the psdrgb to make it work, so if you were doing something clever I was going to listen in. | 18:00.11 |
mvrhel_laptop | It likely will be to fix what ever bonehead thing I did to break it :) | 18:00.46 |
| ick. this RGB simulated overprint stuff is going to be such a hack from a color perpective | 18:05.07 |
Robin_Watts | mvrhel: Just checking my comprehension here: overprint says "when I write one color, remove all the other colorants underneath it", right? And that only makes sense when we are dealing with ink colors (CMYK + spots) rather than transmissive colors (RGB). | 18:15.15 |
mvrhel_laptop | yes | 18:15.23 |
| with an additive device there should be no overprinting | 18:15.50 |
| but cust 532 has an additive RGB device which later is actually CMY | 18:16.05 |
| plus a K generated | 18:16.10 |
| and they want overprint | 18:16.17 |
| so we need to fake things out a bit | 18:16.23 |
Robin_Watts | So whenever you write R you set G and B to full ? | 18:16.46 |
mvrhel_laptop | there will be some RGB to CMYK and back quackery going on yes | 18:17.07 |
| stuff that makes me shudder | 18:17.15 |
Robin_Watts | They can't be color managed here, right? | 18:17.31 |
mvrhel_laptop | right | 18:17.35 |
Robin_Watts | so presumably they just treat C = 1-R etc. | 18:17.43 |
mvrhel_laptop | yes | 18:17.47 |
| overprint and color managmeent is tricky even with CMYK | 18:17.57 |
| if you have different source and destination profiles it really is not possible | 18:18.08 |
| or it is very difficult | 18:18.14 |
Robin_Watts | so where normal overprint is "remove the other colorants", this 'inverted overprint' is "set the other colorants" | 18:18.17 |
| s/inverted overprint/overprint in a additive space/ | 18:18.46 |
mvrhel_laptop | it will be the negative. Right now, just getting us to a point where we push the overprint compositor | 18:18.58 |
| then my thought is to add some rgb <-> cmyk conversions around that stuff | 18:19.23 |
Robin_Watts | that sounds like overkill to me. | 18:19.45 |
mvrhel_laptop | I am not going to rewrite the overprint drawing routines | 18:19.58 |
| to handle RGB | 18:20.12 |
Robin_Watts | presumably they have lots of "draw 0" code ? | 18:20.23 |
mvrhel_laptop | Robin_Watts: I have not gotten to that point yet | 18:20.37 |
Robin_Watts | I think I have linearisation working from mupdf. | 18:21.22 |
mvrhel_laptop | Just making my way now to the proper compositor set up | 18:21.23 |
| need to head out for a bit | 18:21.34 |
| bbiaw | 18:21.36 |
Robin_Watts | do I replace the bubble sort with a better one... | 18:21.37 |
| cyl. | 18:21.39 |
mvrhel_laptop | Robin_Watts: you there? I know it is late | 21:04.03 |
| just wondering what it was you had in mind for overprint. there is a bunch of stuff in overprint to deal with the overprint mode and the non-zero and zero components to make sure that the proper destination colors are (or are not) erased during the drawing. My plan was not to rewrite all of that but to simply keep things in a clean CMYK situation | 21:10.03 |
| but anyway, I have a plan and I believe it is the proper approach. If you need help with the knockout stuff let me know | 21:11.05 |
mace | wonders what time it is in CA | 22:24.14 |
Robin_Watts | mvrhel_laptop: Hi | 22:33.46 |
| mace: here -8 | 22:33.54 |
mace | Robin_Watts: still office hours then.. | 22:34.32 |
Robin_Watts | mvrhel_laptop: If you have a plan, don't let me interfere. | 22:34.37 |
| mace: Should be. | 22:34.49 |
| Forward 1 day (to 2012/05/31)>>> | |