| <<<Back 1 day (to 2012/09/16) | 2012/09/17 |
marcosw | I ran into some unexpected issues while upgrading casper to 12.04. rather than more down time I've restored the original 10.04 based system and will work on the problems later this week. Please let me know if anything is not working. | 05:43.16 |
Robin_Watts | Morning paulgardiner. Good holiday ? | 08:33.37 |
paulgardiner | hi. Yes, pretty place Lynmouth. Kept getting lost and walking further than we planned, but legs mostly recovered now. | 08:36.43 |
| Thought it would be easier walking than the Lakes, but actually the coast paths are just as steep. | 08:39.04 |
Robin_Watts | :) | 08:44.03 |
norbertj | Hello Robin_Watts | 10:52.25 |
Robin_Watts | hi | 10:52.32 |
ghostbot | salut, robin_watts | 10:52.32 |
norbertj | I have a question on pxl detection of empty pages. | 10:53.00 |
| I see that the pxs->have_page is unconditionally set true in pxpaint.c/pxText() | 10:54.02 |
| But I have a job (word docx generated) that has empty pages but with some textcharacters with size 0,0 on it. This looks like an empty page. | 10:55.03 |
| But is not detected as such (because of the have_page already set to true). | 10:55.28 |
Robin_Watts | OK, this is code I'm not that familiar with, henrys is your man, but let me look. | 10:55.48 |
| I see, so pxText says: if we have some text, and we have a brush, we have a marked page. | 10:57.20 |
norbertj | What I did is in pxl/pxfont.c/px_text() retrieve the cursor from penum->path->position (before the gs_text_process())) and check it against the cursor after gs_text_process(). If there is a change then set pxs->have_page = true; | 10:57.37 |
Robin_Watts | And you'd like it to say "If we have some text, and we have a brush and the text size > 0 then we have a marked page | 10:57.45 |
norbertj | length of string passed > 0 is not necessarily cursor moved. | 10:58.26 |
Robin_Watts | Right, but strictly cursor moved is not the same as page marked. | 10:58.53 |
| Consider printing a space on a page. | 10:59.00 |
norbertj | A space would, in my opinion, mark the page | 10:59.26 |
Robin_Watts | It depends on your definition of marking. | 10:59.47 |
kens2 | What about text in white ? | 10:59.55 |
Robin_Watts | At any rate, this is something I don't feel confident about changing myself. | 11:00.05 |
kens2 | Or a non-marking glyph toher than space ( eg /.notdef) | 11:00.15 |
Robin_Watts | This is Henrys area of the code, and so I feel he should be consulted on this. | 11:00.20 |
| kens2: Indeed. | 11:00.25 |
norbertj | But if you did a px_text and it did not move the cursor (char + backspace in pxl ?) it did not mark I think. | 11:00.26 |
| Should I make a bug for this? | 11:00.50 |
Robin_Watts | norbertj: I think that would be best. | 11:01.00 |
kens2 | Possibly an enhancement bug | 11:01.06 |
norbertj | Will do. | 11:01.06 |
kens2 | THis is clearly a heuristic | 11:01.13 |
Robin_Watts | Your solution doesn't sound a bad one, but there may be hidden aspects of this that only Henry knows about. | 11:01.41 |
norbertj | Also note the comment in pxTextPath. | 11:02.08 |
| I'll make a bug and Henry can think on it. | 11:02.44 |
kens2 | Best thing to do I think | 11:03.19 |
joeh_ | Hello all | 14:04.46 |
| have a hopefully quick setting someone can help with. | 14:05.23 |
| I ran several thousand iterations of identify through imagemagick and now Im getting a GS error. | 14:06.11 |
| Ive reinstalled both IM and GS but the error remains. | 14:06.44 |
| Error: /undefinedfilename in (ICE=VICE=pbmraw) | 14:06.51 |
kens2 | Should read DEVICE | 14:07.01 |
| DEVICE=pbmraw | 14:07.27 |
joeh_ | Only during non-verbose indentify of a PDF file. Verbose works fine. | 14:07.37 |
| where can I set that? | 14:07.52 |
kens2 | I can't help with ImageMagick, all I can do is tell you that teh invocation is wrong tehre | 14:08.02 |
| I'm afraid I know very littel about ImagMagick | 14:08.21 |
joeh_ | so this is a setting in IM not GS? | 14:08.39 |
henrys | huh one from norbertj I see | 14:08.47 |
kens2 | That's a GS error, and the error says that teh command line was wrong | 14:08.54 |
| So I woudl say IM is callnig it incorrectly | 14:09.06 |
henrys | has casper recovered? | 14:09.16 |
joeh_ | rgr that, thank you kindly all. | 14:09.18 |
kens2 | henrys, we discussed it briefly in the logs | 14:09.22 |
| It seems lke a not unreasonable change to the heuristic, but we wanted you to see it. | 14:09.46 |
henrys | yes when I did that I didn't get into that level of checking, lacking documentation. | 14:10.45 |
kens2 | As Norbert syas, its never going to be 100% (withtou checking pixels) | 14:11.12 |
henrys | yes, we used to use the bounding box but that was problematic and slowed things down also. | 14:12.17 |
| bounding box device that is. | 14:12.32 |
kens2 | Yes. Well its up to you but it looked reasonable from the description | 14:12.53 |
Robin_Watts | henrys: Marcosw said that he hit problems with upgrading casper, and so has reverted to an old version. | 14:15.49 |
| Bah. First MSVC crash of the week. | 14:16.17 |
henrys | Robin_Watts:odd wonder what went wrong. | 14:16.28 |
Robin_Watts | The upgrade worked fine for me locally. | 14:16.43 |
| Still, it must be better to upgrade often, rather than waiting until the rpms aren't available any more etc. | 14:17.04 |
chrisl | debs, not rpms..... | 14:17.33 |
| I had a problem upgrading to 12.04, IIRC, I had to do it in two stages - run the dist-upgrade, then do a "repair" upgrade | 14:18.14 |
henrys | tor8:have you upgraded to mountain lion? | 14:26.44 |
tor8 | henrys: no | 14:26.55 |
| I guess I have to upgrade now, to build a version of mupdf for iPhone 5 | 14:27.18 |
henrys | I dread it, but I got apple tv and want to mirror. | 14:27.36 |
tor8 | henrys: the new airplay display mirroring thing? | 14:27.52 |
henrys | yes can't believe you have to upgrade the os to send a display over the network, but ... | 14:28.45 |
Robin_Watts | I've found ML to be better than L. | 14:32.27 |
| Just be aware that: | 14:32.32 |
| 1) VMWare Fusion 3 doesn't work. You need 4. | 14:32.46 |
| 2) valgrind doesn't work on ML. | 14:32.53 |
tor8 | Robin_Watts: but is ML better than SL? | 14:34.50 |
Robin_Watts | Sorry, I should have said: I've experienced no problems in my light use of the mac from moving from L to ML. | 14:35.34 |
| I did encounter problems moving from SL to L. | 14:35.42 |
| but I don't seem to see those any more. | 14:35.53 |
| I'm prepared to believe that ML is better than SL, but I wouldn't like to have to offer evidence myself. | 14:36.25 |
henrys | Robin_Watts:oh I assumed you didn't upgrade which is why I just asked tor8, thanks for the info. | 14:36.36 |
Robin_Watts | henrys: I'm a version junkie :( | 14:36.49 |
henrys | Robin_Watts: I just remember you going from SL -> L -> SL or soemthing like that and thought you'd abandoned the exercise with that. | 14:39.45 |
| the ML articles I've read have been positive, fixing a number of lion gotchas and a bit faster. | 14:40.55 |
norbertj | hello henrys | 14:41.27 |
henrys | oh hi norbertj, I'll have a look at your problem sometime this week. | 14:41.56 |
norbertj | as I put in the bug 693337, the solution is working for me. I'm just not sure whether a char + backspace would still work (I don't know it the backspace would be handled in the pxtext (can that be in pcl-xl?) | 14:42.41 |
| And if this would be something to be put in standard pxl (and something similar for pcl?) | 14:43.28 |
Robin_Watts | henrys: No, I never backtracked. tor8 might have done. | 14:44.25 |
henrys | I've never seen a backspace in xl - all text is "xyshow" or text has coordinates. | 14:45.29 |
ianneub | Howdy all, quick question (I hope), is it possible to get PDF bookmark data using muPDF? | 14:46.30 |
henrys | anyway I have to leave for about an hour - my cousin has dropped by, a surprise visit and I'm going to have breakfast with him. norbertj I'll check the logs when I return in case you have something new about the problem. | 14:46.44 |
Robin_Watts | ianneub: Yes and no. | 14:46.58 |
tor8 | henrys: Robin_Watts: I downgraded my laptop from L back to SL | 14:47.00 |
kens2 | enjoy breakfast henrys | 14:47.01 |
norbertj | henrys: ok, going home myself too. (End of workday). | 14:47.23 |
Robin_Watts | ianneub: No, we don't have an API specifically for retrieving pdf bookmark information. | 14:47.32 |
| Yes, you can get it using the generic PDF object reading/navigation functions. | 14:47.50 |
tor8 | Robin_Watts: we do, if by bookmark he means the "Outline" | 14:48.03 |
ianneub | Robin_Watts: ah ok. I will check along those lines then | 14:48.05 |
Robin_Watts | I thought bookmarks were something different to outline? | 14:48.25 |
tor8 | ianneub: mudraw -l file.pdf will print the outline to stdout, if that's what you're asking about | 14:48.42 |
ianneub | tor8: Thanks! I'll check that out right now | 14:48.58 |
tor8 | Robin_Watts: I believe the reader uses a different terminology from the spec | 14:49.02 |
Robin_Watts | Indeed, you're right. the spec says outlines == bookmarks. | 14:49.37 |
| kens: if you do a windows debug build of language_switch, does it build OK ? | 14:50.46 |
ianneub | tor8: Thanks, that's perfect! | 14:50.56 |
kens2 | No idea | 14:50.57 |
| not tried recently | 14:51.02 |
ianneub | Robin_Watts: thanks as well :) | 14:51.05 |
Robin_Watts | chrisl: ping | 15:25.05 |
chrisl | Robin_Watts: poing | 15:30.38 |
Robin_Watts | I think the windows build of language_switch is broken. | 15:31.00 |
kens | mty builsd was ok | 15:31.16 |
chrisl | language_switch is broken | 15:31.18 |
Robin_Watts | chrisl: IN that it fails to link? | 15:31.30 |
chrisl | In that it doesn't work, it should not be used..... | 15:31.43 |
kens | My build is OK but the executable doesn't work | 15:33.01 |
chrisl | I can't build it at all since it doesn't seem to understand the BUILD_SYSTEM parameter | 15:34.56 |
Robin_Watts | Ok. It's failing cos of WINDOWS_NO_UNICODE. | 15:37.50 |
| Given that my current (hooge) changes haven't broken it, I'm going to ignore it until I get to the utf8 api changes etc. | 15:38.17 |
kens | OK seems reasonale | 15:38.50 |
chrisl | Erm, that's confusing, the makefile has /DWINDOWS_NO_UNICODE - unless XCFLAGS is being defined elsewhere...... | 15:39.45 |
Robin_Watts | chrisl: Yeah. | 15:39.55 |
| and that's why it's dying for me. | 15:40.04 |
| My own makefile crapness, sorry. | 15:40.29 |
chrisl | Okay, I'll leave it with you for now - you can let me know if you need further confusion ;-) | 15:41.19 |
| I am confused about the BUILD_SYSTEM issue - if the pcl and xps makefiles handle it, how can ls not, I wonder...... | 15:42.01 |
apineda | would anyone recommend the best way to separate stroke lines of a particular color to a separate pdf? | 15:44.02 |
Robin_Watts | What input file? ps? pdf? | 15:45.40 |
mvrhel | I have to run a few errands this morning. bbiaw | 15:52.58 |
| anniversary today... | 15:53.15 |
kens | Ah, congratulations | 15:53.25 |
Robin_Watts | The trick is to buy a present BEFORE the day :) | 15:53.31 |
kens | How many years ? | 15:53.33 |
mvrhel | hehe | 15:53.36 |
| yes | 15:53.38 |
| 18 years | 15:53.40 |
Robin_Watts | Coo. Longer than me. | 15:53.53 |
mvrhel | unfortunately the main present is not going to arrive in time | 15:54.02 |
| so I need to get a small backup | 15:54.09 |
Robin_Watts | Things I have learnt about presents: | 15:54.31 |
| 1) Not too small, cos they think it might be jewellry, and then you're in trouble. | 15:54.44 |
mvrhel | yes :) | 15:54.52 |
Robin_Watts | 2) Nothing related to cleaning. | 15:54.52 |
mvrhel | ha | 15:54.56 |
| I will keep those thoughts in mind | 15:55.26 |
kens | I find chocolate is always acceptable | 15:56.59 |
| Preferably expensive chocolate | 15:57.07 |
mvrhel | yes. that is a good idea | 15:57.17 |
Robin_Watts | Depends on the woman; sometimes you can just be storing up pain for yourself later... | 15:57.37 |
kens | A hamper of Green&Black's went down well at Xmas :-) | 15:58.05 |
mvrhel | ha | 15:58.43 |
| ok. need to run | 15:58.49 |
apineda | @Robin_Watts input is pdf | 15:58.51 |
kens | bb | 15:58.55 |
Robin_Watts | apineda: How can you identify the color you want to keep? | 15:59.34 |
| I mean, it might be an RGB color, or a CMYK or a spot colorant. | 16:00.01 |
kens | Might be possible by redefineing stroke in Ghostscript to test the colour and discard the path if its wrong | 16:00.02 |
apineda | separation color name | 16:00.18 |
kens | Of course you have to make sure to test the stroke colour | 16:00.20 |
Robin_Watts | apineda: So basically you want to process a PDF and strip a particular separation out? | 16:00.40 |
| Do you need to get a PDF at the end? | 16:00.53 |
apineda | yes, I might have to do this myself. I need both pdfs. One stripped out of a stroke, the other just the stroke. | 16:01.07 |
kens | That's at least 2 passes then. | 16:01.29 |
Robin_Watts | Well, as kens says, you might be able to do it by hacking gs. | 16:01.39 |
| s/gs/the ghostscript postscript interpreter/ | 16:02.01 |
kens | I'm sure *I* could do it by hacking GS, but it might take me some time. | 16:02.05 |
| Probably a PostScript prologue would do it. | 16:02.22 |
Robin_Watts | http://git.ghostscript.com/?p=user/robin/ghostpdl.git/.git;a=commitdiff;h=6ce1c98a92b158baec04107fcaf57331a2420556 | 16:02.42 |
kens | Good grief.... | 16:03.12 |
Robin_Watts | Gents, I have a large change to gs (see the above link) that tweaks lots of printing functions. | 16:03.12 |
| Could I ask you all to take a quick look at the areas of code that you own within that commit to make sure I haven't done anything you horribly object to? | 16:03.50 |
chrisl | Robin_Watts: with that, can we build multi-threaded without the contrib drivers? | 16:06.28 |
kens | Hmm, I'm surprised there is so little of mine. Clearly pdfwrite needs more debug prints. It all looks OK to me. | 16:06.46 |
Robin_Watts | chrisl: If you define GS_THREADSAFE it defines all the non-threadsafe printing away to nothing. | 16:07.08 |
| hence, while I haven't touched the contrib drivers, it is at least possible that they will work. | 16:07.28 |
| We could make GS_THREADSAFE a build system thing (Rather than a CFLAGS thing), and use that to exclude contrib drivers that aren't known to work. | 16:08.49 |
kens | It seems to me that the contrib devices mostly drive specific printers, so excluding them from a multi-threaded build makes some sense to me. | 16:09.31 |
chrisl | Great. It looks less horrible than it might have been - I guess the memory pointers are pervasive enough to cover almost everything....... | 16:09.40 |
kens | Of course, if someone is building a print server, that might limit them a bit, but they are unsupported.... | 16:09.55 |
chrisl | I'm not sure what to do about the contrib devices - I suspect they only way they'll become thread safe is if we make the effort :-( | 16:11.22 |
kens | Yes, I'm sure that's true | 16:11.51 |
Robin_Watts | kens: Well, if someone was making a cheap print server box they'd probably only have 1 printer on it at a time. | 16:12.04 |
| hence a single api instance isn't a problem. | 16:12.11 |
kens | Robin_Watts : yes, I was thinking of a more complete implementation, driving several printers simultaneously | 16:12.43 |
Robin_Watts | yeah. | 16:13.09 |
| Well, there is no magic bullet if these things use statics :( | 16:13.23 |
kens | But like I said, these are basiclaly unsupported. THe only one I cna think of being a problem might be the opvp device. | 16:13.31 |
| chrisl I seem to remember we settled on oly 3 or 4 being 'important' ? | 16:14.05 |
chrisl | Well, based on information from Till, yes | 16:14.33 |
kens | Well, we might want to consider doing the work for those. | 16:14.56 |
| OK need to go start cooking, back later. | 16:15.33 |
chrisl | Yeh, devopvp has quite a few static variables :-( | 16:15.38 |
ray_laptop | happy anniversary, mvrhel. (ours is 24 yrs today) | 16:27.17 |
| mvrhel: when you are available later, I want to talk about the RGB overprint simulation. | 16:28.23 |
Robin_Watts | Top top for the day: rm out * is not the same as rm out* | 16:58.53 |
marcosw | Robin_Watts: I sometime have that happen when tab completion only finds one result and inserts a space after it (i.e. I type "rm test\t*\n" expecting there to be several files starting with 'test'). | 17:03.45 |
Robin_Watts | And I've just deleted the perl script I spent ages writing last week. | 17:05.16 |
| Well, Robins second law comes into play. I guess I know what I'll be doing later. | 17:06.31 |
| ray_laptop: (For the logs) Could you install helgrind on peeves please? | 17:10.10 |
Robin_Watts | pops out to fetch helen. | 17:10.19 |
marcosw | Robin_Watts: isn't this what time machine is for? | 17:15.14 |
Robin_Watts | marcosw: It was on peeves. | 18:05.49 |
marcosw | there are linux versions of time machine, I keep meaning to try one out but never get around to it. | 18:06.52 |
mvrhel | ray_laptop: I am back. happy anniversary for you and karen | 19:11.43 |
Robin_Watts | ray_laptop: Dunno if you saw my mail. Could you install helgrind on peeves please? | 19:18.44 |
ray_laptop | Robin_Watts: about to do that. Sorry I (also) had anniversary errands to do. | 19:19.21 |
Robin_Watts | no worries. | 19:19.27 |
| and thanks. | 19:19.56 |
ray_laptop | Robin_Watts: E: Couldn't find package helgrind (from apt-get install helgrind). Any idea what the package is called ? | 19:20.56 |
Robin_Watts | Hmm. | 19:21.22 |
| Oh, it's part of valgrind, not a separate exe. | 19:22.12 |
| Sorry. | 19:22.14 |
ray_laptop | Robin_Watts: looks like -- do I need to update the valgrind on peeves ? (currently 3.5.0-Debian -- latest release 3.8.0) | 19:23.15 |
Robin_Watts | ray_laptop: It's throwing up lots for me to look at. | 19:24.46 |
| So I'm guessing it's OK. | 19:24.50 |
| Thanks. | 19:24.52 |
ray_laptop | Robin_Watts: sudo apt-get upgrade valgrind says: 0 upgraded, 0 newly installed, 0 to remove and 0 not upgraded. | 19:24.58 |
| so maybe ubuntu (at least the rev on peeves) doesn't have an update | 19:25.24 |
| Robin_Watts: BTW, feel free to SMS me to make sure I notice what you need (in case I am off at cust 532, which probably will happen soon) | 19:26.32 |
henrys | Robin_Watts:as I recall Stefan Kemper made all those code changes before and then they were backed out on Tor's recommendation that the debug code not be overwhelmed with another parameter. Other than hearing tor8's views I didn't see anything wrong with the code. By way of convention it seems the memory parameter should come before the letter designator but I don't feel strongly about it. | 19:44.47 |
tor8 | henrys: I don't remember that... :) | 19:46.18 |
Robin_Watts | henrys: in the if_debug? Yeah, you may be right. | 22:35.57 |
| I'll canvas opinions at the meeting tomorrow. | 22:36.09 |
| Forward 1 day (to 2012/09/18)>>> | |