| <<<Back 1 day (to 2015/08/12) | 20150813 |
dalcde | Has anyone else experienced an xserver crash when using mupdf? | 02:27.18 |
| logs: http://pastebin.com/NM3WDtqk | 02:27.20 |
sebras | dalcde: I doubt that mupdf is to blame | 02:38.12 |
| dalcde: but I did find this report https://bugzilla.redhat.com/show_bug.cgi?id=1118540 | 02:38.18 |
dalcde | sebras: yes by definition an x server crash is a bug in the x server | 02:38.37 |
sebras | dalcde: the backtrace appears to be similar, so maybe it is the same (or a related) cause. | 02:38.42 |
dalcde | it's just that it is only triggered when using mupdf | 02:38.44 |
| usually when I scroll | 02:38.57 |
| (it doesn't crash immediately) | 02:39.14 |
sebras | dalcde: maybe the redhat bug report can help you trace and fix the problem in the x server. I doubt that anyone here can really help you with that. :) | 02:39.44 |
| dalcde: I need to go to bed now. | 02:40.20 |
dalcde | ok thanks | 02:40.23 |
| I just wanted to know if it is just me | 02:40.35 |
rayjj | hurray. At least that worked. | 04:17.57 |
| bbiaw | 04:18.23 |
jogux | hi ghostbot. I guess casper is back then :-) | 10:19.14 |
chrisl | So, I wonder if that was marcosw or coincidence.... | 10:19.37 |
jogux | hm, I can't actually get into casper yet. guess it may still be booting. | 10:20.47 |
chrisl | You can get to it if you use the AWS host name | 10:21.19 |
| And it's back..... | 10:23.05 |
kens | ah welcome back ghostsbot | 13:12.15 |
| D'oh I forgot to pull Henry's changes, no wonder I'm showing 2 files fialng :-( | 13:13.24 |
HD|Laptop | Hey all | 13:47.07 |
| I'm using the command gs -o '/tmp/pdf_Ma24c7_initialStamp.pdf' -sDEVICE=pdfwrite -g3200x600 -c "/Helvetica-Bold findfont 20 scalefont setfont" -c "0 .8 0 0 setcmykcolor" -c "100 0 moveto" -c "(VERTRAULICH) show" -c "0 20 moveto" -c "(üser näme) show" -c "80 40 moveto" -c "(13.08.2015 15:46:04) show" -c "showpage" to generate a pdf stamp | 13:47.47 |
| however the umlauts are messed up | 13:47.53 |
kens | define 'messed up' | 13:48.09 |
HD|Laptop | LANG=de_DE.UTF-8 and terminal is utf8 too | 13:48.15 |
| kens: umlauts get weirdly displayed | 13:48.24 |
kens | almsot certainly irrelevant | 13:48.27 |
| probably a missing font, cna you share the file ? | 13:48.44 |
rayjj | so, it looks like casper went down again after I had a problem (at around 0400 GMT) | 13:49.14 |
HD|Laptop | "Ëâ¦ser nËâme" is the output | 13:49.15 |
| kens: sure | 13:49.19 |
kens | rayjj yes | 13:49.27 |
rayjj | goes to check to logs | 13:49.31 |
kens | marcos mailed details of login | 13:49.41 |
chrisl | HD|Laptop: You're actually doing the text on the command line? | 13:49.49 |
rayjj | wonder why we are having such problems | 13:50.00 |
kens | Oh I missd that | 13:50.05 |
jogux | rayjj: the logs will be dodgy as ghostbot as been on/off a bit. | 13:50.07 |
HD|Laptop | chrisl: from a php script actually, but yes, commandline | 13:50.11 |
kens | No font selection there, so you get the dedault font | 13:50.19 |
chrisl | HD|Laptop: that won't work | 13:50.21 |
HD|Laptop | hmm but I have a font selection in there c "/Helvetica-Bold findfont 20 scalefont setfont" | 13:50.57 |
kens | You're doing several -c sections, much better to do it as 1 | 13:51.21 |
| and a whole lot clearer too | 13:51.27 |
chrisl | Postscript doesn't understand UTF-8 | 13:51.54 |
kens | You don't close them with -f either apparently | 13:51.55 |
| Indeed it doesn't | 13:52.02 |
HD|Laptop | https://www.dropbox.com/s/o4av5lm97vtp78b/41.pdf?dl=0 is the file | 13:52.21 |
kens | And with no Encoding other than whatever the font's default is, you migth get 'something' | 13:52.26 |
| As chrisl says you can't just send UTF8 to a PostScript interpreter | 13:53.07 |
HD|Laptop | okay | 13:53.18 |
kens | It won't work properly | 13:53.18 |
HD|Laptop | do I have to convert it? | 13:53.26 |
| also, is it possible in any way to center the username? | 13:53.35 |
kens | The character codes need to be appropriate to the Encoding of the font. | 13:53.49 |
| If you Google for 'center text in PostScript' you will find many answers, teching PostScript programmign is a bit off-topic here | 13:54.22 |
HD|Laptop | yeah I tried a couple of these but I failed miserably ^^ | 13:54.46 |
kens | Well stringwidth returns the width of the text. Take the media width, subtract the stringwidth, divide by 2 and use that as the x-offset for a moveto | 13:55.40 |
| then 'show' the text | 13:55.46 |
HD|Laptop | kens: how do I find out the encoding of the font? | 13:57.33 |
| shouldn't all fonts be utf8 supporting anyway, these days? | 13:57.48 |
kens | Usually you don't, you create an Encoding and tell the font to us it | 13:57.56 |
| HD|Laptop : No. | 13:58.01 |
chrisl | HD|Laptop: Postscript predates Unicode by quite some time | 13:58.06 |
kens | PostScrip predates UTF8 and does not support it | 13:58.12 |
chrisl | So, one could argue Unicode should be more cognisant of Postscript! | 13:58.41 |
kens | Look at the Acumen Trainign site for the Acumen Journal, especially November and December 2001: | 13:59.49 |
| http://www.acumentraining.com/acumenjournal.html | 13:59.49 |
HD|Laptop | hmm managed to get it running, sorta *cough* | 14:35.41 |
Gabengar | Hello | 17:19.37 |
ghostbot | Welcome to #ghostscript, the channel for Ghostscript and MuPDF. If you have a question, please ask it, don't ask to ask it. Do be prepared to wait for a reply as devs will check the logs and reply when they come on line. | 17:19.37 |
Gabengar | Anyone here? | 17:19.39 |
| How exactly does ghostscript interact with .jpgs | 17:20.05 |
chrisl | Ghostscript can create jpeg images, with appropriate options - I wouldn't say "interact" | 17:20.51 |
Gabengar | What kind of options? Can it create double-layered images... So an .jpg underneath a .jpg | 17:23.20 |
| kind of like an encrypted image | 17:23.28 |
chrisl | No | 17:23.34 |
Gabengar | :( | 17:23.37 |
chrisl | I've never even heard of such a thing | 17:23.47 |
Gabengar | So generally speaking, Ghostscript has nothing to do with cryptography? | 17:23.57 |
chrisl | Ghostscript contains cryptgraphics elements as required by the PDF specification | 17:24.24 |
| s/cryptgraphics/cryptographic | 17:24.50 |
ndiddylaptop | so there's this online puzzle that me and some guys from another chan are trying to solve that revolves around a bunch of images that look like pages full of runes and were created by ghostscript (the icc profile says Artifex in it). Here's the link https://www.dropbox.com/sh/lkta4q921vliyuw/AADmZ1YUHXWSjSizlMGZHXVMa . The interesting thing about the images is that if you look at them with a hex editor, you see blocks of rep | 18:27.58 |
| eating data, and if you delete all the blocks, the image is just gray. I'd like to know if A: it's possible for jpeg images to work this way (displaying a complex image using just repeating data sounds odd) and B: if ghostscript typically produces jpeg images with similar qualities (repeated data displaying an image). | 18:27.58 |
henrys | rayjj: can you access the cluster now? | 18:31.47 |
rayjj | henrys: yes, why | 18:32.23 |
| It's been working for me all day | 18:32.41 |
| (and I see you and chris on it sporadically) | 18:32.53 |
henrys | I've been wait for a while for a test to finish and the page doesn't work. Something must be weird with my browser, I can reach bmpcmp stuff but the dashboard | 18:33.29 |
rayjj | henrys: the dashboard is working fine. Your run just finished and mine is next | 18:34.03 |
| then chris' | 18:34.10 |
| bmpcmp | 18:34.15 |
Gabengar | I'm also interested in hearing the answer to ndiddylaptop's question :) | 18:35.38 |
henrys | rayjj: thanks working now just browser weirdness | 18:36.24 |
kens | ndiddylaptop : most of the JPEG data is not repeated patterns at all. I suspect that's 0 padding or somethign | 18:48.28 |
| You cannot edit a JPEG by deleting chunks of data, you break it if you do that. | 18:48.50 |
| Its encoded using a high pass filter to separately encode the high and low frequency components | 18:49.20 |
| I'm surprised the file will even open if you delete chunks of it. | 18:49.30 |
ndiddylaptop | kens: about half of the image is repeated data | 18:49.57 |
kens | Not the one I'm looking at, 0.jpg | 18:50.11 |
| But all that probably means is there are comparatively few colurs (which is true) | 18:50.41 |
| JPEG is a very bad choice for ths kind of image | 18:51.01 |
| FWIW the runes look very much like the Futhark rune set to me. | 18:51.24 |
ndiddylaptop | I think they chose jpeg because there's outguess data in it | 18:51.51 |
kens | I have no idea what you mean by 'outguess' | 18:52.05 |
ndiddylaptop | It's a program to hide text or other data in a jepg | 18:52.24 |
kens | Its possible that the image data has been manipulated to steganographically include data, but I very much doubt if what you are lookign at is hidden data | 18:53.12 |
ndiddylaptop | http://webcache.googleusercontent.com/search?q=cache:fHbR4q98fusJ:www.outguess.org/+&cd=1&hl=en&ct=clnk&gl=us | 18:53.45 |
kens | As I said, since there are only 3 colours (at least on the first page) there is liable to be very little low frequency data | 18:53.49 |
| So OUtguess is a treganographic program. | 18:54.10 |
| In any event, ths is absolutely *not* somethign Ghostscript does | 18:54.37 |
Gabengar | One more follow up kens | 18:54.57 |
| Correct me if I am wrong, but ghostscript can me used to convert a pdf to jpg and jpg to pdf | 18:55.15 |
kens | Wrong | 18:55.23 |
Gabengar | ahh | 18:55.26 |
kens | You cna render a PDF to a JPEG, you can create a PDF from a JPEG, but if you aren't careful what you get isn't the same JPEG | 18:55.54 |
| With care it can be. | 18:55.58 |
| You can include JFIF data directly as image data ion a PDF file | 18:56.14 |
| Ths shouldn't be considered a conversion though | 18:56.42 |
Gabengar | hmm gotcha | 18:56.56 |
kens | I very strongl;y suspect (but am not interested enough to check) that the repeated blocks of data are the low frequency components of the images. Because there are only three colours, you end up with a lot of duplicaton. | 18:58.06 |
| THe high frequency components are teh edges of all the runes and decoration | 18:58.35 |
Gabengar | So if one used it to render a PDF to a JPEG, would it be plausible that some pixels of a font type get messed up... Someone was noticing some rune letters appear differently, despite being the same rune. I thought, that it may just be whatever Artifex software was used that may have altered a few pictures | 18:59.17 |
| a few pixels* | 18:59.50 |
kens | JPEG is a lossy compression scheme, so sharp edges will be blurred, there will often be 'ringing' around sharp discontinuities | 19:00.05 |
| In short, JPEG is a really bad compression scheme for ths kind of image. Its intended for Photographc data (hence the Joint Photographc Experts Group, or JEPG) | 19:00.46 |
| JPEG* | 19:00.55 |
Gabengar | Ahhh | 19:01.42 |
| Okay :) Thanks for all the help | 19:01.49 |
kens | OK | 19:01.53 |
ndiddylaptop | yeah, thanks. people on our chan don't know as much about image formats and stuff as you | 19:02.31 |
kens | At high zoom factor I can see individual pixels in light grey round the sharp edges, pretty much waht I woudl expect to see | 19:02.35 |
| I'd suggest that the original was itself an image format of some kind, the edges are not very smooth | 19:03.09 |
| Or it could have been a rotten font :-) | 19:03.27 |
kens | runs away again | 19:04.59 |
Robin_Watts | Just running through the email that's come in in my absence (while trying not to fall asleep at the keyboard) | 19:08.37 |
| I reckon we've dropped support requests for MuPDF at least twice. | 19:08.57 |
| "MuPDF and questions on multipage" on the 3/8/2015 14:16 needs a reply. | 19:09.19 |
| "Commercial License of MuPDF" from Scott on the 3/8/2015 21:03 needs a reply about Android Korean text entry. | 19:09.50 |
rayjj | OK, I found an issue with 12-07B.PS (which was badly broken with the previous HT code) when num_repeat > 1, but I have to set this aside, so I am going to stop for now. | 20:55.07 |
| on the fast-HT code fixes. | 20:55.23 |
| mvrhel (for the logs): I will email the patch (cc tech) for comment, but may commit it. I have several other patches stacked up and need to get back to bug 695925 | 20:57.06 |
| mvrhel (for the logs): (also emailed) Please have a look, at http://git.ghostscript.com/?p=user/ray/ghostpdl.git;a=commitdiff;h=ca8018087abedfd2161d66a526813f001d8ecc41 | 21:41.45 |
Robin_Watts | fredross-perry: OK, there should be a 'forfred' branch on my casper mupdf repo that has the commit you've been waiting for in. | 22:29.14 |
| Sorry about that! | 22:29.17 |
fredross-perry | hey welcome back! | 22:29.25 |
| Robin_Watts: mvrhel: I have a couple of questions that would impact the UI. (a) whatâs the maximum number of separations Iâll encounter? Like 3, or 4, or more? and, is it possible/desirable to enable/disable them separately? As opposed to one at a time. | 22:32.56 |
Robin_Watts | fredross-perry: Maximum supported by photoshop is 56. | 22:33.20 |
| gs supports 64. | 22:33.27 |
| I would imagine that CMYK + less than 8 others will cover 99.9% of the cases you'll see. | 22:34.17 |
| but ideally we should cope with more. | 22:34.26 |
| The only vague idea I had for the UI for this was to have a top bar that looked like: | 22:34.58 |
fredross-perry | Are there commonly-known sets of these? | 22:35.11 |
| I was thinking weâd have RGB (3) or CMYK(4) for starters, and have 3 or 4 extra icons at the top, that youâd click on and off. | 22:35.45 |
Robin_Watts | a row of squares (C,M,Y,K,etc) across the top showing the different colors, and tapping each one to turn it on and off. | 22:35.58 |
fredross-perry | same. | 22:36.05 |
Robin_Watts | And then maybe to have the name of each spot shown diagonally down from it at 45 degrees. | 22:36.20 |
| but that might be hard to code. | 22:36.40 |
| and ideally that bar should scroll left/right if there are too many to fit on one screen. | 22:36.56 |
fredross-perry | but if itâs more than 4 in any common scenario, you want more of a floating toolbar or something. I was actually thinking weâd display the actual color in each button. | 22:36.57 |
Robin_Watts | Yes, each square would be the 'equivalent' color. | 22:37.15 |
| (either hollow or solid depending on whether it was enabled or not) | 22:37.28 |
fredross-perry | and maybe replace the âPâ icon sith something like: /Users/fredross-perry/Desktop/colors.png | 22:37.58 |
| well, thatâs not helpful. http://www.ross-perry.com/colors.png | 22:38.46 |
| weâre thinking the same it seems. Thanks. | 22:39.02 |
Robin_Watts | fredross-perry: That doesn't really fit with the 'style' of the mupdf icons. | 22:39.20 |
| But having someone other than me draw the mupdf icons would be good :) | 22:39.36 |
fredross-perry | Yes, I know that. | 22:39.40 |
| ;-) | 22:39.45 |
Robin_Watts | I am not a graphic artist :) | 22:39.45 |
fredross-perry | But I think that since weâre dealing with colors, then why not? | 22:40.06 |
Robin_Watts | I agree that in a revised UI, that would be a good thing to use. | 22:40.37 |
fredross-perry | also thinking, when we return from the âproofâ activity, weâd also retutn to the default top bar, rather than making the user tap âXâ. | 22:40.50 |
Robin_Watts | fredross-perry: Yeah, probably. | 22:41.11 |
fredross-perry | hope you had a nice time away? | 22:41.25 |
Robin_Watts | Had a fab time. | 22:41.37 |
| Thanks. | 22:41.39 |
| But I'm massively sleep deprived now, so I've gotta go flake out for a bit. I should be back tomorrow during normal UK hours or so. | 22:42.22 |
fredross-perry | go crash then. cheers | 22:42.50 |
| anyone: I am dying for a mac terminal replcement where I can set up a toolbar full of commands, or some equivalent idea. Thoughts? | 22:43.48 |
| Forward 1 day (to 2015/08/14)>>> | |