| <<<Back 1 day (to 2015/10/28) | 20151029 |
bvx89 | chrisl: I've checked with pdf_info.ps, and you're right, it's the boxes of the PDF that is the problem | 12:53.22 |
chrisl | bvx89: so, hopefully setting the media size will get you where you need to be | 13:02.21 |
bvx89 | Yes. I just need to find a way to calculate from WxH of the original JPG to points that PS uses | 13:03.01 |
| For a JPEG with the resolution 14394x13157, the correct box had the resolution 2590.92x2368.26 | 13:03.42 |
Robin_Watts | bvx89: So, you have JPEG with *dimensions* 14394x13157 | 13:04.29 |
bvx89 | Correct | 13:04.34 |
chrisl | points are 1/72 of an inch | 13:04.34 |
Robin_Watts | If you're displaying that in the PDF at 300dpi then it would appear as 14394/300 inches wide. | 13:06.33 |
| So the box would need to be 14394/300*72 wide. | 13:07.10 |
bvx89 | Thanks! | 13:07.40 |
| It's actually 400dpi, which makes the numbers work out | 13:07.53 |
chrisl | The slight issue, if you are using viewjpeg is that it doesn't get the resolution JFIF tag - which, IIRC, may not be there to start with...... | 13:13.01 |
bvx89 | Well, I was thinking about using Graphicsmagick for extracting the dimensions and the DPI from the JPEG | 13:23.58 |
chrisl | kens: 696305 is a clipping problem | 13:34.29 |
kens | Oh joy | 13:34.37 |
chrisl | Just before we draw the text, we set a clip. Intersecting that new clip path with the existing one yields an empty clip | 13:35.28 |
| Which clearly it shouldn't | 13:35.47 |
kens | I presume it isn't supposed to, and that somehow the bonkers CTM is involved | 13:35.52 |
chrisl | The bonkers CTM is probably a major factor | 13:36.24 |
| The thing is at most resolutions the new clip and the existing one round to the same values, hence we just ignore the new clip. Only at some resolutions is there a very small difference in the coordinates, and thus we go through the process of intersecting the paths, and end up with an empty clip | 13:37.53 |
kens | Yeah that's almost certainly something to do with the insane CTM | 13:38.18 |
| Well if you want to habd it to me, change the assignment and oput that in the bug report and I'll see if there's anything we can do about it. | 13:38.54 |
| I'm halfway through the broken font in 696306 at the moment and don't want to stop right now | 13:39.14 |
chrisl | It's hardly your area - this is graphics library stuff. | 13:39.25 |
kens | Oh, I was sort of assuming it was the PDF interpreter | 13:39.40 |
| FWIW I don't believe we can get the file for 696306 to work 'properly' I think Acrobat must be ignoring the error in the type 1 font. And only whichever version of Acrobat Marcos used (he hasn't said). My version throws a wobbly and prints bullets..... | 13:40.49 |
chrisl | No, we have two valid, *almost* identical paths coming from the interpreter, and when the clipping code tries to intersect them, it results in an empty clip, which is really very wrong | 13:40.53 |
kens | Oh yeah that seems like a major problem all right, I misundersttod, I thought they were supposed to be the same | 13:41.22 |
chrisl | Well, they probably are meant to be same, but with two different sets of scaling going on, rounding differences are not surprising, and should be handled (albeit slightly more slowly) | 13:43.15 |
kens | Hmm, might have to look t the source again | 13:44.04 |
chrisl | At the time we set the clip, the ctm is [-4.16666651 0.0 0.0 4.16666651 6496.06299 -0.275072902] - so not really that bonkers | 13:46.11 |
kens | Yeah I thought it was worse, I saw some really big values in there | 13:46.44 |
chrisl | TBH, I don't see much that's too nutty going on here at all | 13:47.56 |
kens | Are you looking at the original file now ? | 13:48.20 |
chrisl | No, I've cut down your one even further | 13:48.37 |
kens | COUld be that I'm thinking of the original | 13:48.54 |
chrisl | Oh, there's plenty of nutty sh*t in the original, but that doesn't seem to cause us a problem! | 13:49.25 |
kens | Probably because of all the other times we've run across it and had ot work around it | 13:49.50 |
| Hmm, well that caused a slew of diffs | 14:19.49 |
henrys | the convention with git log messages is less than 50 characters for the first sentence, then a blank line followed by multiple paragraphs wrapped to 72. I fear trying to get everyone to "sign on" to this one but I'm going to give it a go. | 15:33.30 |
kens | doesn't fancy counting characters individually | 15:34.23 |
Robin_Watts | kens: No one minds if you're out by a few chars. | 15:36.27 |
mvrhel_laptop | I try to limit the first line. I had not noticed if we were having an issue with this | 15:36.42 |
henrys | kens: I really feel doing commits with a real editor that wraps and spell checks is not unreasonable | 15:36.50 |
Robin_Watts | But I've seen people that regularly either do an essay in the first line, or that don't bother to wrap subsequent paragraphs. | 15:37.25 |
henrys | what are we doing these commits with, sed? | 15:37.58 |
mvrhel_laptop | Just so I know, am I one of the problem people? | 15:39.06 |
| I use the commit box with tortoise git which has a spell checker and I do returns on my lines | 15:40.35 |
henrys | mvrhel_laptop: the goal is not to have any ellipses here: http://git.ghostscript.com/?p=ghostpdl.git;a=search;s=Michael+Vrhel;st=author | 15:40.52 |
mvrhel_laptop | I see | 15:41.20 |
kens | henrys I'm using Git Gui | 15:41.50 |
mvrhel_laptop | I can make sure that happens more often | 15:41.59 |
kens | I don't think 50 characters is enough | 15:42.22 |
mvrhel_laptop | It just means the title is not going to tell you too much | 15:42.43 |
kens | Exactly | 15:42.51 |
mvrhel_laptop | And will not be a real sentence | 15:42.55 |
henrys | all the git tools use 50 character for that stuff so everyone will see truncated output when they see the logs (all tools) ... | 15:43.13 |
kens | I try to keep the first line short and I try to put line breaks in at respectaqble intervals in the following paragraphs | 15:43.21 |
henrys | thought this might be controversial | 15:43.43 |
Robin_Watts | mvrhel_laptop, kens: 50 chars is plenty. | 15:43.50 |
kens | henrys so woult you prefer a truncated title or a useless one ? | 15:43.50 |
mvrhel_laptop | Robin_Watts: I can do it. I am just stating that there will be less information. | 15:44.21 |
henrys | that's a false dichotomy | 15:44.25 |
mvrhel_laptop | But that is fine | 15:44.35 |
| I don't have a problem with it | 15:44.42 |
kens | I like to put in the title what component the commit is for and what it fixes | 15:44.43 |
| Because I feel that is useful to see quickly | 15:44.55 |
Robin_Watts | kens: "bug 69999: pdfwrite; fix missed SMask case\n\n<then describe it as long as you want>" | 15:45.03 |
henrys | kens: nobody sees it quickly because they use standard git tools to look at logs so they just a truncated sentence. | 15:45.40 |
Robin_Watts | looks for an example of a bad first sentence and finds: "Fixes PCL crash when a new ICC directory is specified on the command line" :) | 15:46.00 |
kens | Yeha but they see the start of it where I put the majority of the info | 15:46.00 |
| My commits generally start 'PDF interpreter - ' or 'PS interpreter' or 'pdfwrtei - ' | 15:46.19 |
henrys | Robin_Watts: I am guilty of it as well | 15:46.27 |
Robin_Watts | kens: TBH, your lines seem pretty good to me. | 15:46.37 |
kens | I don'#t feel I can reliably put a useful sentence together in 50 characters in all cases | 15:46.52 |
| I do *try* to keep the commit lines short | 15:47.09 |
Robin_Watts | kens: Well, you come close. And if people are using the standard git tools, they'll just lose the last few chars. | 15:47.24 |
kens | But I don't want to get into the Stack Overflow nonsense of counting characters | 15:47.36 |
Robin_Watts | But: "Some tidy ups, add individual targets for each product have configure check for pcl and xps sources, and use that to decide if we add them to the "all" target." | 15:47.47 |
| That's a bad one. | 15:47.52 |
| (just scanning through git logg here - not trying to pick on anyone) | 15:48.19 |
kens | Regarding spelling, I did point out before that I am using a spell checker, but lots of the stuff I put ion log messages gives spell checkers fits, so I sometimes miss typos | 15:49.12 |
| Thngs like CTM TextMatrix and FontMatrix don't sti weell with spell checkers | 15:49.43 |
henrys | I guess wrapping at 72 is good enough. | 15:51.03 |
kens | I don't know what GIt gui wraps at :-( | 15:51.18 |
| Ah, I cna set it, its currently 75 | 15:52.06 |
mvrhel_laptop | question for everyone | 15:52.22 |
| Bug 695277 is an odd one, since it is a PS file that invokes the PDF14 device | 15:52.58 |
| It does not set up the number of spot colors on the page so the PDF14 device fails with the separable devices that depend upon this knowledge | 15:53.35 |
kens | Personally I thnk you should feel free to modify the PostScript as required. Anyone that sumbmits a bug using our (undocumented) POstScript extensions will get laughed t cruelly | 15:53.49 |
mvrhel_laptop | OK. I am going to do that. I am feeling like I am heading down a rat hole getting this to work | 15:54.12 |
kens | I don't thnk you shold bother | 15:54.23 |
henrys | but keep in mind folks will read your stuff with tools like git.ghostscript.com so they'll see a truncated sentence. the convention is 50. | 15:54.25 |
mvrhel_laptop | and I don't see any real up side to it | 15:54.28 |
kens | Its intended to be used from PDF, if you use it from POstScript you're on your own IMO | 15:54.43 |
mvrhel_laptop | That is my feeling too | 15:54.53 |
kens | henrys I *really* don't want to have to sit and count the characters in my first line every time, it'll waste a load of time | 15:55.26 |
| Not to mention irritating me every time | 15:55.37 |
henrys | yeah that would drive me nuts too. | 15:56.15 |
rayjj__ | mvrhel_laptop: let me know if setting PageSpotColors doesn't take care of it for you (I noticed that the file is already doing a setpagedevice, so you can just put it in the same dict) | 15:57.03 |
kens | I try top keep the first line short, and I try to get useful info in early, I'm happy to carry on with that | 15:57.22 |
henrys | fair enough | 15:57.43 |
kens | Drat, my fix for ths bug breaks a Sumatra file :-( | 15:58.14 |
henrys | shelly's commit that I just put in is a bad one | 15:58.25 |
robmv | Greetings, has anyone tried the new signed ghostpdf.cat on Windows 10. It still says the driver is not | 15:59.37 |
| digitally signed | 15:59.40 |
kens | There was a problem due to the file being copied to a Linux machine as part of the releaser process, run unix2dos on the .inf file or search our bug database and pick up a file from there | 16:00.41 |
robmv | ohh thanks, will check that | 16:01.21 |
kens | http://bugs.ghostscript.com/show_bug.cgi?id=696244 | 16:02.19 |
| You can still install the printer of course | 16:02.39 |
| Hi Pete | 16:04.01 |
robmv | unix2dos did the fix, thanks again | 16:06.28 |
kens | NP | 16:07.19 |
pedro_mac | hi Ken - feels like Iâve been absent from here for ages. Too much time spent on skype! | 16:07.31 |
kens | A pleasure I'm managing to avoid so far :-) | 16:08.31 |
pedro_mac | good move - you might get dragged in if you raise your head above the parapet ;) | 16:10.51 |
marcosw | henrys: ping | 16:25.57 |
henrys | marcosw: yup | 16:26.18 |
marcosw | i'm cleaning old emails and can't remember what we decided at the staff meeting to do about non-staff not being allowed to comment on customer bugs. Specifically htl and <http://bugs.ghostscript.com/show_bug.cgi?id=695074>. | 16:27.42 |
henrys | they aren't allowed they can create a new bug if need be. | 16:28.54 |
mvrhel_laptop | ok adding /PageSpotColors 0 to the setpagedevice fixed the issues | 16:29.14 |
| so that is fixed | 16:29.22 |
henrys | marcosw: ^^^ | 16:29.23 |
marcosw | yeah, saw | 16:29.29 |
kens | Sounds good to me mvrhel_laptop | 16:29.29 |
marcosw | thx | 16:29.30 |
henrys | marcosw: I added my machine, see the logs but I guess I missed something. | 16:30.08 |
marcosw | okay. | 16:30.17 |
henrys | marcosw: added to the cluster | 16:30.18 |
marcosw | henrys: oh, you are running windows? | 16:30.52 |
henrys | marcosw: no | 16:31.14 |
| linux | 16:31.21 |
marcosw | then you don't need those scripts. they were hacked (and I do mean hacked) together by Robin_Watts to allows windows cluster nodes. I'm sure windows cluster compatibility bit-rotted away a long time ago so I should remove mention of those scripts. | 16:33.09 |
| what's your node name? I'll add it to the cluster master. | 16:33.35 |
henrys | marcosw: henrysi7 I did add it to .ssh blindly copying how it's done for other machines | 16:34.36 |
| marcosw: is there more? If so it should be in the readme. | 16:35.10 |
| marcosw: and the build example needs to be updated since chrisl's change to the directory structure. | 16:38.21 |
marcosw | The monitorPort.pl script needs to be restarted so that it picks the new machine. I'll fix the code so that won't be necessary going forward. | 16:38.46 |
henrys | marcosw: cool | 16:39.57 |
marcosw | actually the monitorPort script is automatically restarted if it's not running, so just killing it is enough. I'll add that to the readme and also fix the test build instructions. | 16:41.32 |
henrys | marcosw: oh the customer can edit the customer bug. | 16:47.00 |
marcosw | the reporter can | 16:47.23 |
henrys | marcosw: right | 16:47.28 |
marcosw | and the assignee | 16:47.36 |
henrys | off to cross fit class back in an hour or so | 16:47.49 |
marcosw | have fun (or whatever passes for fun around here). | 16:48.04 |
| mvrhel_laptop: regarding <http://bugs.ghostscript.com/show_bug.cgi?id=695277>, did you want me to replace transparency_example.ps with transperency_edit.ps in our regression test repository (and also in ghostpdl/examples)? | 16:57.43 |
mvrhel_laptop | marcosw: oh I did not realize is was in ghostpdl/examples | 16:58.16 |
| Yes it should be replaced. But keep the name transparency_example.ps | 16:58.33 |
marcosw | will do. | 16:58.40 |
mvrhel_laptop | thansk | 16:58.42 |
| thanks | 16:58.46 |
marcosw | mvrhel_laptop: turns out the file is not in our regression test repository, which surprised me since I thought all the files in ghostpdl/examples were. I will add it. | 17:02.19 |
| actually it was, just not in the same place as the other files. | 17:07.55 |
mvrhel_laptop | ok. This is why I left the file replacement for you. I have no idea where these things go | 17:10.27 |
marcosw | chrisl: what is the git magic to be able to see the commit log for files that moved as part of the build consolidation? e.g. "git log examples/alphabet.ps" | 17:28.56 |
Robin_Watts | --follow | 17:31.17 |
| (I think) | 17:31.26 |
marcosw | Robin_Watts: that was it, thx. A follow-on question: how do I make that the default for git log? | 17:33.21 |
Robin_Watts | You can't. | 17:33.38 |
| Cos --follow with more than 1 path will give you an error. | 17:33.50 |
| What you could do is make an alias like: git history | 17:34.19 |
marcosw | actually, the google says: | 17:34.24 |
| Teach the command to behave as if "--follow" was given from the command line when log.follow configuration variable is set and there is one (and only one) path on the command line. | 17:34.29 |
| git config log.follow true | 17:34.34 |
Robin_Watts | Oh, right. That's nice. | 17:34.50 |
marcosw | wonder if it works... | 17:34.56 |
| requires git 2.6, so no, it doesn't work for me. | 17:36.56 |
Robin_Watts | git config alias.history "log --follow" | 17:38.51 |
| Then you can: git history file | 17:39.01 |
marcosw | yeah, but then I have to remember to use git hstory instead of git log. | 17:39.43 |
| since it easier asking than looking on the internets, is it possible to add default options to git commands? I prefer git log to always have --name-only and --format=fuller. | 17:41.25 |
Robin_Watts | marcosw: Dunno offhand, sorry. | 17:41.57 |
| There may be variables you can set using git config, but I don't know them. | 17:42.47 |
marcosw | apparently git config alias is the only solution. | 17:47.28 |
marcosw | needs to pop the stack and try to remember what he was working on 2 hours ago | 17:48.47 |
mason4444 | Does anyone know if the javascript in Android version of mupdf will allow manipulation of the textfield.richvalue setting via span? Trying to set different characters to 2 different colors. | 21:45.37 |
mvrhel_laptop | rayjj: when you come back there is the commit on my repos for you to have a look over | 22:23.38 |
| hi rayjj: | 23:32.59 |
| so my regression testing came back fine | 23:33.07 |
| and the patch is on my repos on casper | 23:34.11 |
rayjj | mvrhel_laptop: thanks. I'm looking at the code. Just a few more minutes... | 23:34.32 |
| Forward 1 day (to 2015/10/30)>>> | |