| <<<Back 1 day (to 2014/06/12) | 2014/06/13 |
kens | Oh joy "please bisect between 9.07 and 9.14 with this 251 page file'.... | 08:39.50 |
chrisl | thinks: "I'm sorry, but the changes involved are too extensive...... etc" | 08:43.26 |
kens | Well at the moment I don't know, I'm running timing tests right now, so I can't do a bisect (I need my binary of master) | 08:45.04 |
| FWIW the latest changes you put in look good chris, the times are more or less the same as 9.05. More when I get the finished numbers | 08:45.42 |
| OK well the startup times were 0.17 seconds for 9.05, 0.29 for 9.14 and .24 for master. | 08:49.42 |
| That's running a simple 'quit' using the same setup as the real ps2write test, averaged over 100 runs | 08:50.26 |
| the dumb job they sent takes .24 on 9.05, .29 on 9.14 and .27 on master, averaged over 100 runs | 08:51.39 |
| So we're somewhere in the middle and almost the entire time is taken up starting and shutting down Ghostscript. | 08:52.22 |
| Now to repeat for all the jobs they sent | 08:52.32 |
kens | isn't going to repeat the startup times each time | 08:53.04 |
chrisl | Hmm, I'd hoped those changes were a bit better than that...... | 08:55.50 |
kens | Well, that's what I see here (windows 32-bit release version) | 08:56.12 |
| Here's the raw figures: | 08:56.32 |
| testing startup times | 08:56.32 |
| 9.05, start 9:40:13.56 | 08:56.32 |
| end 9:40:30.64 | 08:56.32 |
| 9.06, start 9:40:30.64 | 08:56.32 |
| end 9:40:47.76 | 08:56.32 |
| 9.07, start 9:40:47.76 | 08:56.33 |
| end 9:41:04.88 | 08:56.33 |
| 9.09, start 9:41:04.88 | 08:56.34 |
| end 9:41:23.31 | 08:56.34 |
| 9.10, start 9:41:23.31 | 08:56.35 |
| end 9:41:41.64 | 08:56.35 |
| 9.14, start 9:41:41.64 | 08:56.36 |
| end 9:42:00.42 | 08:56.36 |
| Master, start 9:42:00.42 | 08:56.37 |
| end 9:42:24.49 | 08:56.37 |
| That's 100 executions | 08:57.13 |
chrisl | Well, better is better, I suppose | 08:57.43 |
kens | actually the 9.14 result looks anomalous | 08:58.01 |
| I'll be rerunning the test anyway, I have to add all the other files and outptu combinations :-( | 08:58.39 |
chrisl | kens: do you want me to do the bisect? | 09:00.25 |
kens | chrisl if you have the time, otherwise I'll get to it htis afternoon | 09:00.40 |
chrisl | It won't take long | 09:01.01 |
kens | I can't even imagine which change(s) might have caused that | 09:01.02 |
chrisl | I have a vague memory of some scaling related changes, but I could be off base...... | 09:01.33 |
kens | Your memory is better thanmine, but of course in the span of 3 years..... | 09:01.54 |
chrisl | Mind you, many pages of text in a Type 3 font? Doh! | 09:02.45 |
kens | I know, but its possible the original wasn't a type 3, I haven't looked at the file yet | 09:03.13 |
chrisl | True - but the crazy bbox doesn't hint at a "conventional" original font type, either | 09:03.55 |
kens | Yeah that's true. I do have an open enhancement request to try and do something about mad Font BBoxes in type 3 fonts, it could be this is related | 09:04.30 |
| After all, this is a customer with history on mad files | 09:04.51 |
chrisl | One amongst many..... | 09:05.04 |
| Hmm, can't seem to reproduce the problem on Linux x64...... | 09:11.36 |
kens | THat's kind of surprising | 09:12.25 |
chrisl | And the failed to mention the raft of errors it spits out | 09:12.54 |
kens | It looks like the PDF file has *already* been 'converted' from PDF to PDF once. Why do people think repeated conversion is a good idea ? | 09:13.10 |
chrisl | kens: I don't see missing text on Windows, either, so <shrug> | 09:31.36 |
kens | I'll give it a try later, I have the release binary for 9.07 here | 09:31.58 |
chrisl | The only think I didn't do was install 9.07 - I just used the "bare" binary | 09:34.51 |
| s/think/thing | 09:35.02 |
kens | Like I said, I'll give it a bash later, I do have the binary installed. | 09:35.19 |
| Though I doubt that makes a difference. I wonder if they built the binary themseleves... | 09:36.01 |
| Or if there is something else they aren't telling us | 09:36.09 |
chrisl | Like random other patches | 09:38.42 |
kens | Quite possibly, yes | 09:38.53 |
| Leave it with me and I'll test it when my PC is free | 09:39.22 |
| Well, when my build isn't doing other things anyway | 09:39.38 |
Shelly_ | chrisl: kens: morning guys. Chris, free for a chat? check other window... | 10:07.50 |
kens | Hi SHelly | 10:07.58 |
Shelly_ | kens: good to finally get back here, just catching up on the logs, looks like you guys are going to be busy for a while! | 10:09.48 |
kens | Yeah the Smart Office stuff is eating resources :-( | 10:12.39 |
Shelly_ | buying tech from another company sounds good in the sales pitch but always comes with strings attached. | 10:15.15 |
kens | In fairness I think Robin and Paul tried to make that clear :-) | 10:15.38 |
chrisl | Argh, go out to buy coffee, and..... | 10:33.01 |
kens | and you missed Shelly :-( | 10:33.14 |
chrisl | Yeh, that's what I meant - I'll drop him a txt | 10:33.41 |
kens | SOrry, didn't realise you were out or I'd have told him | 10:33.58 |
chrisl | I didn't expect to be as long. Supermarket was *packed* - everyone seemed to be buying entire trolley-fulls of BBQ supplies! | 10:35.14 |
kens | Oh, World Cup fever I guess | 10:35.34 |
chrisl | And decent weather | 10:35.50 |
kens | Well, I cna't reproduce the 'missing text' problem either. My guess is they've changed something. | 10:37.01 |
chrisl | At least I'm not hallucinating | 10:38.21 |
kens | No, I checked carefully. There are some strange things happening with the display in Acrobat (flips from 'full size' to page mode at soem point for example, and the page scrolling is bonkers in some places) but that seems to be more of an Acrobat problem than anything else | 10:39.21 |
| Their command line doesn't work either (needs " " on the fielname) which doesn't exactly inspire confidence | 10:40.05 |
chrisl | I suspect it's because the PDF is the result of <ahem> "merging" several files together | 10:40.06 |
| They have "" around the file name | 10:40.36 |
kens | Hmm, you think that's what hey've done ? You could be right | 10:40.39 |
| Hmm, not on my email they don't | 10:41.05 |
chrisl | gswin32c -sDEVICE=pdfwrite -o gs.pdf -f "01 Classifier Gearbox.pdf.pdf" | 10:41.29 |
kens | Odd, mine has no "" | 10:41.36 |
chrisl | They are there for me in both plain text and "original HTML" views | 10:42.16 |
kens | Not for me :-) | 10:42.47 |
lama2014 | Hello | 11:28.20 |
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. | 11:28.20 |
lama2014 | How to investigate this: Error: /invalidfileaccess in --file-- | 11:28.48 |
| ? | 11:28.51 |
kens | You could report a bug, but the most likely problem is exactly what the error says 'invalid file access' either the fiel doesn't exist, or the path doesn' exist,or Ghostscript doesn't have permissions to acess it. | 11:29.53 |
chrisl | If you are running with -dSAFE or -dSAFER you could try leaving those options off | 11:31.17 |
lama2014 | chrisl i removed -dSAFER and then i got Error: /undefinedfilename in --file-- | 11:37.40 |
kens | So that tells you the file doesn't exist | 11:37.52 |
lama2014 | kens How to find out which file it's looking for? | 11:38.10 |
kens | O.O Its your PostScript isn't it ? SUrely you know..... | 11:38.25 |
chrisl | lama2014: the file name will on stack listing in the error output | 11:39.25 |
lama2014 | Operand stack: --nostringval-- (_region_.prv/tmp24295_5s/preview.ps) (r) | 11:39.32 |
| this one? | 11:39.38 |
kens | There you are then | 11:39.40 |
| THat's the filename | 11:39.53 |
lama2014 | this file does exist. That's strange... | 11:40.09 |
kens | THat file existis in the current directory ? | 11:40.24 |
| Or more accurately; the current directory has a sub-directory called '_region_.prv' which has a further sub-directory called '24295_5s' and there is a file in there called preview.ps ? | 11:41.56 |
| Err 'tmp24295_5s' | 11:42.15 |
lama2014 | v@laptop:~/tmp$ ls -al _region_.prv/tmp24736zNu/preview.ps | 11:42.51 |
| -rw-rw-r-- 1 v v 67540 Jun 13 11:41 _region_.prv/tmp24736zNu/preview.ps | 11:42.56 |
kens | THe 'tmp' directories differ | 11:43.42 |
| "/tmp24736zNu" and "/tmp24295_5s" | 11:44.09 |
lama2014 | i tried to work on my truecrypt ntfs filesystem and 5 mintes ago i moved everything to my ext3 home folder | 11:44.39 |
| the whole problem began when i installed acutex to emacs and now i'm trying to reproduce the problem that appeared inside emacs | 11:46.13 |
| that's the reason of the difference | 11:47.52 |
chrisl | But that means that the file gs is told to open doesn't exist | 11:48.43 |
kens | That's all pretty much Greek to me. As far as I cna see the reason the filename is not defined is because the file does not exist with teh spcified name/path | 11:48.49 |
| That's pretty much 'not our fault' really | 11:49.15 |
lama2014 | hmm... now i got gs ran successfully. Is that because -dSAFER was leaved off? | 11:51.16 |
kens | Aha, I see the customer has indeed been using a modified GS | 11:51.17 |
| lama2014 : Not a clue, but possibly yes. | 11:51.45 |
| -dSAFER prevents certain file accesses (see the documentation for details) | 11:52.34 |
chrisl | kens: I haven't seen a mail from the customer - possibly forgot to cc support (or maybe slow reaching me...)? | 11:53.24 |
kens | Oh let me look | 11:53.31 |
| Yeah they forgot to CC support, I'll forward it | 11:53.44 |
| So he was comparing a modified 9.07 against stock 9.14. THe problem only occurs on their modified 9.07 | 11:54.32 |
| Which of course makes it not unlikely that it will exist in a putative 9.14 modified the same way | 11:54.55 |
chrisl | Indeed - I think I'd wait for the "developer responsible for Ghostscript ports" to return from vacation..... | 11:55.49 |
kens | Well, it would have seemed most sensible to me.... | 11:56.02 |
kens | boggles to consider what they might have done that had this effect....... | 11:57.34 |
| coffee | 11:57.47 |
lama2014 | Thanks! | 13:35.37 |
kens | You're welcome | 13:36.09 |
rayjj | Robin_Watts: so my guess about the messed up Chinese characters in SO was correct -- the misuse of hinting to do parts of glyphs ? | 14:08.18 |
Robin_Watts | rayjj: Yes. But I'd already sent the customer a fix by the time you wrote to them suggesting what the problem might be. | 14:08.57 |
rayjj | Robin_Watts: does SO generally enable font hinting ? | 14:09.46 |
Robin_Watts | SO does no font hinting at all. | 14:09.56 |
rayjj | Robin_Watts: could that contribute to the layout differences ? I'm pretty sure that MS always uses the font hints | 14:10.54 |
Robin_Watts | rayjj: It could. If we were close enough to blame the hinting for the errors, I'd be happy :) | 14:11.26 |
rayjj | that makes sense. I didn't know how far off we were | 14:11.53 |
| (not really 'we', thankfully) | 14:12.48 |
Robin_Watts | I found the cause of significant differences in powerpoint this morning. | 14:12.55 |
rayjj | Robin_Watts: great ! | 14:13.03 |
Robin_Watts | But that's like saying: "I've found the problem with your car... it only has 3 wheels" etc. | 14:13.18 |
rayjj | I definitely don't envy you having to muck around in that | 14:13.44 |
el_mendi | Robin_Watts: I get an error that says library "libmaliinstr.so" is not found, do you have any idea of what that means? | 14:21.32 |
Robin_Watts | nafc, sorry. | 14:21.42 |
el_mendi | Is an android library :S | 14:23.23 |
| thank you any way | 14:23.34 |
Robin_Watts | libmali is for 3d acceleration, I think. | 14:26.18 |
rayjj | The guy (Ron) is on the phone and wants to know about adding something to the artifex | 14:59.02 |
| hi, ron | 14:59.49 |
Robin_Watts | hi ron. | 15:00.00 |
ronbo | hello | 15:00.15 |
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. | 15:00.15 |
jogux_mac | welcome ronbo | 15:00.40 |
ronbo | Hi. Ray suggested I hop on to inquire about getting some answers to questions... | 15:01.14 |
Robin_Watts | ronbo: Ask away. | 15:01.29 |
ronbo | Miles has asked me to help set up a forum for smart office, and later other artifex products ... | 15:01.37 |
rayjj | so, Ron is working for Miles and wants to know how / who can get phpBB up on wherever the artifex server lives | 15:01.50 |
Robin_Watts | ronbo: right, and you want to use phpBB for that. | 15:01.55 |
ronbo | AT the moment, I'm interested in installing phpBB on the artifex domain | 15:02.00 |
Robin_Watts | I would be really reluctant to see phpBB installed on our server. | 15:02.16 |
rayjj | Robin_Watts: I don't think I have admin for the artifex server -- you and miah were involved, iirc | 15:02.29 |
| Robin_Watts: do you know of a better thing to operate the forum ? | 15:03.01 |
ronbo | Miah gave me access to the CMS and for FTP, but not domain access | 15:03.03 |
henrys_ | have we nixed discourse.org? | 15:03.08 |
Robin_Watts | That is to say, I would really reluctant to see phpBB installed on casper.ghostscript.com (that's our central server) | 15:03.08 |
rayjj | Robin_Watts: artifex.com isn't on casper, is it ? | 15:03.29 |
Robin_Watts | but using phpBB on the godaddy hosting (that handles artifex.com) wouldn't offend me. | 15:03.47 |
rayjj | Robin_Watts: right. | 15:04.00 |
jogux_mac | robin: I think I'd tend to keep it separate from that. | 15:04.07 |
rayjj | and I agree that phpBB on casper would probably not be best | 15:04.23 |
| jogux: separate from artifex.com ? | 15:04.42 |
| jogux: why | 15:04.47 |
ronbo | phpBB could reside anywhere, so if there are security concerns we can find another location | 15:04.54 |
Robin_Watts | jogux_mac: Using the same hosting account, but a different top level domain should be OK, right? | 15:05.12 |
| so forums.artifex.com or something ? | 15:05.20 |
jogux_mac | rayjj: I wouldn't put it on the server server as anything; it's now for security issues, I would say it should be kept separate. ie. comprosing phpbb shouldn't let someone deface the main website. | 15:05.31 |
jogux_mac | makes an impressive number of typos/brainos :( | 15:05.47 |
| I'll just try that again from scratch | 15:05.58 |
rayjj | jogux: that's OK. I got the gist of it | 15:06.09 |
jogux_mac | ok :) | 15:06.14 |
ronbo | I thought I was a good touch-typist, but you guys are REALLY fast :-) | 15:06.27 |
paulgardiner_lap | It would be desirable if all that can be attacked via the forum is the forum | 15:06.29 |
jogux_mac | paul> yes | 15:06.37 |
rayjj | paulgardiner: agreed | 15:06.41 |
henrys_ | paulgardiner_lap: is phpBB what you wanted to use? Do you have a position on the forum? | 15:07.33 |
rayjj | so, do we need to have it actually hosted somewhere else ? | 15:07.41 |
ronbo | I kind of like the idea of keeping it completely separate | 15:07.44 |
jogux_mac | my only objection to phpbb is that it needs constant vigilance to keep it up to date and spam free; if someone is going to handle that I have no objection. | 15:08.15 |
Robin_Watts | If it was me, I'd open a new hosting account (I've used ix in the past, but there are many alternatives) and set it up on that. | 15:08.20 |
rayjj | it's easy (and cheap) enough to get a godaddy account to host it separately | 15:08.21 |
jogux_mac | the other option which is worth considering imho would be dedicated phpbb hosting. | 15:08.37 |
| https://www.phpbb.com/hosting/ | 15:08.59 |
Robin_Watts | That way we don't have to worry about attacks/leaks etc. | 15:09.06 |
ronbo | I thought discourse was attractive and functional, but it's relatively new. I found phpBB and discovered it's not only the most popular, there is a huge body of support information | 15:09.16 |
paulgardiner_lap | henrys_: I have no strong opinion. I was originally attracted to use it because is was so commonly used and users would be used to it | 15:09.18 |
henrys_ | jogux_mac: I like the hosting | 15:09.59 |
Robin_Watts | I don't know much about phpBB, except that it's known to be 'troublesome' with spam/attacks etc. But that may just be because it's been around long enough for there to be automated tools for attacking it. | 15:10.04 |
ronbo | I was reading that since versoin 3.0, security problems are largely a non-issue. But I would like to read more ... | 15:10.39 |
Robin_Watts | discourse looks newer/shinier, but whether that will subsequently prove to be a spam magnet, who knows... | 15:10.42 |
paulgardiner_lap | But then discourse looked good too. The attraction now is that it's Ron's preference | 15:10.45 |
henrys_ | I thought discourse.org looked good Iâve seen a lot of phpBB stuff that doesnât look like it belongs in this millenia | 15:10.55 |
paulgardiner_lap | I mean phpBB is Ron's preference | 15:11.00 |
Robin_Watts | http://www.discoursehosting.com/?gclid=CjgKEAjwwuqcBRCSuoivmIPnkwQSJACpqj3ku5HT6Ce_9cJeykZBlMt1L5cuTLVwM_darPBactExV_D_BwE | 15:11.25 |
jogux_mac | robin> oo, that's new. | 15:11.48 |
Robin_Watts | $20 a month, free trial. | 15:12.05 |
henrys_ | wow really new | 15:12.06 |
| I just looked for something like that a few weeks ago | 15:12.23 |
ronbo | really new | 15:12.48 |
Robin_Watts | I like shiny. shiny wins everytime! | 15:13.14 |
jogux_mac | :-) | 15:13.29 |
rayjj | so, could we have forums.artifex.com go to them ? | 15:13.30 |
Robin_Watts | rayjj: so they say. | 15:13.46 |
jogux_mac | how about rombo sets up that free trial then, and we bash at it for 10 minutes? | 15:13.54 |
ronbo | I have no preference yet. I want easy and a want functional. Miles instructions to me were: DO NOT MAKE WORK FOR THE ENGINEERS..... | 15:14.29 |
henrys_ | ronbo: now you can have miles pay you to manage a hosted site ;-) | 15:14.39 |
rayjj | ronbo: right, and this looks like we don't have to do anything | 15:14.54 |
ronbo | I can set up two demo sites, one with phpBB and one with discourse and we can try 'em out. | 15:15.06 |
Robin_Watts | ronbo: I second jogu's idea. Let's try the free trial. | 15:15.06 |
| sounds good. | 15:15.17 |
jogux_mac | I think we all know how phpbb works already tbh :) | 15:15.28 |
ronbo | But Miles doesn't. | 15:15.40 |
| Thanks for finding the discourse hosting site. I did not see that last week. (I probably did not look hard enought) | 15:16.15 |
el_mendi | Robin_Watts: I get the error on line 57 of this code: http://pastebin.com/yBGgr4VP | 15:16.34 |
Robin_Watts | ronbo: I rather suspect it wasn't around last week :) | 15:16.36 |
el_mendi | I don't know what I'm doing wrong | 15:16.57 |
henrys_ | ronbo: yes I looked too this is new | 15:17.04 |
ronbo | Thanks..... bye.... | 15:17.12 |
Robin_Watts | el_mendi: As indeed you should. | 15:17.16 |
| el_mendi: You have fz_point *pos = NULL; and you're doing: pos->x = blah. | 15:17.41 |
| fz_point pos; pos.x = blah; | 15:18.09 |
| and then later where you use pos use &pos; | 15:18.16 |
el_mendi | U.U" | 15:18.24 |
| thank you | 15:18.27 |
Robin_Watts | np. | 15:18.47 |
| ronbo: Come back anytime :) | 15:19.23 |
rayjj | ronbo: and when you get a test forum running, let us know here or email tech at artifex.com | 15:20.08 |
rayjj | goes back to work debugging the improvements for cust 532 ... | 15:20.43 |
ronbo | Thanks Robin. Now that I know where you all hang out, I will. I'll be back in touch with when I've got something to show. | 15:20.53 |
jogux_mac | btw, in case it's not clear, discoursehosting is /not/ run by / affliated to the official project at discourse.org - the official discourse hosting is still 'coming soon'. | 15:23.22 |
henrys_ | ronbo: I hope milesâ instructions donât make you hesitant to drop by. Usually a little chatting makes for a lot less work for engineers in the long run. | 15:24.26 |
ronbo | thanks Henry. I understand. | 15:25.01 |
henrys_ | kens:should I follow up with John today? | 15:27.46 |
kens | Chris is going to follow up on the build stuff | 15:28.05 |
| THe forms is going to take longer (possibly much longer) but he has a work-around so I don;t see it as urgent | 15:28.36 |
henrys_ | hallelujah on the build stuff ;-) | 15:28.50 |
el_mendi | Robin_Watts: Now I get the error in line 82: http://pastebin.com/MEbWnLgg | 15:34.38 |
| Just when I call to clean_page | 15:34.52 |
Robin_Watts | What error? | 15:35.08 |
el_mendi | the same of android library | 15:36.14 |
Robin_Watts | libmali? | 15:37.36 |
henrys_ | kens: surprising how many aix folks we seem to have | 15:38.24 |
el_mendi | That error gave me with the pointers error of before but after change it now the error is on this line | 15:38.27 |
| yes, libmali | 15:38.30 |
kens | henrys, surprises me..... | 15:38.43 |
Robin_Watts | el_mendi: libmali isn't anything to do with us. | 15:38.48 |
| Are you sure that control is not reaching my_process_fn? | 15:39.01 |
el_mendi | yes | 15:39.16 |
Robin_Watts | I can't help you here. | 15:40.12 |
el_mendi | OK thank you | 15:40.33 |
Robin_Watts | "Cannot find a rule to create target | from dependencies." That's their make util being old, right? | 15:46.36 |
chrisl | it's not gmake | 15:47.01 |
Robin_Watts | ok, so "use gmake" :) | 15:47.18 |
chrisl | yes | 15:47.25 |
mvrhel_laptop | kens2: hoping to have v2 profile creation out to you by early next week. | 17:48.49 |
| it is going to be very easy for you. | 17:48.57 |
| basically a single call to get a pointer to a buffer with the v2 profile data and the size of the data. you don't need to do any mem management either | 17:49.24 |
| as it will be part of the profile structure and released when the profile is destroyed | 17:49.45 |
el_mendi | Robin_Watts: The problem of the libmali was because my Android device has an MTK6592 processor, but running the aplication in a Nexus 4 gives me this error in the same point: http://pastebin.com/yu6t87Z9 | 17:56.41 |
mvdan | Hello, are any of the mupdf guys here? | 17:58.37 |
| Specifically anyone involved in the android version | 17:58.46 |
mvrhel_laptop | blah. a bit more code then I had hoped for the icc v2 creation stuf | 18:10.24 |
| bbiaw | 20:42.24 |
| need a break from icc conversion... | 20:42.33 |
| Forward 1 day (to 2014/06/14)>>> | |