| <<<Back 1 day (to 2012/10/01) | 2012/10/02 |
kens | Robin_Watts : ping | 09:14.34 |
Robin_Watts | pong | 09:23.11 |
kens | aha | 09:23.15 |
| robin got s funny message wheil trying to do a cluster run | 09:23.29 |
| In MingW suing gitpush.sh: | 09:23.37 |
| $ /usr/bin/gitpush.sh | 09:23.53 |
| warning: LF will be replaced by CRLF in clusterdatestamp. | 09:23.53 |
| The file will have its original line endings in your working directory. | 09:23.53 |
| error: git checkout-index: unable to create file gs/Resource/Init/pdf_main.ps (P | 09:23.53 |
| ermission denied) | 09:23.53 |
| fatal: Could not reset index file to revision 'HEAD'. | 09:23.53 |
Robin_Watts | ok. | 09:24.25 |
| Does that happen every time? | 09:24.31 |
kens | It only happened this time, I don't normally alter pdf_maion.ps | 09:24.47 |
Robin_Watts | It sounds like something is holding gs/Resource/Init/pdf_main.ps open. | 09:24.50 |
kens | Hmm, could be Visual Studio I guess | 09:25.10 |
Robin_Watts | or AV software. | 09:25.20 |
kens | Well I often cluster push with VS holding the modified file open | 09:25.39 |
| I suppose it could be AV software | 09:25.54 |
| SHould I try it again ? Do you know if my changes are being teted ? :-) | 09:26.10 |
Robin_Watts | (I've seen cases where if you change a file, AV software then jumps in to scan it. If you then try and change it again, you can't cos it's mid scan) | 09:26.32 |
kens | Well, I guess its possible :-) | 09:26.46 |
Robin_Watts | Well, it's testing something :) | 09:26.47 |
kens | Yes, I just don't know what its testing, exactly | 09:26.58 |
Robin_Watts | You can look at: /home/regression/cluster/gitbridge/mupdf | 09:27.10 |
kens | mupdf ? | 09:27.18 |
Robin_Watts | You can look at: /home/regression/cluster/gitbridge/ghostpdl | 09:27.28 |
| Your changes should be on the 'ken' branch there. | 09:27.37 |
kens | Yes, good point, let me go look | 09:27.47 |
Robin_Watts | Or log into peeves and check /home/marcos/cluster/users/ken/ghostpdl | 09:28.08 |
| to see what actually made it to the remote end. | 09:28.20 |
kens | I'll try peeves for absolute ertainty | 09:28.40 |
| Looks like my changes made it in | 09:31.01 |
| THanks Robin_Watts | 09:31.09 |
Robin_Watts | np | 09:38.35 |
| sebras, tor8: New version of those patches up. Only the "Refactor compressed stream loading" has changed since last night, which is to fix the function renaming that sebras spotted. | 10:00.06 |
sebras | Robin_Watts: I'll look at it in more detail tonight. if tor8 or you happen to merge it before that is ok with me. :) | 10:01.58 |
Robin_Watts | sebras: Thanks. | 10:02.36 |
| chrisl: Your bigtiff commit is tabtastic | 12:17.23 |
chrisl | Eh? git didn't give any warnings :-( | 12:17.48 |
Robin_Watts | gdevtfax.c at least | 12:18.14 |
| sorry. gdevtifs.c I mean | 12:18.29 |
chrisl | I wonder why the pre-commit checks didn't flag it, then..... | 12:19.02 |
Robin_Watts | pass. | 12:19.18 |
chrisl | Robin_Watts: gdevtifs.c I see one tab, gdevtfax.c has loads, but I didn't add them - is that what you see? | 12:24.00 |
Robin_Watts | TAB(void)fd; ? | 12:25.37 |
| TAB(void) pbase; | 12:25.48 |
| etc | 12:25.50 |
chrisl | Nope, four spaces here..... | 12:26.16 |
Robin_Watts | I pulled --rebase and it whinged at me about those lines. | 12:26.22 |
| OK, let me look again. | 12:26.57 |
| ok, so it must have been git having a paddy. No idea why. | 12:30.41 |
| sorry. | 12:30.47 |
chrisl | No probs. I've a compiler warning to fix, so I'll remove the tabs while I'm doing that | 12:31.13 |
Robin_Watts | chrisl: In general, we removed all *leading* tabs, but kept ones in the middle of lines. | 12:34.33 |
| Not entirely sure why. | 12:34.47 |
chrisl | So I see. Not sure about the fine line between that and "tab-in-indent", though...... | 12:36.00 |
Robin_Watts | I'd take a wild guess that the script to replace tabs in indents was easier to make sure nothing went wrong than one that replaced tabs everywhere in case someone had a tab in a string. | 12:37.31 |
chrisl | Could be, yes. I don't like tabs, but I *especially* don't like tabs *and* spaces, so I'm happy to have removed a few | 12:39.13 |
Robin_Watts | Gah. Horrible bug. Will only reproduce in non debug builds with multithreading. | 13:00.46 |
slestak_work | good day guys | 13:08.35 |
| i am using ghostpdl 9.06 on aic that i compiled myself with gcc | 13:09.07 |
kens | AIX ? | 13:09.25 |
slestak_work | im sry, yes, aix | 13:09.33 |
| no coffee yet | 13:09.37 |
| most of my sDEVICE args work, but I really need txtwrite to extract data from a pcl file | 13:10.10 |
| does not appear to be working. so I need to specify an encoding for latin-1? | 13:10.32 |
| i thought that might be the issue | 13:10.43 |
kens | No, almost certainly not | 13:10.54 |
| It works best if the PDF has a ToUnicode CMap | 13:11.08 |
slestak_work | the resulting output file has just whitespace, and strings of ^@ | 13:11.13 |
kens | Its likely that the PDF simply can't have the text extracted, because we don't know how to. | 13:11.34 |
slestak_work | this is ghostpDl, so I am going from pcl | 13:11.38 |
kens | I'd have to see the PDF to know htough | 13:11.41 |
| PCL ? Well no Unicode info then. | 13:11.53 |
| We assume its ASCII, if it isn't then we are stuck | 13:12.02 |
slestak_work | when I open the pcl in vim (with -b) I can see the text within it | 13:12.05 |
kens | Yes, because we assume its ASCII and write it that way | 13:12.33 |
slestak_work | I considered using re's to try to get the data. wonder if that will work | 13:12.44 |
kens | But again, I would have to see the PDF (or PCL, whatever) | 13:12.46 |
slestak_work | mind if I posit it up somewhere? it is small | 13:12.59 |
kens | Ofg course, just give me a URL | 13:13.09 |
slestak_work | thx, sec | 13:13.14 |
kens | Mind you, I don't speak very good PCL, I may have to ask for help from a colleague | 13:13.28 |
slestak_work | oh any insight at all is appreciated. | 13:14.04 |
kens | txtwrite is very new, so I would not be surprised if there are bugs. | 13:15.26 |
slestak_work | here is a link in dropbox. https://dl.dropbox.com/u/4384825/sbr.pcl | 13:15.30 |
| it is a 2 page pick ticket. | 13:15.42 |
kens | Just saving the file now | 13:15.48 |
slestak_work | well if i can help test and maybe patch, all the better | 13:15.53 |
| output to png and tiff work great | 13:16.02 |
kens | There is ASCII text there, certainly | 13:16.13 |
| Just give me a minute to rebuild ghostpcl | 13:16.29 |
slestak_work | this is the command I used. | 13:16.38 |
| pcl6 -LPCL -sDEVICE=txtwrite -sOutputFile=sbr.txt -dNOPAUSE sbr.pcl | 13:16.40 |
| i think the regexes might work, but I would rather parse it first | 13:17.14 |
kens | Thanks, that's helpful, people often don't give the command line | 13:17.26 |
| Binary building now | 13:17.42 |
slestak_work | i also considered maybe I need to tweak a dependency when building | 13:17.54 |
kens | There shouldn't be any dependencies with that code, its pretty simple | 13:18.10 |
slestak_work | i dont recall my version of gcc | 13:18.11 |
| i thikn 4.06 | 13:19.05 |
kens | Again, sholdn't be a problem, if it gets this far it ought to work | 13:19.31 |
slestak_work | here is what my resulting text file looks like in bash http://screencast.com/t/0eAapUBQIa | 13:19.37 |
kens | still building.... | 13:20.44 |
chrisl | Hmm, seems to produce readable text on Linux | 13:20.45 |
kens | interesting | 13:20.53 |
| Maybe it is a build problem then | 13:20.59 |
slestak_work | i bet it is text encoding | 13:21.21 |
chrisl | http://pastebin.com/H1MMMsAK | 13:21.24 |
slestak_work | seems to always be a problem with six | 13:21.29 |
| aix | 13:21.31 |
kens | There is no text encoding here really | 13:21.33 |
| Maybe AIX doesn't like our output of course | 13:21.46 |
slestak_work | if i strace it or sth, can we tell if I am extracting text properly? | 13:22.09 |
kens | Looks like artificail emboldening but otherwise seems correct chrisl | 13:22.17 |
| slestak_work: , hmm its tricky, but I can point you to some places you can put breakpoints if you have a full debugger | 13:22.41 |
chrisl | Yes. Could it be an endian issue? | 13:22.46 |
slestak_work | i am using gnu screen in bash | 13:22.47 |
kens | slestak_work : means very little to me | 13:22.56 |
slestak_work | i an not a pro with gdb, quite the noob | 13:22.58 |
kens | Me either, but I have to use it soemtiems.... | 13:23.08 |
slestak_work | i can try | 13:23.14 |
| this is a power5 proc, i think it is littleendian | 13:23.43 |
kens | Well, my output (Windows) matches chrisl's | 13:23.46 |
| I believe PowerPC is little-endian | 13:24.05 |
slestak_work | does anyone have a mac to try it on? pre intel mac | 13:24.08 |
kens | No, only Intel | 13:24.23 |
chrisl | slestak_work: powerpc workstations are usually bigendian | 13:24.24 |
kens | it shouldn't be an endian-ness problem, since this is all byte oriented. | 13:24.53 |
slestak_work | does the fact the tiff and png writer work prove out the endian thought? | 13:25.23 |
chrisl | kens: gs_char and gs_glyph are larger than bytes..... | 13:25.24 |
slestak_work | would .configure or any build output help? | 13:25.45 |
kens | I don't think we use them | 13:25.45 |
| You can put a breakpoint on txtwrite_process_plain_text | 13:26.15 |
| pte->text.data.bytes should contain the text | 13:26.34 |
| looks like it arrives one byte at a time | 13:26.50 |
chrisl | kens: does txtwrite accumulate text in "UTF" buffers? | 13:27.21 |
kens | First one should be 'R' then 'E' then 'L' | 13:27.39 |
slestak_work | in /gs/base/gdevtxtw.c? | 13:27.41 |
kens | yes | 13:27.45 |
| chrisl its more complicated than that :-( | 13:28.11 |
| it 'encodes' the character first | 13:28.20 |
| THen it tries to find a Unicode code point for it. | 13:28.47 |
| If that fails it treats hte text as ASCII | 13:29.05 |
chrisl | I was just wondering if it stores in buffers of multibyte codes, and then truncates it to ASCII | 13:29.18 |
kens | either way it adds teh Unicode to a char * buffer | 13:29.20 |
| Later on it outputs to the designated format | 13:29.37 |
slestak_work | can I just put the verb break inside the function? | 13:29.53 |
kens | Well, I usually use ddd | 13:30.16 |
chrisl | Me, too | 13:30.25 |
kens | Which I guess you can't use, so I am in the dark on gdb | 13:30.29 |
slestak_work | so you do it at runtime? | 13:30.31 |
kens | Yes, in the debugger | 13:30.38 |
slestak_work | is ddd an X tool? | 13:30.40 |
kens | Yes | 13:30.45 |
slestak_work | yeah, that is a no go | 13:30.50 |
kens | Its an X front end for gdb | 13:30.51 |
slestak_work | i am reviewing http://www.ofb.net/gnu/gdb/gdb_1.html | 13:31.23 |
chrisl | slestak_work: you should just be able to do "break txtwrite_process_plain_text" at the gdb prompt | 13:31.45 |
kens | I think you should also be able to inspect the contents of pte | 13:32.16 |
chrisl | I'm building on a G4 MacMini now - it will take a while...... | 13:32.41 |
slestak_work | alright, I am in gdb, see if I can stumble through this | 13:32.53 |
kens | accidentally built a Memento build | 13:32.59 |
slestak_work | complains about lack of symbol table | 13:33.35 |
| i dont think I have instantiated pcl6 yet | 13:33.46 |
kens | So you haven't made a debug build | 13:33.50 |
slestak_work | i can rebuild | 13:33.59 |
kens | try make debug | 13:34.04 |
chrisl | "make pcl-debug" will be quicker | 13:34.21 |
kens | :-) | 13:34.34 |
chrisl | Well, no sense building xps and language_bitch when all we need is pcl! | 13:35.41 |
kens | Another thought; try setting -dTextFormat=0 | 13:37.48 |
slestak_work | oh yeah, I didnt mention, i dont have the fastest machine around lol | 13:37.50 |
| i can do that in another terminal | 13:38.45 |
kens | OK, the output isn't immediately useful but it might give some insight | 13:39.09 |
| Instead of text being accumulated and a fake layout, it just gives the raw data | 13:39.36 |
chrisl | So, on my G4 MacMini I get a file full of garbage | 13:39.43 |
kens | OK that's interesting and suggests it *is* an endian-ness problem | 13:40.01 |
slestak_work | want me to pastibin a sample, | 13:40.16 |
kens | yes please | 13:40.22 |
slestak_work | looks like span and char records | 13:40.30 |
kens | first 20 lines or so is enough | 13:40.31 |
chrisl | kens: what's the format argument for UTF? | 13:40.31 |
kens | cannot remember | 13:40.38 |
| valid values are 0->3 | 13:40.51 |
slestak_work | http://paste.mitechie.com/show/815/ | 13:41.01 |
kens | Probably 2 for UTF-8 | 13:41.06 |
| Hmm that looks like NULLs | 13:41.31 |
slestak_work | is that 0, 1, and 2? 3 is not included right? | 13:41.34 |
kens | 3 is valid | 13:41.48 |
chrisl | I get the same - loads of NULLs | 13:41.48 |
kens | I have to guess its a short being read the wrong way round or something | 13:42.04 |
slestak_work | i dont see anythng but x0's | 13:42.11 |
| so the mac mini and aix are similar in this respect? | 13:42.47 |
kens | Yes, looks like it is an endian-ness problem | 13:43.03 |
| Trouble is, I haven't looked at this code in months | 13:43.12 |
slestak_work | do you have access to a machine to replicate the issue? | 13:43.55 |
kens | I don't, no | 13:44.16 |
chrisl | So, TextBuffer is an array of shorts | 13:44.40 |
kens | OK so that's probably the problem | 13:44.48 |
| Its probably reading them as bytes somewhere | 13:44.57 |
slestak_work | my debug build failed | 13:45.44 |
kens | chrisl format 3 is UTF-8 2 is UCS2 1 = XML == MuPDF 0 = debugging | 13:45.51 |
slestak_work | ld: 0711-593 SEVERE ERROR: Symbol C_BSTAT (entry 1487) in object ./debugobj/plmain.o: | 13:46.11 |
| The symbol refers to a csect with symbol number 0, which was not | 13:46.11 |
| found. The new symbol cannot be associated with a csect and | 13:46.12 |
| is being ignored. | 13:46.12 |
kens | TextFormat2 might be more interesting | 13:46.49 |
| As its a 2-byte format | 13:47.00 |
slestak_work | i'll rerun with format 2 | 13:47.00 |
chrisl | That gives me even more nulls | 13:47.27 |
slestak_work | yeah | 13:47.32 |
kens | OK well I'm guessing its the way the data is written (as its single bytes) then | 13:47.48 |
slestak_work | is there a make clean target? | 13:48.14 |
kens | chrisl if you can set berakpoints you could look in write_simple_text with -dTextFormat=3 and see wht hte buffer looks like | 13:48.22 |
chrisl | kens: write_simple_text() isn't being called | 13:49.38 |
kens | with -dTextFormat=3 ? | 13:50.01 |
chrisl | kens: yes | 13:50.24 |
kens | odd | 13:50.30 |
| let me look here | 13:50.56 |
| It gets called for me.... | 13:51.44 |
chrisl | Huh, the text format setting isn't getting through..... | 13:52.46 |
kens | boggles | 13:52.55 |
| How are you getting different output then ? | 13:53.06 |
| The default is 3.... | 13:53.18 |
chrisl | No, it's working now - I had a second TextFormat parameter in the debugger command line | 13:53.46 |
kens | <relief> | 13:53.53 |
chrisl | Yep, there is a bunch of bitwise ands and bitshifts which will be endian specific :-( | 13:55.38 |
kens | In where ? | 13:55.53 |
chrisl | write_simple_text | 13:55.58 |
kens | OK but I don't think that is the problem | 13:56.07 |
| because the decorated text is producing NULLs too | 13:56.19 |
chrisl | For exmaple: UTF8[0] = *UTF16 & 0xff; | 13:56.26 |
kens | Hmm, yes, I must have been in a huirry when I wrote that | 13:56.53 |
slestak_work | this is way over my head so I am being productive elsewhere ;) | 13:58.23 |
kens | chrisl can you put a breakpoint at get_unicode and see what is getting stored in the buffer of shorts ? | 13:59.46 |
| "*Buffer = unicode" | 14:00.06 |
| at the end of the function | 14:00.13 |
chrisl | Ah, lots of zeros - that probably explains the nulls, then | 14:00.59 |
kens | Yes, that's what I was wondering | 14:01.07 |
| The unicode values seem incorrect before we even get to emitting the text | 14:01.19 |
| That suggests that font->procs.decode_glyph iis also endian dependent | 14:01.42 |
| And I don't think I wrote that one :-) | 14:01.59 |
chrisl | Well, it ends up in pl_decode_glyph() which just returns "last_char" | 14:02.30 |
kens | Seems it just returns 'last_char' | 14:02.33 |
| Lol | 14:02.38 |
| So at base its a PCL interpreter problem, possibly exacerbated by problems in txtwrite. | 14:02.59 |
| last_char is extern of course :-( | 14:03.12 |
| Wheee, does that mean last_char is a global ? | 14:04.16 |
chrisl | It seems so :-( | 14:04.30 |
kens | Oh dear :-) | 14:04.37 |
chrisl | It defaults to "0" and I think it should be GS_NO_CHAR | 14:04.54 |
kens | sdeems to be in plchar.c | 14:04.56 |
| It shouldn't remain at 0, it should be set in p_tt_encode_char, it is for me. | 14:06.17 |
| pl_tt_encode_char | 14:06.25 |
| THe code is passing around a uint and treating it as a gs_char | 14:07.07 |
tor8 | I fucking hate this apple bullshit crap! | 14:07.23 |
chrisl | Ooopsee - that's not a txtwrite issue, then! | 14:07.29 |
tor8 | tim cook can go die in a fire! | 14:07.32 |
kens | Oooh, tor8 is *really* cross today | 14:07.46 |
slestak_work | glad i didnt stow my pitchfork | 14:07.48 |
sebras | kens: he was fine at lunch, this must be a sudden case of misspelling by mountain lion. | 14:08.13 |
kens | chrisl I think this is a bug for henrys, though its entirely likely that I'll need to modify txtwrite as well. | 14:08.14 |
tor8 | yeah. mountain lion just made my old wireless base station i worthless piece of plastic. | 14:08.29 |
chrisl | kens: yeh, I agree - one for Henry | 14:08.43 |
sebras | tor8: your linksys? | 14:08.53 |
tor8 | sebras: my apple airport express | 14:09.09 |
kens | chrisl I will write up a bug report, mention that tis txtwrite related and ask henrys to re-assign it to me when his bit is fixed. | 14:09.19 |
tor8 | "This version of AirPort Utility doesn't support this base station. Use AirPort Utility 5.6 to configure or manage this base station. Download version 5.6 at apple.com." | 14:09.25 |
| mountain lion ships with version 6.1 | 14:09.37 |
kens | You cna't downgrade ? | 14:09.57 |
tor8 | "AirPort Utility can't be installed on this disk. The version of Mac OS X on this volume is not supported." | 14:10.00 |
kens | Ah... | 14:10.08 |
chrisl | kens: okay. If you remind me, I'll test it before you start hacking on it..... | 14:10.09 |
kens | chrisl OK thanks | 14:10.16 |
tor8 | so, no. I can't downgrade and I want to reformat my mac book pro to install linux. if only I didn't have to maintain that damn iOS app. | 14:10.42 |
chrisl | tor8: I wonder if you can run MacOS in a virtual machine on Linux on Apple hardware? | 14:11.38 |
slestak_work | kens: want me to submit a bug report, not sure if I could describe it properly | 14:11.40 |
Robin_Watts | tor8: http://www.macworld.com/article/1167965/mountain_lion_and_the_ancient_airport_base_station.html | 14:11.48 |
| It seems you can still use that version. | 14:11.55 |
sebras | tor8: they also mention this link in the comments: http://support.apple.com/kb/DL1482?viewlocale=en_US but maybe you already have that exact version..? | 14:13.19 |
tor8 | if anybody says "apple stuff 'just works'â¢" I'm going to shoot them. | 14:13.44 |
| sebras: trying that one now | 14:13.59 |
Robin_Watts | tor8: It does. As long as you're prepared to continually lubricate the process with money. | 14:14.12 |
tor8 | sebras: it doesn't. same error. too new osx. | 14:14.27 |
slestak_work | is it worth getting a new base station (security, features, etc) | 14:14.37 |
sebras | tor8: apparent 5.6.1 is available for win7 too..? maybe that's an option for you? | 14:14.40 |
tor8 | Robin_Watts: it does, but only for 3 months, you mean ;) | 14:14.47 |
sebras | Robin_Watts: so does linux, but you need to use another lubricant. :) | 14:15.23 |
Robin_Watts | AirPortUtility5.6.1.dmg ? fed into http://frank.is/mountain-lion-and-the-old-airport-utility/ | 14:15.29 |
sebras | tor8: https://launchpad.net/ubuntu/+source/airport-utils maybe this is an option too..? | 14:15.54 |
kens | slestak_work : I've opened a bug for this problem here: | 14:17.25 |
| http://bugs.ghostscript.com/show_bug.cgi?id=693362 | 14:17.26 |
slestak_work | tyvm | 14:17.39 |
tor8 | the problem with installing any apple software on windows is it drags along a ton of other crap, like apple software update and bonjour | 14:17.40 |
kens | If you want you can add yourelf to the cc list. | 14:17.40 |
slestak_work | i have been singing your channels praises in my lug channel | 14:18.05 |
kens | Thank you :-) | 14:18.15 |
| Now I need some more coffee...... | 14:18.25 |
radistao | hi, | 14:35.18 |
| can anybody help me in PostScript | 14:35.36 |
| i convert JPEG->PDF | 14:35.46 |
kens | Neither is PostScript :-) | 14:35.55 |
| and I am in the middle of answering your stack overlfow post | 14:36.17 |
radistao | you mean this one ? | 14:36.33 |
| http://stackoverflow.com/questions/12690554/how-to-set-pdf-page-size-in-postscript-from-viewjpeg-ps-file | 14:36.34 |
kens | yes | 14:36.40 |
radistao | ok. thx. i'll wait | 14:36.53 |
chrisl | kens: that won't work as the procedure is bounded by save/restore | 14:42.51 |
kens | You can put it inside the proc, or remove save/restore | 14:43.16 |
| Unlike Kurt, I don't usually provide a completely canned solution | 14:43.43 |
| Because, to be frank, I don;t usually have the time | 14:43.57 |
chrisl | I was thinking of adding a "viewJPEGgetsize" procedure | 14:45.59 |
kens | Yes coudl do. | 14:46.06 |
radistao | In which place i have to add your answer? inside viewjpeg.ps ? | 14:47.44 |
kens | yes | 14:48.13 |
henrys | I must stop beginning my morning with that customer - have to save him for later in the day. | 14:53.45 |
tor8 | Robin_Watts: the commit message for the image params, shouldn't that be fz_compression_params rather than fz_image_params? | 14:53.55 |
kens | henrys after first coffee :-) | 14:54.03 |
Robin_Watts | tor8: probably. will check in a mo. | 14:54.21 |
chrisl | or eighth coffee, maybe....... | 14:54.21 |
slestak_work | kens: want to see a second oddity with that pcl file? | 14:56.15 |
| if you dont, that is cool. I cornered a good portion or your and your team members morning already | 14:56.39 |
kens | slestak_work : depends on what it is.... | 14:56.52 |
| I am not a PCL expert | 14:57.15 |
slestak_work | if you use the pdfwrite device, the glyphs in the detail section are black boxes | 14:57.17 |
| same font as in the header | 14:57.29 |
kens | On AIX ? | 14:57.31 |
slestak_work | yes | 14:57.34 |
kens | Have you tried any other platform ? | 14:57.44 |
slestak_work | i can render one and dropbox a sample | 14:57.46 |
kens | And what are you using to view the PDF file ? | 14:57.52 |
chrisl | They might be truetype notdef glyphs | 14:57.59 |
kens | indeed | 14:58.04 |
slestak_work | sumatra | 14:58.09 |
| (which uses mupdf) | 14:58.16 |
kens | Hmm, well that's MuPDF, so tht should be OK | 14:58.21 |
slestak_work | it is a ttf I am using | 14:58.31 |
kens | The PCL file contains a tt font, yes | 14:58.55 |
chrisl | radistao: here's what I came up with the other evening: http://pastebin.com/U1JcZG44 | 14:59.00 |
slestak_work | pcl-> pdf was my original target, but settled on pcl -> tiff with the font issue (and it works) | 14:59.05 |
| i'll try it on my ubuntu machine and report back | 14:59.32 |
chrisl | I would expect a little endian platform to produce better results | 14:59.59 |
slestak_work | its not a testing target. im used to seeing breakage on aix | 15:00.27 |
henrys | paulgardiner:ready for the meeting? Sounds like the odds for a working android demo before next week are pretty low but I think we should still keep pushing for a demo and get it out there as soon as possible. Is everyone on board with that or have new priorities arisen? | 15:00.27 |
chrisl | AIX is a PITA :-( | 15:00.53 |
slestak_work | +1 | 15:01.01 |
Robin_Watts | henrys: I think that's still our best plan. | 15:01.01 |
| My reading of the situation was that the big work for the demo was in a making a 'fast redrawing' version. | 15:01.49 |
paulgardiner | henrys: Yes, that sounds sensible to me. | 15:01.52 |
Robin_Watts | paulgardiner: How huge are the changes for a slow redrawing one ? | 15:02.09 |
| I'd hope that the pain of doing a full redraw shouldn't be too huge for simple forms. I mean, that's what the iOS one does anyway, right? | 15:02.53 |
paulgardiner | Robin_Watts: possibly not that huge. I was just going to say that we might have some sort of demo by the end of the week, but really very basic and slow. | 15:03.00 |
Robin_Watts | OK. | 15:03.14 |
paulgardiner | I have calc.pdf sort of almost working. | 15:03.45 |
Robin_Watts | in what way sort of almost? | 15:04.07 |
henrys | well I'll take the tablet to chicago anyway to show off mupdf and if something interesting breaks while I'm in Chicago send me email and a pointer to a download. Of course if we have something sooner that's great too. | 15:05.37 |
paulgardiner | Working other than clicking the calc buttons also toggles the menu (and sometimes causes a page flick), plus there are some strange bugs in the redraw I'm trying to understand | 15:06.52 |
| I haven't yet looked at input (calc.pdf has no input fields). | 15:07.18 |
Robin_Watts | Ah, right. | 15:07.33 |
kens | slestak_work : I've just realised that pdfwrite uses the same code to retrieve the glyph that txtwrite does, and will suffer the same problem on a big-endian machine | 15:07.33 |
Robin_Watts | Well that sounds like good progress anyway. | 15:07.41 |
slestak_work | kens: mean little endian? | 15:08.00 |
kens | Ohh, actually I see a problem with it on the PC too | 15:08.19 |
slestak_work | kens: err, aix | 15:08.21 |
kens | I suspect raster ops | 15:09.19 |
paulgardiner | henrys: How long are you in Chicago? | 15:09.28 |
henrys | Sat. - Weds. | 15:09.53 |
slestak_work | so you dont need my ubuntu build to proceed? | 15:10.01 |
Robin_Watts | Show is Monday -> Wednesday? or Sunday -> Wednesday? | 15:10.11 |
henrys | paulgardiner:anyway don't sweat it let's just keep moving forward. We certainly don't want V8 static to be a long term solution do we? | 15:10.47 |
Robin_Watts | henrys: What other option is there? | 15:11.02 |
paulgardiner | henrys: not sure about v8. Judging by the apk it didn't enlarge the app as much as I was expecting. | 15:11.42 |
Robin_Watts | If the copy of v8 on the device isn't exported through a standard interface, we have no option but to provide our own copy. | 15:11.51 |
henrys | I wonder how chrome is using it? | 15:12.09 |
Robin_Watts | If we ever sell mupdf to a phone maker, they can arrange for the system v8 to be exported and we can use that. | 15:12.29 |
henrys | is there a way to look at shared library usage on these devices by a program like ldd on linux. | 15:12.49 |
| ? | 15:12.50 |
Robin_Watts | chrome on the desktop? Includes its own v8.dll, I'd imagine. | 15:12.58 |
| henrys: Well, ldd will probably run. Android is linux, right? | 15:13.10 |
paulgardiner | We might need a rooted device to do it on | 15:14.03 |
Robin_Watts | As luck would have it, I have an HTC Desire surplus to requirements now. | 15:14.23 |
| So we can root that etc as required. | 15:14.29 |
| (Helen just upgraded her phone) | 15:14.59 |
paulgardiner | I saw some forum posts saying things like "Yes I tried SpiderMonkey and never got it working. V8 just worked out of the box" | 15:15.02 |
henrys | anyway I guess the static thing isn't an issue if you aren't seeing a noticeable slowdown in startup. To bog doesn the pdf interpreter with functionality used .001 percent of the time is not so great. | 15:15.35 |
Robin_Watts | We could probably build our own v8.dll rather than make a static lib. | 15:17.01 |
| But then we'd need some magic to load it on first use. | 15:17.10 |
paulgardiner | henrys: Oh yes. I forget that. It seems to slow down the start up of a debug build on windows, but actually it seems to make no difference on android. | 15:17.25 |
henrys | well then let's not worry about it for now. | 15:17.50 |
paulgardiner | I hadn't even thought about start up speed because it just started pretty much instantaneously as usual | 15:18.18 |
slestak_work | kens: so is there anything I can provide or do regarding pdfwrite device? | 15:19.06 |
kens | slestak_work : that looks to me like its caused by raster ops in the PCL | 15:19.26 |
slestak_work | was your statement saying "its the same problem as the other bug"? | 15:19.34 |
kens | No its a different problem | 15:19.43 |
| Since it exhibits on teh PC | 15:19.53 |
slestak_work | oh, ok | 15:19.58 |
kens | It cannot be 'fixed' as PDF does not support raster ops | 15:20.14 |
| All you cna do is make an image of the file | 15:20.22 |
slestak_work | ok. I know the src of the pcl, and it is highly suspect | 15:20.38 |
| very bad aix printing solution | 15:20.49 |
kens | I wouldnt' say its 'wrong' just 'silly' | 15:20.51 |
henrys | paulgardiner:anything else for this meeting? seems like you are going along great, tripping over this kind of stuff is expected actually I'm surprised there is so little of it considering the maturity of the mobile development environment. | 15:21.00 |
slestak_work | kens: tyvm | 15:21.07 |
kens | You don't need a raster op to produce this text, but a number of print drivers use them even when not required | 15:21.16 |
paulgardiner | henrys: no, nothing else for the meeting. I'll see if I can get you some sort of demo for Chicago. | 15:22.36 |
Robin_Watts | paulgardiner: Would it help for me to dig out Helens HTC Desire for you to root? Probably not for the demo, but for after that... | 15:23.27 |
paulgardiner | Might well be. I've been nervous of rooting mine. | 15:24.29 |
Robin_Watts | yeah, I wouldn't want to root any device you expect to get mobile coverage on. | 15:24.49 |
henrys | tor8 - anything for the meeting? | 15:26.17 |
tor8 | henrys: so we're going with slow redraw everything for the demo? | 15:28.11 |
Robin_Watts | tor8: That's the shortest path to a working demo, right? | 15:28.33 |
paulgardiner | tor8: for now. | 15:28.41 |
henrys | is it really that bad? | 15:29.15 |
tor8 | yes. it's probably okay for text-only pages. add images and you'll suffer. | 15:29.37 |
| so probably good enough for a demo | 15:29.50 |
paulgardiner | For calc.pdf on my Galaxy S2, I'm seeing maybe a 1/2 second delay for the update. | 15:29.53 |
Robin_Watts | tor8: even with images, we should be OK cos of the caching. | 15:29.59 |
| complex line art though... | 15:30.06 |
henrys | anyway it should be good enough. | 15:31.13 |
paulgardiner | Since working on the app again, I've been thinking doing the fast version might not be so hard, provided we don't mind using twice the bitmap memory per page. | 15:31.31 |
| But there are lots of fiddly changes involved. Not possible for Chicago. | 15:33.11 |
henrys | paulgardiner:fair enough, let's call it. We'll all be around for a while anyway we have another meeting in 20 minutes. | 15:36.34 |
paulgardiner | Ok. I'll let you know how it's going as the week progresses. | 15:37.36 |
Robin_Watts | curses gdb and goes to make tea. | 15:38.33 |
chrisl | Robin_Watts: you can use curses with gdb for an ASCII based pseudo-gui....... ;-) | 15:44.31 |
Robin_Watts | chrisl: It's not the gui. It's what it's showing me. | 15:45.34 |
henrys | chrisl, kens:I saw PCL come up in the logs should I read that stuff or is there a quick summary or bug? | 15:46.10 |
Robin_Watts | gscms_get_link is called with memory=0x1098320 | 15:46.22 |
henrys | oh I see the bug. | 15:46.28 |
kens | henrys new bug assigned to you | 15:46.32 |
chrisl | Just an unfortunate use of "curses" opened up the scope for a poor pun! | 15:46.38 |
Robin_Watts | cmsCreateTransformTHR is then called with ContextID=0x200e788 | 15:46.46 |
| chrisl: Ah! | 15:47.04 |
| but ContextID = (void *)memory so something is going wrong somewhere! | 15:47.25 |
chrisl | Robin_Watts: gdb does have watch points..... | 15:49.31 |
Robin_Watts | threadtastic code. | 15:49.46 |
| so it doesn't go wrong in the same way every time. | 15:50.11 |
chrisl | Robin_Watts: Ah, threads are evil | 15:50.47 |
| Robin_Watts: what about helgrind? | 15:52.50 |
Robin_Watts | chrisl: I will try that if I have to. | 15:53.04 |
| I suspect the correct value IS being passed, just gdb is lying to me (possibly because optimisations are not disabled) | 15:53.35 |
chrisl | Or there's another valgrind tool called DRD aparently for threading, too | 15:54.05 |
Robin_Watts | yeah. From my limited reading DRD is not as advanced as helgrind. | 15:54.26 |
chrisl | Yeh, but it's supposed to be considerably quicker - allegedly...... | 15:54.51 |
henrys | I'm not quite following the endian theory gs_char is just a long I don't see any place where bytes would be getting swapped around is this something chris discovered debugging on his machine? Am I missing something? | 15:55.05 |
kens | henrys well why is the value 0 ? | 15:55.28 |
| It isn't on a little-endian platform | 15:55.38 |
chrisl | henrys: bigendian, 32 bit system - "long" is 32 bit, gs_char is 64 bit | 15:56.08 |
kens | meeting time | 16:00.19 |
| morning mvrhel | 16:00.50 |
henrys | so we pass a uint to a procedure with a parameter declared as a long I still don't see why that would create 0 on a big endian system. | 16:00.51 |
mvrhel | good morning | 16:00.54 |
henrys | howdy. | 16:00.55 |
kens | henrys from my POV I see a 0 being passed int, I was assuming that the problem was hte cast, if its not I don't know what it is, but the value should not be 0 | 16:01.24 |
| And is not 0 on Linux or Windows | 16:01.39 |
chrisl | henrys: I'll debug more on the G4 tomorrow if you like | 16:01.53 |
henrys | kens:okay well I'll have a look certainly the parameters shojld match anyway. | 16:02.03 |
| mvrhel:do you have anything new and interesting I can talk about at chicago. Typically color stuff is important at this kind of show? | 16:02.57 |
mvrhel | henrys: I just sent you an email about things that I am working on for the customer. The big things though are the separation capabilities and also the use of NCLR ICC profiles for these devices | 16:03.58 |
henrys | or if anyone else has thoughts about things I should say when someone says "what's new", let me know. | 16:04.02 |
| so no meeting next week or at least I won't be here. | 16:06.16 |
chrisl | henrys: since when do we count as "new" for this show? | 16:07.23 |
henrys | everyone is quiet today. | 16:07.26 |
kens | Nothing to say really. | 16:07.39 |
henrys | I didn't really have anything for the meeting. | 16:08.10 |
mvrhel | henrys: I dont have much. Just lots to do for 330 | 16:08.16 |
henrys | well anything else. | 16:08.17 |
chrisl | henrys: I have made changes to the PS interpreter so that the CET tests "pass" when we're using 64 bit ints (based on CPSI mode) | 16:08.44 |
henrys | I still don't understand what the customer I'm talking to is doing, doesn't make any sense. | 16:08.55 |
kens | I don't understand him either | 16:09.17 |
| his emails make no sense to me | 16:09.29 |
henrys | I think I'll just work on his other bugs and see if he trips himself up arguing seems futile. | 16:10.04 |
ray_laptop | henrys:which customer is that ? 190 ? (that's something I don't follow) | 16:10.30 |
marcosw | I had a question about encryption key length. for the export license question: as long as we only support key lengths that are less than 56 symmetric, 512 asymmetric, and112 elliptic curve we can self classify. Do anyone know if this is the case? I emailed this question to tech a while ago but didn't see any replies. | 16:10.42 |
henrys | tor8:the viewer ... I forgot to ask at the last meeting? | 16:10.44 |
tor8 | henrys: soon usable. | 16:10.53 |
henrys | tor8:ah great. | 16:11.10 |
ray_laptop | marcosw: does it apply only to encryption or to decrypt as well ? | 16:11.28 |
henrys | marcosw:what is the latest aes stuff I thing Robin_Watts addes that had the largest key size I believe. | 16:11.54 |
marcosw | encryption | 16:11.59 |
Robin_Watts | henrys: Size isn't important. | 16:12.16 |
marcosw | ray_laptop: wait, maybe not. | 16:12.28 |
Robin_Watts | And I don't remember making any AES changes myself. | 16:12.42 |
| sebras added stuff for V6 in mupdf. | 16:12.51 |
marcosw | the document says "Encryption using key length", but now I'm not sure what that means. | 16:13.16 |
henrys | mvrhel:okay I should be able to say something at the show about this business. | 16:15.03 |
mvrhel | henrys: ok. it basically amounts to a hyper level of color control based upon object type | 16:15.45 |
henrys | well that's interesting. | 16:16.32 |
| I don't want to bore our fans. | 16:17.07 |
Robin_Watts | I'm out tomorrow at Apps World. | 16:17.27 |
henrys | alexcher:how's the memory stuff coming any questions I should send luratech? | 16:18.16 |
ray_laptop | mvrhel: 'hyper level' ??? | 16:18.19 |
| mvrhel: is there any extension to the object types we classify ? or what ? | 16:18.57 |
alexcher_ | henrys: I'm about to commit. There are a few indeterministic problems that keep me from commiting. | 16:19.33 |
henrys | alexcher:great | 16:19.58 |
ray_laptop | mvrhel: I have had one customer years ago that wanted to be able to tag shadings (because it changed their trapping). | 16:20.16 |
henrys | marcosw:did you see the thread about the CCITT g4 data, their customer claims it is good data. | 16:20.34 |
mvrhel | hyperlevel meaning an over abundance | 16:20.56 |
| ray_laptop | 16:20.59 |
marcosw | henrys: I saw the bug reports. Their tiff file is certainly good. | 16:21.04 |
mvrhel | control to the extreme | 16:21.08 |
Robin_Watts | Ah, while I remember; artifex.com | 16:21.14 |
mvrhel | in terms of options to do stuff | 16:21.14 |
henrys | kens, chrisl:yes that global is rude, is that something done differently in the new font integration? | 16:21.36 |
mvrhel | ray_laptop: no additional object types | 16:21.43 |
Robin_Watts | mace was on here last week talking about the new website, and the requirements it had for the hosting. | 16:21.53 |
mvrhel | image, graphics and text is still fine with this customer | 16:21.56 |
chrisl | henrys: I haven't changed it, no. I can do so, if you like | 16:21.58 |
kens | henrys no, there's nothing new there. | 16:21.58 |
Robin_Watts | We were discussing it with ray_laptop and then he fell off irc. Did that conversation ever get completed offline? | 16:22.18 |
| mace pointed out that artifex.com resolved to 4 different servers, 3 of which were dead. | 16:22.37 |
henrys | shouldn't mace just report that to miles. | 16:23.23 |
Robin_Watts | possibly, but it's all tied up in the same issue. | 16:23.42 |
| Is the hosting that we have for artifex.com capable of hosting the new site - and it's ray that has the keys to that particular cabinet. | 16:24.19 |
ray_laptop | Robin_Watts: XO has multiple servers and play a common DNS trick to internally load balance their networks. | 16:24.47 |
Robin_Watts | I got the impression that Miles had pointed mace to ray as it was getting technical, but I could be wrong. | 16:24.59 |
| ray_laptop: Right, but having 3 of the 4 servers broken seems a bit crap. | 16:25.17 |
ray_laptop | Robin_Watts: what are the IP addresses you see (from europe) | 16:25.34 |
| Robin_Watts: how are you checking 'broken' ? | 16:25.51 |
Robin_Watts | Well, I've not been able to reach artifex.com for weeks. | 16:26.06 |
henrys | I don't think you need a lot of horsepower to run artifex.com if that is the concern. | 16:26.17 |
Robin_Watts | henrys: It needs MySQL and PHP. | 16:26.30 |
chrisl | ray_laptop, Robin_Watts: and artifex.com was unreachable for me for a few months last year, although it's been fine recently..... | 16:26.54 |
Robin_Watts | Mace has asked ray to confirm that our hosting package has MySQL and PHP, and to date, I have not seen a definitive answer from ray. | 16:26.55 |
| chrisl:still down for me. | 16:27.03 |
ray_laptop | Robin_Watts: we could just create an amazon instance | 16:27.09 |
Robin_Watts | mace suggested a simple test for ray to do, but I haven't heard the response. | 16:27.21 |
ray_laptop | Robin_Watts: I thought we determined that we _don't_ have mysql | 16:27.33 |
chrisl | Robin_Watts: well, you could reach it when it was down for me - it must just be ISP, or which way the wind's blowing.... | 16:27.53 |
Robin_Watts | (It's possible that all this has been sorted offline, and I haven't seen it, in which case tell me to shut up_ | 16:27.53 |
| chrisl: No, I've not been able to reach it at all for ages. | 16:28.11 |
henrys | I'm curious what we would need mysql for? | 16:28.18 |
Robin_Watts | ray_laptop: I did not see that determination. | 16:28.20 |
| henrys: The CMS. | 16:28.23 |
ray_laptop | Robin_Watts: mace proposed a the simple test for phpinfo, but I didn't see the test for mysql | 16:28.27 |
chrisl | Robin_Watts: But you definitely could when it was down for me | 16:28.43 |
Robin_Watts | Content Management System; so Miles/Scott can post stuff without hacking html. | 16:28.49 |
| chrisl: not recently then. | 16:28.57 |
henrys | Robin_Watts:okay thanks just curious. | 16:29.07 |
ray_laptop | can marcos just fire up an amazon instance and we'll dump XO ? | 16:29.09 |
Robin_Watts | ray_laptop: I didn't see the results of the phpinfo | 16:29.09 |
| ray_laptop: An amazon instance would be a bad usage. | 16:29.23 |
chrisl | Robin_Watts: Well, last year was, erm, a year ago or more...... | 16:29.24 |
ray_laptop | Robin_Watts: why so ? | 16:29.40 |
Robin_Watts | Cost. | 16:29.44 |
| The instance would need to run 24/7 and we'd need to pay for transfer bandwidth. | 16:30.02 |
ray_laptop | I thought the low end instance was about $150 a month or so | 16:30.06 |
henrys | it's 10:30 the meeting in formally over, thanks, once again I won't be here next week. | 16:30.19 |
Robin_Watts | Right, and we can get a whole years hostage with ix or someone for $150! | 16:30.26 |
| ray_laptop: Did you run the phpinfo test? | 16:30.50 |
| Cos for me, that shows the MySQL details. | 16:30.57 |
marcosw | henrys: and I don't think we need a support meeting, only two new bugs. | 16:31.06 |
ray_laptop | Robin_Watts: well Miles pays about $100/mo for XO (and they only allow 300Mb of storage for that cost) | 16:31.09 |
Robin_Watts | jeez. Time to get off that horse... | 16:31.20 |
| http://www.wss.co.uk/test.php | 16:31.24 |
henrys | marcosw:agreed. | 16:31.39 |
Robin_Watts | I pay something like $7 a month to ix for all my sites with at least 4Gig of storage. | 16:31.44 |
ray_laptop | Robin_Watts: I would be happy to switch. | 16:32.43 |
Robin_Watts | ray_laptop: Did you run the phpinfo test on artifex.com? (Not that I can see the results, dammit :( ) | 16:32.50 |
ray_laptop | Robin_Watts: I'm searching for the email with that test in it. | 16:33.22 |
chrisl | Robin_Watts: do you know what mace used to test the accessibility of the IP addresses? | 16:33.50 |
marcosw | You can get an amazon micro instance of $0.005/hr (which works out to $3.60/month). That does require a $100 reservation charge, good for three years. So the actually cost is $6.50/month. OTOH, that doesn't include bandwidth and I don't know if a micro instance has enough power to run artifex.com. | 16:34.49 |
Robin_Watts | chrisl: I don't, sorry | 16:34.53 |
chrisl | Oh, never mind, I can ping all four - I really thought they'd disable ping..... | 16:35.04 |
ray_laptop | oops. Was just about ti reply to marcosw :-( | 16:36.32 |
Robin_Watts | We could just stick artifex.com onto my ix hosting. It won't cost me anything. | 16:37.05 |
ray_laptop | for the logs, I don't think anything we do on artifex.com requires a lot of bandwidth or compute power. It's not like we have videos or anything | 16:37.15 |
Robin_Watts | The 'expert plan' on ix is $4 a month. | 16:38.13 |
| 'Infinite' disk space/bandwidth/domains. | 16:38.32 |
| We fell foul of that when we put downloads.ghostscript.com on there, because they don't really mean 'infinite bandwidth' :) | 16:38.57 |
ray_laptop | Robin_Watts: the hassle is switchcing the DNS records in a way that doesn't cause a glitch where it can't be found. | 16:39.00 |
Robin_Watts | ray_laptop: As opposed to the existing situation where it can't be found at all at the moment for me ? :) | 16:39.33 |
| We can set up the new hosting, then when it's ready, swap the IPs. | 16:39.50 |
| In the day or so it takes to propagate, we#ll serve from both places. | 16:40.09 |
| Then we can cancel the XO hosting. | 16:40.21 |
| But we don't need to use ix. Don't we already have a godaddy account? | 16:40.36 |
| (Actually, I'm sure we do, I think it's on my credit card :) ) | 16:40.57 |
chrisl | Our downloads are hosted on godaddy | 16:41.24 |
Robin_Watts | yeah. | 16:41.29 |
chrisl | And realistically, the hosting overlap will likely be more like a month than a couple of days | 16:42.00 |
Robin_Watts | OK. Let me work with mace to pull the new website onto godaddy.com | 16:43.49 |
| and then when it's up and working, we can flip the switch? | 16:44.02 |
ray_laptop_ | Robin_Watts: I think we should set up the *same* (old lame-o) website so people don't get confused if the site flips back and forth depending on the DNS resolution | 16:44.25 |
Robin_Watts | I'd rather set the new one up so we can be sure everything works before flipping the switch. | 16:45.02 |
ray_laptop_ | Robin_Watts: I have some comments (mistakes) for the new website content | 16:45.09 |
Robin_Watts | Have you mailed them to mace? | 16:45.19 |
ray_laptop_ | Robin_Watts: no, haven't finished reviewing all of it | 16:45.41 |
| Robin_Watts: 'flipping the switch' takes a couple of days to settle down, right ? (48 or 72 hours) | 16:46.23 |
Robin_Watts | Something like that. In practise it's always taken <24 for me. | 16:46.44 |
henrys | marcosw or robin_watts:do either of you have the url from which you retrieved the draft pdf spec, I know I told you guys about the line but now I can't find it. | 16:53.51 |
Robin_Watts | henrys: I don't have the URL for where we bought it. | 16:54.20 |
henrys | s/line/link | 16:54.29 |
Robin_Watts | henrys: http://www.iso.org/iso/iso_catalogue/catalogue_ics/catalogue_detail_ics.htm?ics1=37&ics2=100&ics3=99&csnumber=53041 ? | 17:03.45 |
kens | OK I'm off now, goodnight all | 17:07.40 |
henrys | ah thanks robin_watts. | 17:16.38 |
ray_laptop | Robin_Watts: I ran phpinfo and the results are at: http://www.artifex.com/phpinfo.htm (if you can get to it) | 17:39.50 |
Robin_Watts | I cannot :( | 17:40.08 |
ray_laptop | Robin_Watts: I don't see anything about mysql, but there is a line "DBM: This is GDBM version 1.7.3, as of May 19, 1994." | 17:41.15 |
Robin_Watts | ray_laptop: I'm fighting godaddy at the moment. | 17:41.32 |
ray_laptop | Robin_Watts: try http://69.16.143.110/phpinfo.htm | 17:41.41 |
Robin_Watts | No luck. | 17:42.15 |
ray_laptop | Robin_Watts: or http://66.152.109.110/phpinfo.htm | 17:42.20 |
Robin_Watts | It's virtual hosted, so lots of websites come from the same servers. | 17:42.34 |
ray_laptop | Robin_Watts: that is _very_ strange | 17:42.35 |
henrys | Does anyone understand why gp_open_scratch_file - in particular why doesn't it unlink(filename) after mkstmp. Isn't that the normal usage? So that when the program exits or the file is closed it is deleted? | 17:42.36 |
Robin_Watts | hence unless you have the domain name in the title, it won't know what domain to serve you. | 17:42.54 |
ray_laptop | henrys: because that is used by the clist code and it needs to close and re-open by name. I tried fdopen(fileno(c_file)) and it fails on Windows | 17:43.55 |
| Robin_Watts: oh, that's right. nm. | 17:44.21 |
henrys | ray_laptop:how is the file ever deleted? | 17:44.39 |
| and can't your rewind? | 17:45.12 |
ray_laptop | henrys: explicitly with a call to unlink | 17:45.14 |
Robin_Watts | I can add artifex.com into the downloads.ghostscript.com hosting (so it will also appear as downloads.ghostscript.com/artifex, but I don't suppose that matters). | 17:45.55 |
henrys | oh I see open scratch return the filename it generates along with the file pointer. | 17:46.06 |
| okay thanks. | 17:46.14 |
Robin_Watts | All users of downloads.ghostscript.com ever normally see is downloads.ghostscript.com/public | 17:46.16 |
ray_laptop | henrys: Windows doesn't allow us to open for w+ and then allow other threads to open (as above) for "r". It has to be closed and re-opened (iirc). | 17:46.30 |
Robin_Watts | The only problem is that the new website assumes it's in the root directory :( | 17:46.40 |
mace | ok just about caught up with scrollback | 17:46.52 |
Robin_Watts | Aha. Hi mace/ | 17:47.00 |
ray_laptop | Hi, mace. Can you see the phpinfo.htm ? | 17:47.13 |
henrys | ray_laptop:okay thanks | 17:47.19 |
mace | henrys: i've not reported the IP problem any further than mentioning it here, mostly because it's not something i have any control or input on :) | 17:47.34 |
ray_laptop | mace: can you see the phpinfo at: http://www.artifex.com/phpinfo.htm | 17:47.51 |
mace | ray_laptop: yeah; top line says it all for me.. php 3! | 17:48.24 |
ray_laptop | Robin_Watts: what DNS do you get ? | 17:48.31 |
| mace: what do you need ? | 17:48.37 |
mace | you really need to find a better hosting provider :/ | 17:48.46 |
Robin_Watts | mace: How about: http://downloads.ghostscript.com/artifex/test.php | 17:48.54 |
mace | php 5 ideally, thats the current maintained version | 17:49.07 |
ray_laptop | Robin_Watts: mace: I agree. Telling XO to take a hike would be fine with me. | 17:49.31 |
mace | for what its worth, and i'm sure there are plenty of reasons why this wasn't considered but mentioning it for "the logs" :) i do hosting | 17:50.42 |
| Robin_Watts: your php version is sane, and has mysql support. would need to do further tests to check mysql server is there and sensible version, but that looks positive | 17:51.16 |
ray_laptop | I'll tell Miles that we will be switching from XO and see if he has any particular hosting outfit he wants to use (we have godaddy for some of the domain registrations), or if he wants to just leave it up to us (likely) | 17:51.19 |
mace | nods.. highly recommend moving from XO from what you've said so far.. had to look this up from wikipedia, apparently php 3 was last released end of 2000! | 17:52.56 |
| i'm all for stability, but that's taking it a bit far :D | 17:53.14 |
Robin_Watts | mace: We have a hosting thing on godaddy. | 17:53.47 |
| We can therefore put artifex.com on there without it costing us anything extra. | 17:54.05 |
mace | should work in theory. i've put a cake app on godaddy previously. just make sure that package has mysql support | 17:54.48 |
Robin_Watts | I'm sure it does. | 17:54.57 |
mace | probably billed as number of databases | 17:55.15 |
Robin_Watts | ray_laptop: Are you in control of the artifex.com nameservers ? | 17:55.18 |
| Can you point test.artifex.com to 188.121.46.128 please? | 17:56.03 |
ray_laptop | sorry, was on the phone with Miles (but he got interrupted with another call) | 17:56.08 |
mace | Robin_Watts: i've just checked all of the ip addresses for artifex.com andhttp://www.artifex.com, they're working for me atm | 17:56.45 |
| working insomuch as telnet IP 80 results in a connection | 17:56.59 |
| which is better than the other day | 17:57.06 |
| Robin_Watts: assuming its not working for you just now, can you do a dns lookup on your machine? | 17:57.30 |
| see if you're getting the same addresses | 17:57.42 |
Robin_Watts | 207.155.248.{31,47} and 207.155.250.20 and 207.155.252.31 | 17:58.47 |
mvrhel_laptop | bbiaw | 17:59.39 |
mace | Robin_Watts: can telnet to all of those | 18:00.45 |
Robin_Watts | mace: MySQL is definitely available. | 18:01.57 |
| Do I need to create a database ? | 18:02.02 |
mace | urm, yeah | 18:02.17 |
| you going to set up a test site? | 18:02.29 |
Robin_Watts | I've wget'd from artifex.dsis.co.uk | 18:02.46 |
mace | chuckles | 18:02.54 |
| you won't need a database for that | 18:03.10 |
Robin_Watts | COnfused then. | 18:03.40 |
mace | you need the source code :) | 18:04.00 |
Robin_Watts | Oh, right, of course. | 18:04.13 |
| Can you make the source available somewhere as a zipfile and then I can upload it? | 18:04.30 |
mace | yeah just sorting that now | 18:04.41 |
Robin_Watts | Ta. | 18:04.44 |
| ray_laptop: Are you off the phone now? | 18:05.31 |
mace | Robin_Watts: zip or tar/gz? | 18:09.00 |
Robin_Watts | zip for preference. | 18:09.07 |
| I know the godaddy tools cope with zip. | 18:09.15 |
tor8 | Robin_Watts: tor/master has mubusy -> mutool rename | 18:56.58 |
Robin_Watts | tor8: cool. I've got to fix a commit message referring to fz_image_params, right? | 18:57.43 |
| Other than that were you happy with my changes? | 18:57.49 |
tor8 | Robin_Watts: and sebras function rename (fz_free_compressed_buffer, were there any other?) | 18:58.20 |
Robin_Watts | I did that. | 18:58.31 |
tor8 | ah, forgot to fetch recently | 18:59.02 |
| Robin_Watts: I haven't looked at the fz_function commit yet | 19:00.05 |
| the rest looked fine | 19:00.12 |
Robin_Watts | tor8: OK. thanks. | 19:00.16 |
tor8 | Robin_Watts: oh, I need to amend the manpage too in that mubusy commit | 19:02.09 |
Robin_Watts | ok. I'm web bashing with mace at the moment, so I won't be looking immediately. | 19:03.39 |
tor8 | Robin_Watts: I've amended, so whenever you have time. | 19:03.58 |
| no rush, we've got half a dozen branches sitting on origin/master waiting to be pushed and rebased it looks like :) | 19:04.19 |
Robin_Watts | Hi ray. | 19:31.54 |
| We have the website up and running on the godaddy hosting now. | 19:32.05 |
ray_laptop | Robin_Watts: OK. What's the URL ? | 19:36.38 |
| Miles interrupted the call, then never called me back. :-( | 19:37.06 |
Robin_Watts | ray_laptop: Well, it will respond to test.artifex.com and artifex.com, but only if the DNS is changed. | 19:38.01 |
| I've added a line in my hosts file locally to test it. | 19:38.18 |
| I sent you an email with the DNS details in. | 19:38.47 |
| If you update artifex.com to point to that address, we'll swap over. | 19:39.05 |
henrys | hmm shelly applying zeniko patches ugh | 19:40.26 |
| something to deal with after lunch. | 19:42.32 |
sebras | tor8: two patches on sebras/master | 20:18.02 |
slestak_work1 | henrys: for bug 693362, can I view the repo so I can patch my install to test on aix? | 20:23.20 |
| nm, found the git repo | 20:24.14 |
| would be cool to either replace the sourceforge cvs mirror of gs with a git mirror or just remove the csv one | 20:25.19 |
| henrys: do these casts to uint need to be gs_char also? http://goo.gl/VGLwW | 20:31.00 |
henrys | slestak_work1:yes I'll fix that. | 20:42.10 |
| slestak_work1:has the bug changed? | 20:45.17 |
| oh actually those uints should be gs_glyphs. | 20:47.57 |
ray_laptop | Robin_Watts: BTW, if I can be of help on the MT SEGV, let me know. It's more interesting than DNS and web hosting stuff | 20:54.55 |
Robin_Watts | ray_laptop: I'm still bashing the webhosting. | 20:55.19 |
ray_laptop | Robin_Watts: OK. np | 20:55.29 |
Robin_Watts | ray_laptop: Does that mean you've changed the DNS? | 21:00.57 |
slestak_work1 | henrys: i wonder if kens has updates to make to reflect these changes | 21:01.13 |
| i'll certainly volunteer my aix box to help test this | 21:01.32 |
henrys | I was just told there was a type mismatch and they were seeing a bad parameter (0) on big endian machines. It seems to me unlikely the type mismatch was the problem but we'll see. | 21:04.32 |
ray_laptop | Robin_Watts: no, I am having log-in problems. | 21:06.15 |
slestak_work1 | they specifically said the array that should have held the text (one char at a time) was all Nulls | 21:06.32 |
| worked fine on win and lin, failed on POWER5 aix and g4 mac mini | 21:07.00 |
ray_laptop | Robin_Watts: so do you want to add test.artifex.com, or switch the DNS over to godaddy ? | 21:07.07 |
Robin_Watts | Add test.artifex.com for now. | 21:07.17 |
slestak_work1 | i'll be patient and talk to ken tomorrow. | 21:07.25 |
| ty so much for your work. | 21:07.33 |
ray_laptop | Robin_Watts: OK. I'll try and get that done this afternoon | 21:07.41 |
Robin_Watts | In order to make artifex.com secure, I've had to take downloads.ghostscript.com down, supposedly for just a minute, but... godaddy have screwed the pooch somehow and the change isn't going through, | 21:08.12 |
henrys | are we still against // comments can't remember I never use them in the gs code. | 21:27.05 |
| ? | 21:27.11 |
mace | Robin_Watts: take it you've not bottomed out the godaddy problem yet? | 21:32.05 |
elhoyos | time | 22:25.48 |
Robin_Watts | mace: I'm doing the 'non-root accessible' thing. | 22:56.19 |
mace | Robin_Watts: not going to plan i'm guessing | 23:30.10 |
Robin_Watts | mace: Well, downloads.ghostscript.com is no longer the primary thing. | 23:30.33 |
| It's served from /downloads | 23:30.39 |
| artifex.com is working, and I hope should not now be exposing anything. | 23:30.50 |
| but downloads.ghostscript.com isn't working any more :( | 23:31.00 |
mace | want me to take a look? | 23:32.34 |
Robin_Watts | I think it's working now. | 23:36.31 |
mace | happy days | 23:40.23 |
Robin_Watts | mace: hmm. The funky highlighting images aren't working. | 23:42.00 |
mace | oh that is odd.. they were ok a moment ago | 23:44.19 |
| might've been cached though | 23:44.25 |
| its looking for an /icons/ folder in root | 23:44.58 |
| although it could be a virtual folder too, depends on how they've configured it | 23:45.30 |
Robin_Watts | what is looking for an icons folder? | 23:45.44 |
mace | the highlighting images are fetched from /icons/ | 23:47.29 |
| but they're not web accessible | 23:47.40 |
| ie. the markup is ok, but the files are misplaced | 23:47.55 |
Robin_Watts | really? That's not how I was reading the page code. | 23:48.05 |
| but OK. What do I need to put where? | 23:48.14 |
mace | i'll recheck, i was using firebug | 23:48.19 |
| yeah markup is full of img tags | 23:48.45 |
Robin_Watts | <a class="fadehover" href="/page/printer-products.html"><div style="background-image: url(/theme/Artifex/img/printer.png)"><img src="/theme/Artifex/img/printer-blue.png" width="111" height="148" alt="Printer / MFPs" /></div><span>Printer / MFPs</span></a> | 23:49.25 |
| I read that as: background = /theme/Artifex/img/printer.png foreground = /theme/Artifex/img/printer-blue.png | 23:49.52 |
| and then the foreground fades out on a hover. | 23:50.01 |
| both those are available. | 23:50.05 |
mace | oh are you on about the test site? | 23:50.18 |
| i'm on about downloads :) | 23:50.28 |
Robin_Watts | Sorry, I was on about artifex.com | 23:50.40 |
mace | downloads is broken too then | 23:51.15 |
Robin_Watts | really? downloads looks OK to me. | 23:51.50 |
| gah. now it doesn't. | 23:52.29 |
mace | *chuckle* | 23:52.41 |
| rollovers are working for me after a forced reload | 23:54.18 |
| on artifex website, before i confuse things further | 23:54.50 |
Robin_Watts | And now they are for me too. | 23:55.00 |
| Something is clearly unhappy with caching. | 23:55.07 |
| Uploading the icons from my copy of apache now. | 23:55.25 |
| ok, that seems to be working now. | 23:56.10 |
mace | yeah good here | 23:56.43 |
Robin_Watts | OK, I'm going to bed while the going is good :) | 23:57.14 |
mace | hehe enjoy | 23:57.20 |
Robin_Watts | Thanks! | 23:57.21 |
| Forward 1 day (to 2012/10/03)>>> | |