| <<<Back 1 day (to 2015/08/27) | 20150828 |
tomty89 | chrisl: About urw-core35-fonts. So I see that in the two commits in June, "L"/"No9 L" is added two the "Nimbus Sans"/"Nimbus Roman" families. But then in the latest commit "Nimbus Sans L Regular" become "Nimbus Sans Regular" again; and "Nimbus Roman No9 L Regular" suddenly become "Nimbus Sans L Regular". Are these just mistakes or do they happen for a reason? | 07:14.36 |
| chrisl: I am only checking the afm on the git web though, not sure about the pfb and pfm | 07:15.08 |
chrisl | tomty89: erm, that sounds wrong... I'll check it out. Some of the names have changed, though | 07:16.53 |
tomty89 | chrisl: because I already requested fontconfig to patch against the new names in "1.10" and it's already upstream, so I just want to make sure that I wasn't too reckless to do that. https://bugs.freedesktop.org/show_bug.cgi?id=85225 | 07:33.38 |
| chrisl: They won't remove the old mapping soon though. | 07:33.43 |
chrisl | tomty89: sorry, I don't have control over the names - we (try to) use the fonts as they arrive | 07:34.31 |
tomty89 | Oh I see. So that change wasn't your call. | 07:35.12 |
| okay got to go. hope that the inconsistency/weirdness in the latest commit will be fixed soon. | 07:39.03 |
sh4rm4^bnc | chrisl, do you have the backlog available and time to look into the problem i encountered ? | 08:58.08 |
chrisl | sh4rm4^bnc: I can look - but cross compiling with shared libs isn't something I've looked at so far | 08:59.09 |
sh4rm4^bnc | yeah it appears crosscompiling isn't quite working atm | 08:59.35 |
chrisl | It requires manual intervention to tweak the Makefile | 08:59.56 |
sh4rm4^bnc | <sh4rm4^bnc> btw base/genarch base/genconf base/echogs need to be compiled with $HOSTCC rather than $CC | 09:00.18 |
| <rayjj> sh4rm4^bnc: but genarch is problematic since running that on the host WON'T give the right arch.h for the target | 09:00.20 |
| seems even with manual intervention the arch.h will be messed up | 09:00.36 |
chrisl | Right, so there's a makefile variable called AUXCC which is the compiler used to compile the host based tools | 09:01.00 |
| Also, in the Makefile is a variable called TARGET_ARCH_FILE - if that is set (to a predefined arch.h file), genarch will simply copy that predefined arch.h into the build | 09:02.24 |
| Looking at the configure stuff, I'm not at all sure why lcms2 wouldn't be found, since we are using the pre-defined macro to check for it | 09:10.59 |
sh4rm4^bnc | did you see that in my recipe lcms and lcms2 are rm -rf'd ? | 09:22.32 |
| that part is taken from http://www.linuxfromscratch.org/blfs/view/svn/pst/gs.html | 09:27.15 |
chrisl | sh4rm4^bnc: sure, which is why we end up using the predefined macro to check for a library | 09:28.03 |
sh4rm4^bnc | is that an m4 macro ? | 09:28.33 |
chrisl | Yes - AC_CHECK_LIB() | 09:28.52 |
| Er, thought you were using gs 9.16..... ? | 09:30.09 |
sh4rm4^bnc | in my case, lcms2 installs its headers into $prefix/include and its libs into $prefix/libs as it should | 09:30.13 |
| yeah | 09:30.14 |
chrisl | Well, there is no lcms directory..... | 09:30.34 |
sh4rm4^bnc | so the headers and libs are in the compilers default dirs | 09:30.40 |
| nothing special is needed to include them | 09:30.52 |
| and -llcms2 is sufficient to link against it | 09:31.06 |
chrisl | In which case, the configure check should find it | 09:31.38 |
sh4rm4^bnc | checking for local lcms2 library source... no | 09:32.33 |
| checking for system lcms2 library... checking for cmsGetTransformOutputFormat in -llcms2... yes | 09:32.33 |
| checking lcms2.h usability... yes | 09:32.33 |
| checking lcms2.h presence... yes | 09:32.33 |
| checking for lcms2.h... yes | 09:32.34 |
chrisl | So, it did find it | 09:32.46 |
sh4rm4^bnc | btw when i still had lcms2 2.5 installed the configure check accepted the lib but later i got compile errors | 09:33.47 |
| due to use of a type that was introduced in later versions | 09:34.02 |
chrisl | That's not surprising..... | 09:34.14 |
| lcms2 has been rather a moving target | 09:34.36 |
sh4rm4^bnc | cmsPluginMutex | 09:38.40 |
| so it may sense if the configure script checks for that | 09:38.58 |
| *may make sense | 09:39.05 |
chrisl | Sure, I can add that in. | 09:39.24 |
Robin_Watts | Didn't we add that? :) | 09:39.42 |
chrisl | I think we did, yes | 09:39.59 |
| BTW, I just tried here, and linking to a shared lcms2 works fine...... | 09:40.01 |
sh4rm4^bnc | yeah the link works | 09:40.15 |
Robin_Watts | (or we were the instigators behind Marti adding it at least) | 09:40.21 |
sh4rm4^bnc | but the rule in base.mak leads to a bogus inclusion of -I/include | 09:40.36 |
chrisl | base.mak ???? | 09:41.09 |
sh4rm4^bnc | <Robin_Watts> hence $(II)$(LCMS2SRCDIR)$(D)include$(_I) is turning into -I/include | 09:41.27 |
| base/lib.mak | 09:41.48 |
Robin_Watts | base/lib.mak. | 09:41.49 |
chrisl | Erm, my shared lcms2 doesn't have cmsPluginMutex in it, but Ghostscript linked against it just fine | 09:42.20 |
sh4rm4^bnc | but there's code that uses it | 09:42.39 |
| so the compile fails | 09:42.47 |
chrisl | Nope, for me, it linked fine | 09:43.03 |
sh4rm4^bnc | ./base/gsicc_lcms2.c:static cmsPluginMutex gs_cms_mutexhandler = | 09:43.41 |
| it fails at compile time due to an unknown tyoe | 09:44.01 |
| that type exists only in the headers | 09:44.56 |
| it's not an external symbol | 09:45.06 |
chrisl | It should be - it's part of the LCMS2 API | 09:45.32 |
| I need to poke around at this, work out what's a good symbol to check for | 09:49.29 |
| I think possibly _cmsCreateMutex() is a decent option | 09:51.45 |
| sh4rm4^bnc: surely the -I/include is benign? | 09:52.55 |
sh4rm4^bnc | or just do #include <lcms2.> / <stdio.h> int main() { cmsPluginMutex foo; printf("%zu", sizeof(foo); return 0;} | 09:53.13 |
| if the type is unknown the compile should fail | 09:53.44 |
chrisl | I prefer to use the stock macros where it's viable | 09:53.46 |
sh4rm4^bnc | ok | 09:54.32 |
| <chrisl> sh4rm4^bnc: surely the -I/include is benign? | 09:54.36 |
| hmm ? in my case it's "malicious" as it pulls in the host headers | 09:54.53 |
chrisl | Oh, okay..... | 09:55.22 |
sh4rm4^bnc | which have differently sized types (32 vs 64 bit) | 09:55.30 |
chrisl | Fixing that is going to mean quite a few changes | 09:59.25 |
sh4rm4^bnc | :( | 09:59.56 |
chrisl | Every third party lib is going to need an explicit include directory, as well as the currently used source directory | 10:01.05 |
| Actually, it's worse than that, every third party library will need an explicit "-I/search/here" value, not just the include directory :-( | 10:04.18 |
sh4rm4^bnc | hmm | 10:05.08 |
| isnt a conditional if the LCMS2 dir is set sufficient ? | 10:05.22 |
| like | 10:05.29 |
chrisl | We can't use conditionals | 10:05.34 |
| Robin_Watts: did you not add the gproof device to the unix build? | 10:05.39 |
Robin_Watts | chrisl: I don't think it's in by default, no. | 10:06.28 |
chrisl | Hmm, well that explains part of Fred's problem | 10:07.00 |
| Huh, but you added it to the Windows build...... | 10:07.26 |
Robin_Watts | Adding gproof to the unix device involves touching configure scripts, right? bugger that. | 10:08.36 |
| s/unix device/unix build/ | 10:08.50 |
chrisl | I'll apply the same logic next time someone wants something in the Windows build changed...... | 10:09.06 |
Robin_Watts | I mean, I happily hack things in for my own use, but... :) | 10:09.11 |
chrisl | Hrm, I'm not keen on the internal device name being different from the actual device name string..... | 10:14.44 |
Robin_Watts | chrisl: Possibly it should be the gprf device. | 10:18.21 |
| I'd not be offended by that change. | 10:18.29 |
chrisl | Robin_Watts: that would be preferable, IMO.... | 10:31.12 |
Robin_Watts | chrisl: I'll do that later today. | 10:31.35 |
chrisl | Robin_Watts: let's make sure Michael and Fred are fully up to speed with it, too | 10:32.19 |
| sh4rm4^bnc: so, as I tried to say before I was ejected: I wonder if we could hack around the include path problem by setting the source directory to a dummy that we know couldn't possibly contain header files | 10:33.08 |
| sh4rm4^bnc: actually, that is basically how the other third party libraries work when doing a shared lib build, so I guess it's workable..... | 10:56.00 |
sh4rm4^bnc | chrisl, sounds good | 10:57.48 |
chrisl | sh4rm4^bnc: Okay, I've committed and pushed those changes | 11:11.26 |
sh4rm4^bnc | chrisl, nice. do you have a link to the git commit (web itf) handy ? | 11:11.56 |
chrisl | sh4rm4^bnc: http://git.ghostscript.com/?p=ghostpdl.git;a=summary | 11:12.15 |
| sh4rm4^bnc: as for setting the CCAUX variable from the configure command line, I know what needs done for that, but it's a more involved change, so I'd rather not look at the until after the pending release | 11:14.07 |
sh4rm4^bnc | chrisl, is there any specific reason you call it AUXCC rather than HOSTCC which is pretty much standard ? | 11:17.16 |
| (HOSTCC is used by the kernel, busybox and pretty much anything that supports x-compile OOB) | 11:17.40 |
chrisl | sh4rm4^bnc: probably because the person who added it wasn't aware of HOSTCC | 11:17.58 |
sh4rm4^bnc | i see | 11:18.11 |
chrisl | In either case, we have to take action to set it, so the name isn't very important | 11:18.36 |
| sh4rm4^bnc: and it makes sense since we build those tools in "obj/aux", so it's the "CC for AUX" | 11:20.50 |
| I have to post a letter - bbiaw..... | 11:22.56 |
Robin_Watts | chrisl: Say when you get back... | 11:29.55 |
tor8 | kens: You mentioned you had found another airline for the Nassau to Freeport leg? | 12:00.50 |
kens | tor8 yep | 12:00.59 |
| Foudn 3 in total | 12:01.07 |
| BahamasAir gets *really* bad reviews | 12:01.22 |
| There's also Western Air | 12:01.35 |
| But although their timetable says they fly in the evening, you can't book such a flight | 12:01.52 |
| So we're going with SkyBahamas | 12:02.00 |
| You might want to consider carefully whether you feel like risking it | 12:02.16 |
tor8 | Yeah. The mailboat backup is a 9h trip :) | 12:02.37 |
kens | We figure worst case the flight is cancelled (or we miss it) and we end up staying the night in Nassau. | 12:02.51 |
| tor8 there's a hgh speed ferry too at 3 hours | 12:03.00 |
| but it departs once a day at 1:30 pm | 12:03.09 |
tor8 | ah, that's more reasonable. | 12:03.17 |
kens | Troulbe is, the BA flight lands at 3 pm | 12:03.31 |
| We're booked on SkyBahamas (reviews were OK) on a 33 seat propeller plane | 12:03.55 |
| Departing at 17:30 | 12:04.06 |
tor8 | kens: Thanks. Going to look at booking tickets for me and sebras soon, so knowing which options you picked is helpful. | 12:04.40 |
kens | If we miss that we might (reluctantly) try for the BahamasAir flight at 18:00. Failing all that, we find a hotel in Nassau and get a flight the next morning | 12:04.46 |
| tor8 if you need teh flight details I can dig them out. | 12:05.44 |
| Oh yeah the last point is that there's no BA flight on the Monday | 12:06.11 |
| So we're flying Sunday and paying an extra night | 12:06.22 |
tor8 | The travel agencies I've looked at only seem to have BahamasAir. I can't find any SkyBahamas or Western Air flights. | 12:06.28 |
kens | I booked direct :-) | 12:06.38 |
| -dUseCIEColor -dNOPAUSE -sDEVICE=jpeg -dFirstPage="1" -dLastPage=2 -sOutputFile=outputImage_%0d_A.gif -dJPEGQ=100 -r300 -q inputFile.There are also (at least) 2 Western Air outfits | 12:06.54 |
| OOps, paste error | 12:07.00 |
| http://www.skybahamas.net/ | 12:07.22 |
| Going via the US you will get more choice, and more reliability | 12:08.03 |
tor8 | kens: Biggest problem there is the immigration. Sebras has never been to the US, so I fear we won't be able to use the new kiosks that Robin mentioned. | 12:09.08 |
Robin_Watts | tor8: No, you won't be able to. | 12:10.06 |
kens | Well the immmigration is a turn off. We also thought it might be fun (possibly in the Chinese sense) to fly between islands in a small plane | 12:10.11 |
tor8 | The reliability might outweigh that concern though; it's going to be a very long trip no matter what and I don't cope well with trouble when I'm that tired. | 12:10.11 |
Robin_Watts | But in theory... if lots of people are using the kiosks, the normal immigration line should be shorter too. | 12:10.32 |
tor8 | kens: Ah, more trouble. The LHR - NAS flight departs at 10 in the morning, which is a bit too early for connecting flights from Copenhagen. | 12:13.10 |
kens | tor8 looks like you'll want the Miami flights | 12:13.30 |
tor8 | It does look that way :( | 12:13.44 |
Robin_Watts | THe LHR->MIA flight I'm on leaves at 9:10 | 12:14.00 |
kens | Shame, we'll let you know how the little aircraft experience went | 12:14.02 |
jogux | me & Pete are expecting an overnight stay in heathrow on the wya out I think | 12:15.45 |
kens | ick | 12:16.02 |
Robin_Watts | jogux: Presumably elaine and yukko are coming? | 12:17.19 |
jogux | I think Elaine is. Yukko is undecided currently I think. | 12:17.35 |
kens | free trip to the sun in winter, undecided ? O.O | 12:17.54 |
jogux | It's a problem planning things 3 months in advance when we've got two cats undergoing regular and unpredictable vet treatment :( | 12:18.26 |
Robin_Watts | kens: They get quite enough sun in Glasgow, right? :) | 12:18.38 |
tor8 | kens: Looks like Finnair or Airberlin are the main options; the London flights all involve a one night layover somewhere. | 12:18.42 |
| Finnair has better seat pitch. | 12:18.49 |
| Too bad Finnair is twice the price :( | 12:19.31 |
chrisl | Robin_Watts: I'm back.... big queue at post office :-( | 12:20.08 |
Robin_Watts | chrisl: No worries. | 12:20.14 |
| I have commits ready to go for mupdf and gs to change the name of the device. | 12:20.30 |
kens | Robin_Watts : more than we do in winter | 12:21.00 |
Robin_Watts | chrisl: If you want to nod at them, I'll commit them both at the same time, rebuild libgs.so for fred, and send an email out to mvrhel/fred warning them. | 12:21.58 |
chrisl | Robin_Watts: I'm not seeing the gs commit? | 12:23.10 |
Robin_Watts | Ahem. Try now, sorry. | 12:23.50 |
| Not the android one. | 12:23.58 |
chrisl | Yep, I see it - I'm good with both | 12:24.13 |
| I did mention to Fred via e-mail, that the name might change... | 12:24.36 |
| WTF? Why on earth should installing an Internet Explorer update cause Windows to have to restart? | 12:37.17 |
kens | everythign causes tht | 12:37.55 |
chrisl | Hmph :-( | 12:38.22 |
Robin_Watts | cos things like the windows update daemons depend on the system level http fetchers etc. | 12:38.29 |
chrisl | But not *every* update needs a reboot, and this is a freakin' web browser! | 12:40.35 |
| I'm guessing it's because the HTML rendering engine is used all over the joint | 12:41.04 |
sh4rm4^bnc | lol windows | 13:14.57 |
marcosw_ | sebras: Iâm going to reboot casper to install updates. | 22:38.13 |
rayjj | Robin_Watts: mvrhel: or anyone else: Comments solicited on a removing the 'compressed color list' junk (primarily from the pdf14 compositor). It's been moribund for quite a while and I'd like it out of the way. | 22:54.12 |
| ^^^ the commit is on http://git.ghostscript.com/?p=user/ray/ghostpdl.git;a=commitdiff;h=55203d93e86974ee974c441d567c27b06507e98e | 22:54.24 |
Robin_Watts | rayjj: It sounds like a good idea to me. I'll look over the commit tomorrow. | 22:54.42 |
rayjj | Robin_Watts: thanks | 22:54.52 |
| no rush. | 22:54.53 |
| I'll go forth on the real reason for getting rid of it (overprint simulation)... | 22:55.22 |
| Robin_Watts: you will see lots of lines in the diff starting with '-' | 22:56.20 |
| Forward 1 day (to 2015/08/29)>>> | |