| <<<Back 1 day (to 2011/08/11) | 2011/08/12 |
robin_watts_mac | tor8: char is unsigned on all ARM platforms. | 00:35.50 |
| I put it to you that char being signed is the "wierd" case :) | 00:36.04 |
tor8 | oh darn it | 00:36.26 |
| better go over all structs where I use plain chars then. hopefully not so many. | 00:36.46 |
| robin_watts_mac: shouldn't you be out in the jungle? | 00:37.07 |
robin_watts_mac | just about to go to breakfast. | 00:52.19 |
robin_watts_mac | heads out of the day. bbl. | 01:44.55 |
kens | Hi robin_watts_mac | 08:47.08 |
tkamppeter | Can someone have a look at https://bugs.launchpad.net/bugs/825054 | 08:54.06 |
| GS 9.04 seems not to build on PPC. | 08:54.27 |
kens | I don't know if we have a PPC machine, maybe chrisl has an old Mac still ? | 08:54.44 |
| But the error is a re-definition which suggests a platform problem. | 08:55.10 |
| Could be a configure problem or similar, looks like we are including a definition of a system structure which is already defined. | 08:55.44 |
tkamppeter | kens, what do you mean with "platform"? PPC or Ubuntu? | 08:55.49 |
kens | Sorry, I mean the compiler/library in this case. | 08:56.12 |
| Do we have the version of gcc and glibc being used ? | 08:56.40 |
| We are including the library definition in 'time.h' and also our own definition in 'time_.h' | 08:57.35 |
| I'm afraid this kind of build problem is beyond me, you really need chrisl, who has disappeared for now. | 08:58.35 |
| chrisl did you see Till's message about a build problem with PPC on Ubuntu ? | 09:43.49 |
chrisl | No, I missed that - but I've been having connection problems this morning...... | 09:46.09 |
kens | chrisl Yes, my connectin died. I take it you've seen Till's comment in the logs though ? | 10:19.19 |
chrisl | kens: yes, I'm looking at it now. | 10:21.28 |
kens | OK thanks, its beyond me.... | 10:21.38 |
chrisl | tkamppeter: 9.04 built without a problem on my PPC box running Ubuntu 10.10. My guess is that time.h has been moved on the latest PowerPC build, and thus the test in unix-aux.mak fails ( if ( test -f $(INCLUDE)/sys/time.h ); then $(ECHOGS_XE) -a $(gconfig__h) -x 23 define HAVE_SYS_TIME_H; else true; fi ) - FWIW, I've no idea what it is done that way :-( | 10:22.48 |
| kens: did you want the patched 9.04 installer done, as well? | 10:29.50 |
kens | yes please, but there's no rush, teh customer is in the US. | 10:30.09 |
chrisl | 32 bit only okay? For now at least? | 10:30.26 |
kens | 32-bit Windows is what they say they are using. | 10:30.38 |
| I'm sure that's all they need | 10:30.53 |
chrisl | Okay, I'll kick it off in a few minutes - laptop wants to update <sigh> | 10:31.03 |
kens | My PC did that when I started it up yesterday.... | 10:31.15 |
chrisl | kens: mail on way - just a sanity check of the patch | 10:54.51 |
kens | THanks chrisl | 10:55.03 |
| chrisl it looks fine, as long as it compiles OK it should be good. | 11:01.46 |
chrisl | kens: yep, it builds fine. http://www.ghostscript.com/~chrisl/gs904w32-bug692422.zip has the installer in it, it's password protected, the password is their customer number. | 11:03.10 |
kens | Right I'll pick it up and send it to them later, thanks! | 11:03.30 |
robin_watts_mac | I believe, it's policy that we don't do builds for customers. | 11:08.04 |
| Not that we aren't allowed to do it, just that they aren't allowed to expect it :) | 11:08.26 |
kens | robin_watts_mac : but we can't test the fix ;-) | 11:08.31 |
robin_watts_mac | sure. I've done builds and passed them along before. | 11:08.58 |
chrisl | kens: it might be an idea, when you pass it along, to explain this is a special case (because we can't test it) and we're doing it as a courtesy. | 11:10.21 |
robin_watts_mac | exactly. | 11:10.28 |
kens | Yes, will do so. | 11:10.29 |
| Woudl like to hear from Marcos though | 11:10.41 |
chrisl | going to lunch...... | 11:13.43 |
tkamppeter | chrisl, thanks, have found out where they are. Is there a ./configure or make options to override this check or to direct it to something like /usr/include/x86_64-linux-gnu/? | 11:26.47 |
| Someone knows how I tell GS that I have a system-supplied time.h and overriding the check in unix-aux.mak? | 11:42.31 |
chrisl | tkamppeter: sorry, I had to step out for a bit. | 11:57.43 |
| Do you see the same build problem on x86_64 then? | 11:57.59 |
| tkamppeter: you might try "make XCFLAGS=-DHAVE_SYS_TIME_H=1" to override the automatic setting | 12:01.18 |
tkamppeter | chrisl, thank you, I will try it. | 12:33.58 |
| chrisl, seems to work. Thanks. | 12:58.18 |
chrisl | tkamppeter: cool. I think I can make configure set the value properly - I'll open a bug to remind me. | 12:59.04 |
tkamppeter | chrisl, great. | 13:09.29 |
chrisl | tkamppeter: http://bugs.ghostscript.com/show_bug.cgi?id=692426 | 13:10.21 |
tkamppeter | chrisl, thanks. | 13:11.14 |
| chrisl, everything compiled now. Perfect. Thanks. | 13:32.13 |
chrisl | tkamppeter: great, thanks for letting me know | 13:32.27 |
henrys2 | kens:we have a potential customer that wants a minimal pdfwrite build, FEATURE_DEVS and DEVICE_DEVS would only include what is needed, I think it best we do it and test it, I am concerned something is going to be "missing" at run time. just FYI. | 13:46.26 |
kens | OK henrys | 13:46.39 |
robin_watts_mac | henrys2: Did you sort out a con-call with Company R ? | 14:08.00 |
chrisl | tkamppeter: do you know what the status of the contrib/pcl3 device is? Is it widely used? | 14:08.12 |
robin_watts_mac | wonders if that was in fact where the request for a pdfwrite only gs came from. | 14:09.17 |
henrys2 | kens:is there some chance you could have a patch for that today tested? | 14:12.44 |
kens | Err, patch for what henrys2 ? | 14:12.59 |
henrys2 | the stripped down pdfwrite | 14:13.10 |
kens | OPh right, I'll try, I'm fighting with Dan Grillo and their installer and the handles for someone elese | 14:13.34 |
henrys2 | thanks. | 14:13.47 |
| robin_watts_mac:yes that is what I'm talking to kens about. | 14:14.29 |
kens | Is this on Windows ? | 14:14.39 |
henrys2 | oddly they want pdfwrite not what I was hoping ... | 14:14.45 |
| yes windows | 14:14.52 |
kens | OK thanks | 14:14.57 |
henrys2 | I don't understand 692128 - hintak seems to think we will auto rotate I thought that was just pdfwrite - or is it all vector devices? | 14:20.30 |
kens | OK finally got installer sent to Dan Grillo, I'll try the cut down build now. | 14:23.43 |
| henrys2 I assume the customer will build GS tehmselves, and just need a modified make file ? | 14:24.14 |
| brb | 14:24.18 |
henrys2 | right or a patch against 9.04 - probably complete files they are net people ;-) | 14:25.40 |
kens | henrys2 not sure what kind of patch you mean ? | 14:25.59 |
| Or do you mena make an executable ? | 14:26.16 |
henrys2 | just send me the modified makefiles. | 14:26.38 |
kens | OK, I'll go start playing then. | 14:26.52 |
henrys2 | patch usually means something like "git diff" | 14:27.10 |
kens | Yes, can do that OK. | 14:27.21 |
henrys2 | but I don't know if they want that. | 14:27.23 |
kens | :-) | 14:27.27 |
| I expect they will wnat executable/DLL already built | 14:27.39 |
henrys2 | no they do need to use the api. | 14:28.01 |
kens | Really ? OK | 14:28.18 |
henrys2 | I guess I could ask them again I sort of pushed them towards the api because they wanted to make all these device changes and I assume there will be more changes later. | 14:30.19 |
kens | Not a problem. Almost all FEATURE_DEVS will be requied though. | 14:30.38 |
henrys2 | they do not need pdf input I don't know if that can be separated out. | 14:31.23 |
kens | Really ? OK then I can drop that too. What about EPSF ? | 14:31.37 |
henrys2 | he just told me ps -> pdf so I think that should stay. | 14:32.14 |
kens | OK. | 14:32.21 |
| Build started, lets see how badlyt wrong it goes ;-) | 14:34.56 |
henrys2 | oh it isn't going to be tested with the clusterpush you are going to have to look at it locally. | 14:36.09 |
| duh, didn't think of that. | 14:36.22 |
kens | I figured that. | 14:36.24 |
| Windows only after all | 14:36.34 |
henrys2 | I really want a windows cluster machine. | 14:36.53 |
kens | It ought to be possible | 14:37.03 |
| Hmm.... | 14:38.41 |
| Our standard test harness assumes it has hte display device, and uses some library calls. | 14:39.00 |
henrys2 | I can leave X11 out of the unix build and it works fine. | 14:41.07 |
kens | The Windows test harnes uses copy_dib copy_palette copy_draw etc. | 14:41.41 |
| Or somethign anyway | 14:41.49 |
| I'm getting unresolved externals on those | 14:42.03 |
henrys2 | what test harness? | 14:42.14 |
kens | Just the usual Windows one, gswin32 | 14:42.29 |
| I think. | 14:42.32 |
| In fact its gsdll32.def | 14:43.01 |
henrys2 | well the console app should be okay right? | 14:43.08 |
kens | Its hte creation of the DLL. | 14:43.22 |
henrys2 | maybe best to just include the display stuff | 14:43.53 |
kens | Maybe, but it will make the app bigger | 14:44.14 |
henrys2 | do you have a simple stub to build with the api? | 14:44.54 |
kens | Well, now it throws a wobbly on ICC profile stuff. | 14:44.56 |
| No, no simple stuff | 14:45.04 |
| Anyone know where the ICC profile stuff gets built ? I'd have thought a feature dev, but doesn't seem to be. | 14:46.05 |
| Lets do this a different way. | 14:46.29 |
henrys2 | for now I'd just make it as small as possible without C code changes just makefile and we'll get something to them and see if it is sufficient. | 14:46.57 |
kens | That's my intention. | 14:47.07 |
| Goign to do it by removing stuff one at a time. | 14:47.18 |
henrys2 | tkamppeter:if you could weigh in on 692128 that would be great. | 14:48.47 |
kens | Drat, I just threw away a pending change :-( | 14:53.08 |
henrys2 | kens:chrisl makes the customer binaries now shouldn't we just make a 9.04a or something for Dan. | 15:07.58 |
kens | henrys2 chrisl made a special installer for him, 9.04 plus hte patch. | 15:08.17 |
henrys2 | oh so that should be fine I just saw the question to marcos. | 15:09.01 |
kens | Well, they asked if it was OK to ship it, I can't see why not, but thought I should check. | 15:09.23 |
| Turns out we need practicallly all the FEATURE_DEVS | 15:10.01 |
henrys2 | okay well we should get some space relief getting rid of output devices. | 15:10.44 |
kens | will find out shortly.... | 15:11.03 |
| Hmm, well its built, lets see if anything works..... | 15:15.44 |
| Nope. doesn't work | 15:16.47 |
| henrys2 I'm not going to get this finished today. | 16:42.13 |
| I keep running into problems. | 16:42.19 |
| Actually, maybe I can finish after dinner | 17:04.33 |
| Off now, have a good weekend everyone | 18:49.34 |
tkamppeter | henrys, bug 692128 fixed. Patch approved by real iron. | 20:27.25 |
henrys | thanks I saw that, you can commit it if you like or I'll do it later. | 20:28.22 |
| tkamppeter | 20:28.30 |
tkamppeter | henrys, already committed. | 20:30.14 |
| Forward 1 day (to 2011/08/13)>>> | |