| 05:01.11 | *** join/#nslu2-linux eiffel (~eiffel@dslb-084-061-036-252.pools.arcor-ip.net) |
| 08:33.23 | *** join/#nslu2-linux jomaster (~jomaster@f053084025.adsl.alicedsl.de) |
| 09:13.57 | *** join/#nslu2-linux kolla (kolla@haugum.uninett.no) |
| 11:37.56 | jomaster | eno: at some packages file '.build' will removed on cleaning and on other packages not. is this necessary? than i add it always i note this missing line. |
| 11:38.16 | jomaster | also some packages have no .PHONY |
| 11:47.28 | jomaster | i add my unstage-proposal to zlib.mk / template.mk / template-cvs.mk. i hope you look at this and inform me about your mind. |
| 12:02.51 | *** join/#nslu2-linux nslu2-log (~nslu2-log@limax.nslu2-linux.org) |
| 13:25.38 | *** join/#nslu2-linux CIA-10 (cia@208.69.182.149) |
| 14:40.04 | *** join/#nslu2-linux GPSFan (~kenm@64.92.145.112) |
| 16:47.31 | eno | jomaster: not all packages support "make uninstall" |
| 16:48.30 | eno | probably half of them don't |
| 16:49.29 | jomaster | that's much more than i thought |
| 16:50.20 | jomaster | but we have to test all package before commiting. |
| 16:50.29 | eno | i don't think we're at a phase to add -unstage to template.mk or template-cvs.mk yet |
| 16:50.55 | jomaster | ok. feel free to revert it |
| 16:51.15 | eno | it would just confuse new developers |
| 16:51.39 | jomaster | you are right |
| 16:52.00 | eno | maybe you can experiment on the packages that absolutely require -unstage |
| 16:52.09 | eno | such as mysql and mysql5 |
| 16:53.32 | eno | i rarely use the -clean target |
| 16:53.48 | eno | it's not used in official build either |
| 16:53.53 | jomaster | rssh.mk add "--with-cvs=/opt/bin/cvs" but don't depends on cvs or suggest it |
| 16:54.28 | jomaster | but if it will used removing '.build' is important? |
| 16:55.09 | eno | official build just uses scripts/optware-autoclean.pl and invokes -ipk |
| 16:55.25 | eno | i'd say if we remove -clean, nobody will notice |
| 16:55.50 | eno | so -clean is not important at all |
| 16:56.18 | jomaster | ok |
| 16:57.05 | jomaster | template-cvs.mk use "cvs" but should be better "$(CVS)", isn't it? |
| 16:57.08 | eno | you're right about rssh, please correct it |
| 16:57.33 | eno | probably suggest it |
| 16:58.23 | eno | it was implied that cvs in PATH |
| 16:58.37 | eno | in that case cvs or CVS makes no difference |
| 16:59.07 | eno | only in the context of using host tool, we need $(CVS) with full path |
| 16:59.20 | jomaster | but than defining $(CVS) at makefile doesn't make sense |
| 16:59.52 | jomaster | ok, only want to ask about a hidden reason. |
| 16:59.59 | eno | what do you mean? |
| 17:00.07 | CIA-10 | 03jomaster * r11232 10optware/trunk/make/rssh.mk: rssh: add missing cvs suggesting |
| 17:00.46 | eno | oh i c |
| 17:01.01 | eno | there is a "CVS=cvs" line in Makefile |
| 17:01.02 | jomaster | nothing important. my english seems to bad for you :-D |
| 17:01.27 | eno | that predates my involvement with optware |
| 17:02.49 | jomaster | i will add $(CVS) to template-cvs.mk with my first host-tool-commit. |
| 17:03.57 | eno | so you're CVS as the first host tool example? |
| 17:05.01 | jomaster | yes, it is pretty easy. ;-) |
| 17:05.36 | eno | i'd hate to have to build svn as host-tool though |
| 17:05.47 | jomaster | futhermore, checking standard-tools and abort with an error-msg if one doesn't exist |
| 17:05.58 | jomaster | this would be my second tool :-D |
| 17:06.26 | eno | I think I have a different opinion on this |
| 17:07.10 | eno | cvs or subversion, these are fundamental development tools that all host build machine supports |
| 17:07.27 | eno | they're not giving us any trouble, why fix it? |
| 17:07.49 | eno | by using our own build, we might introduce problems |
| 17:08.12 | eno | how about working on something that DOES give us trouble? |
| 17:08.26 | eno | like autoconf? |
| 17:09.33 | eno | we need (maybe incorrectly) a specific version |
| 17:10.32 | eno | do you know how many dependencies subversion has (recursively)? |
| 17:11.12 | eno | I doubt we will ever do a better job than debian developers given the current resource |
| 17:11.33 | eno | in theory, everything can be built from scratch |
| 17:11.58 | eno | we have to work on things that matter |
| 17:13.18 | jomaster | you are right. but i think it isn't finished, if the build works without an error on the build-server |
| 17:14.55 | jomaster | on native build we should test for all tools at least. |
| 17:15.35 | eno | no, even for native build, we have cvs/svn included in optware-devel |
| 17:15.52 | eno | so we don't need to rebuild it |
| 17:16.15 | eno | neither cvs or svn is giving us ANY trouble at all |
| 17:16.34 | eno | so these are not priority |
| 17:17.03 | eno | i don't give a *** about theoretical perfection |
| 17:17.25 | jomaster | :-D |
| 17:18.04 | jomaster | i leave it out and go directly to aclocal... |
| 17:18.14 | jomaster | no cvs-commit |
| 17:18.22 | eno | thank you for understanding |
| 17:19.22 | jomaster | what about $(CVS) |
| 17:19.55 | eno | search all the .mk files, if it's not used, feel free to remove it |
| 17:20.14 | eno | but again, no rush (since nothing would be broken by it) |
| 17:21.28 | *** join/#nslu2-linux toi (~toi@d54C2A96D.access.telenet.be) |
| 17:26.42 | CIA-10 | 03bzhou * r11233 10optware/trunk/make/ (template-cvs.mk template.mk): reverted template{,-cvs}.mk -unstage change, not to confuse new developers |
| 17:28.37 | jomaster | $(CVS) is used. |
| 20:40.21 | CIA-10 | 03jomaster * r11234 10optware/trunk/Makefile: Makefile: add host-tool-stage for aclocal-1.9 / automake-1.9 if necessary |
| 20:41.36 | CIA-10 | 03jomaster * r11235 10optware/trunk/make/ (autoconf.mk automake19.mk libtool.mk m4.mk): automake19 / autoconf / m4 / libtool: add host-tool-stage for aclocal-1.9 / automake-1.9 if necessary |
| 20:42.41 | CIA-10 | 03jomaster * r11236 10optware/trunk/make/ipkg-opt.mk: ipkg-opt: use aclocal-1.9, automake-1.9 host-tool if necessary |
| 20:43.17 | jomaster | eno: what do you think about this? example is ipkg-opt. |
| 20:44.06 | jomaster | do we append host-staging-bin to PATH instead of calling $(ACLOCAL19)? |
| 21:17.17 | *** join/#nslu2-linux Reedy (~Reedy@wikimedia/pdpc.active.reedy) |
| 21:21.57 | *** join/#nslu2-linux ByronT (~byrontodd@nslu2-linux/ByronT) |
| 21:42.42 | *** join/#nslu2-linux ka6sox (~ka6sox@nslu2-linux/ka6sox) |
| 21:47.59 | eno | jomaster: your change does not work on machine with aclocal-1.9/automake-1.9 |
| 21:48.36 | eno | if we can avoid conditionals, we should |
| 21:48.54 | eno | especially $(SHELL), these are expensive |
| 21:51.03 | eno | your change could send optware build into infinite loop |
| 22:02.02 | ka6sox | shudders |
| 22:04.22 | CIA-10 | 03bzhou * r11237 10optware/trunk/ (Makefile make/ipkg-opt.mk): correction on HOST_TOOL change |
| 22:10.18 | ka6sox | its working! w00t |
| 22:10.35 | jomaster | i am so stupid. forget to test with automake-1.9 again |
| 22:11.15 | eno | ka6sox: thank you for fixing CIA and svn |
| 22:11.50 | eno | jomaster: it's alright, I'm testing the host_tool case on another build machine |
| 22:12.13 | eno | the change overall looks good |
| 22:12.44 | jomaster | little typos on automake |
| 22:13.54 | CIA-10 | 03jomaster * r11238 10optware/trunk/Makefile: makefile: correction on HOST_TOOL_AUTOMAKE19_STAGE |
| 22:15.08 | jomaster | what about appending $PATH versus $(AUTOMAKE19)? |
| 22:15.45 | jomaster | it's late here. changing packages with aclocal-1.9 tomorrow. |
| 22:16.24 | eno | yeah, let's sleep on it |
| 22:18.18 | ka6sox | eno, that would be rwhitby who did the fix. |
| 22:19.01 | eno | great |
| 22:19.52 | eno | is wondering what's so special about aclocal-1.9 and automake-1.9 that we have to use this 1.9 particular version |
| 22:21.43 | jomaster | that was not my idea, really ;-) |
| 22:22.58 | jomaster | i read on automake-page that the versions aren't compatible in all cases. so sometimes a special version is required. |
| 22:23.21 | jomaster | but we can test the pages, working with current automake. |
| 22:25.24 | eno | i bet 90% of the places we don't need to use specific version |
| 22:25.40 | eno | just tested with ipkg-opt, and it builds fine |
| 22:26.21 | jomaster | but not always. i remember that i test a build for some months which don't work with current automake. |
| 22:26.51 | jomaster | (of course not ipkg-opt) |
| 22:28.27 | jomaster | to using $(shell is the idea of an experience-less non computer-scientist. seeing a better way makes me happy. |
| 22:29.22 | jomaster | (reference to me) |
| 22:30.42 | jomaster | about 50 packages using aclocal-1.9 |
| 22:31.07 | jomaster | good night. su |
| 22:33.26 | eno | night jomaster |
| 22:40.36 | CIA-10 | 03bzhou * r11239 10optware/trunk/make/ipkg-opt.mk: ipkg-opt: use autoreconf, avoid specific version of aclocal/automake |