[05:07:17] [connected at Mon May 21 05:07:17 2007] [05:07:17] <> *** Looking up your hostname... [05:07:17] <> *** Checking ident [05:07:17] <> *** Found your hostname [05:07:17] <> *** No identd (auth) response [05:07:17] <> *** Your host is anthony.freenode.net[anthony.freenode.net/6667], running version hyperion-1.0.2b [05:07:19] [I have joined #peak] [06:07:31] ** GertS69 has left IRC (Remote closed the connection) [07:07:50] ** apoirier has joined us [09:02:46] ** zooko has joined us [12:08:02] Good morning, people of #peak! [12:08:12] (Mountain Daylight Time) [12:08:19] UTC-6 [12:09:37] ** pje has joined us [12:10:18] Good morning, pje! (UTC-6) [12:10:25] 'lo [12:27:47] ** apoirier has left IRC (Remote closed the connection) [13:25:26] As mentioned here: http://allmydata.org/trac/tahoe/browser/README [13:25:43] setuptools < v0.6c3 is incompatible with how I use setuptools in this package. [13:25:55] Is there a way to configure ez_setup.py to force an upgrade if the current version is too old? [13:27:18] use_setuptools(version_string_here) [13:27:36] more precisely, it will abort installation and ask them to upgrade. [13:35:52] Thanks. [13:37:39] Waitasec, that's the "DEFAULT_VERSION". I already set that to 0.6c3, and I thought what happened was that it left 0.6c1 -- or whatever comes with dapper -- in place and then the build errored out... [13:38:26] Hopefully my new changes to stop using --root and start using --single-version-externally-managed will make it work even with Dapper's version of setuptools anyway... [13:48:33] So since --root causes distutils to put "wrong" paths into .pyc's, the recipe that I posted to my blog to install either distutils or setuptools packages under GNU stow is no longer a good idea: [13:48:38] http://zooko.com/log-2007.html#d2007-04-27-distutils_or_setuptools_with_GNU_stow [13:48:57] Instead one should use one recipe for setuptools (using --single-version-externally-managed) and another for setuptools (using --prefix). [19:41:14] ** coderanger has left IRC (anthony.freenode.net irc.freenode.net) [19:43:42] ** coderanger has joined us [19:48:59] Is it normal for a package's ./setup.py test to error out until you've done ./setup.py build ? [19:49:15] Because I don't see how to make the test suite and the code loadable from setup.py without building it. [20:30:46] ** coderanger has left IRC (anthony.freenode.net irc.freenode.net) [20:35:49] ** coderanger has joined us [21:22:11] ** coderanger has left IRC (anthony.freenode.net irc.freenode.net) [21:23:48] ** coderanger has joined us [21:25:47] zooko, the "test" command runs build_ext; I don't see what other build subcommands would need running. [21:49:35] pje: thanks. [21:49:46] I'll see if I can make build_ext be sufficient... [21:49:53] The thing is that the src is not in ./$package name. [21:49:57] It is ./src/$package name. [21:50:10] So it is normally installed into ./instdir by the build process. [21:50:18] After which ./setup.py test loads it from ./instdir...