[torqueusers] TORQUE 2.4.10 released

Nico Kadel-Garcia nkadel at gmail.com
Tue Aug 31 10:22:51 MDT 2010

On Tue, Aug 31, 2010 at 11:56 AM, Garrick Staples

> "spec file manager"? That's me.
> My goal is create rpms that match the upstream default build as close as
> possible in the easiest way possible.
> If someone is editing the spec file, they get to keep the pieces.
>> .spec files usually assemble their "configure" options, for example
>> with the '%configure' macro, which works well, and by adding or
>> disabling components such as drmaa with a flag controlling the use of
>> the ''--enable-drmaa" configure option.
> Unfortunately, we can't use %configure because you can't pass arbitrary flags
> to it on some distros because it is an enourmous multi-line string of commands.
> Keeping it simple also meets the goal of matching the upstream default build.

Then don't publish it, or at least don't put it in the tarball. You're
overriding RPM macros without announcing your shift, and ignoring
macro resetting in the .spec file.

> What ever happened to the original complaint that drmma didn't work with my
> 'make rpm'? That's all I'm interested in. I don't care what EPEL does. I don't
> care about people editing the spec file. I care about my spec file working.

Which it does only in messy fashion: in particular, it doesn't work
with "rpmbuild -ta torque-2.4.10.tar.gz". You have the drmaa
featuresets disabled in your ./configure options, so the drmaa
packages get disabled in the "./configure" options, but they get built
with no files in them because the '%if' statements don't include the
package itself, merely the files for the packages.

And, oh yes, on the test sytem I'm on right now, because you didn't
manually disable the gui in your configuration arguments with
"--disable-gui", the "build_gui no" setting in the .spec file is
passively ignored and the xpbs components get built anyway, causing
havoc. So no, on systems with all the GUI components available, the
local ./configure operations outsmart your published arguments.

More information about the torqueusers mailing list