On Tue, Feb 7, 2012 at 3:32 PM, Oliver Burger <oliver.bgr@googlemail.com> wrote:
Am 07.02.2012 14:52, schrieb Reinhard Tartler:
On Tue, Feb 7, 2012 at 2:09 PM, Oliver Burger<obgr_seneca@mageia.org> wrote:
I wonder why do you require these changes? Don't you use something like fakeroot for package building? that should the INSTALL_OWNER changes unnecessary.
Well, I know, that fakeroot is used in deb packaging but not in rpm packaging, and since you don't build as root, it's quite impossible to use those -g and -o options.
Just curious, is fakeroot not available *at all* on fedora/mageia systems.
In addition, they are not needed at all, the package building process does fix the permission of these files on its own, at least rpmbuild does.
That maybe right, but if we go this route, your patch is also unnecessary, because you can also easily directly override the INSTALL_DIR and INSTALL_FILE macros during the makefile invocation/
The SYSCONFDIR/ETCDIR might make sense if you want to install somewhere else than '$(DESTDIR)/etc/x2go'. However, I'm pretty sure that installing those files somewhere else will break x2go in more or less subtle ways as I think those paths are hardcoded at more other places. Therefore, I wonder again what this patch fixes for you.
You are partly right there. While rediffing the patch from 3.0.99.9 I did overlook, the $(DESTDIR) was already added. The patch read
- in the previous version. So you are right, this one isn't really needed anymore. But I have to wonder, why a variable $(ETCDIR) is defined instead of hardcoding it, when it will break x2goserver, if the user should define it during build? Your argument there seems to fire backwards.
I never claimed the Makefile to be perfect.
If we wanted the to make the location /etc/x2go/... really configurable (which I think would be great) then this is mechanism is a great start. Nevertheless, a lot of additional places needs to be fixed, and I see a lot of other way more important things that need fixing first. Therefore, I'd personally recommend to focus on things that really need fixing first.
Your proposed patches in this thread seem rather cosmetic to me, because the functionality that you require can be achieved without any patching as well. I'm not saying they are bad or wrong, I'd rather prefer the makefile to be fixed properly instead of making it more complicated.
-- regards, Reinhard