Evolve 5.5.0 released
Danek Duvall
danek.duvall at oracle.com
Tue Nov 1 17:12:35 UTC 2016
Pierre-Yves David wrote:
>
>
> On 11/01/2016 05:46 PM, Augie Fackler wrote:
> >On Tue, Nov 1, 2016 at 12:03 PM, Pierre-Yves David
> ><pierre-yves.david at ens-lyon.org> wrote:
> >>>>How many other user are using the setup.py directly ? Could we have the
> >>>>setup.py abort unless run through pip ? (or any other package manager)
> >>>
> >>>
> >>>I'm using it as part of the Solaris build (as I am for all other Python
> >>>packages).
> >>
> >>
> >>Does Solaris has a package manager that is able to properly remove the old
> >>package file when installing the new one ?
> >
> >Yes. I think what Danek was pointing out is that the Solaris packaging
> >scripts need to use 'setup.py install' to stage things so the package
> >can be built. The deb rules work the same way, for what it's worth.
> >We'd need a workaround for that if we want to try and recommend pip
> >with some force.
>
> Yes, we'll need that (this was covered by the "(or any other package
> manager)" in the previous message.
Except that no package manager is involved when we build mercurial. It's
just a bunch of makefiles that drive the build, and cause mercurial to be
installed in its final form, but in a staging area; that staging area is
then what's packaged up. So there's no way to detect that it's being run
through our packaging system, since at the time setup.py is invoked, it's
not.
Danek
More information about the Evolve-testers
mailing list