On 16 Jul 2001 22:06:51 +0200, Christoph Egger wrote:
>
> On Mon, 16 Jul 2001, Christoph Egger wrote:
>
> >
> > On 16 Jul 2001, Thayne Harbaugh wrote:
> >
> > > Let's see if we have agreement.
> > >
> > > Directory structure:
> > >
> > > <libname>/
> > > <libname>/dist/
> > > <libname>/dist/rpm/
> > > <libname>/dist/debian/
> > > <libname>/dist/win32/
> > > <libname>/dist/bsd/
> > > <libname>/dist/solaris/
> > >
> > >
> > > Where <libname> expands to libgii, libggi, libgalloc, etc.
> >
> > Yep, agree. That is, what I am trying to say the whole time.
> >
> > > Do we create the entire directory skeleton even for empty dirs or do we
> > > create each subdir as needed when packagine files are created?
> >
> > No empty dirs, please.
>
> Usually, I don't reply myself, but I forgot to write my reasons for this
> decision here:
>
> I don't know, if this possible, but supposed you can install rpm's on
> solaris, then the solaris directory is not needed. I don't know, if
> all *BSD system have one common package-format or if each *BSD system has
> its own package-format. In the last case we need a directory for each
> supported *BSD system.
I used to install RPM's on Solaris all the time. I'd much rather see
the names of the subdirs refer to the packaging system and not the
operating system: rpm, deb, pkg, etc.
> Therefore, creating only NON-empty directories prevents you and all other
> packagers to deal with directory cruft in CVS.
>
>
> > > RPM spec files are reference files that may or may not be used by a
> > > particular distribution, but will work for a general system.
> >
> > If you define a LSB-conform system a "general system" then I agree.
> > (LSB = Linux Standard Base).
> >
> > > Any thing else?
> >
> > No. I see, you got me finally. :)
>
>
> CU,
>
> Christoph Egger
> E-Mail: [EMAIL PROTECTED]
--
Thayne Harbaugh
