Re: Building same package twice and package forking
Stefan Fiedler <[email protected]> Mon, 13 Nov 2006 12:00:24 +0100
| Newsgroups | gmane.linux.distributions.rock.devel |
|---|---|
| Message-ID | <[email protected]> |
Am Montag, 13. November 2006 06:56 schrieb Luca: > Hi Stefan and good morning to you and everyone! ... > I don't actually understood one thing about forking: > let's take the example I did about the gcc forks (with some minor > adjustments): > > pkgfork gcc gcc-pass1 stages 01-------- priority 101.000 > pkgfork gcc gcc-pass2 stages 01-------- priority 101.000 > > If, for example, gcc not forked is being built only in native stages, > let's say: > " stages --2--5---9 priority 150.000 " > is it really necessary to add the "pkgremove gcc" below the declaration > of the forks? > > Again, thanks for help and feedback and have a nice day to everyone! > > Yours, > Luca Hi! Technically this is not necessary, but base packages like gcc are used differently than forked packages. How forks are built depends on each package. In the case of gcc, the 'fork' gcc=gcc (equal to gcc) cannot be built, mainly because it has no version and there is no source file for it in gcc.desc. When running scripts/Config, an enabled gcc simply adds its forks to the package list (config/<config>/packages) and removes itself. Another example of forks is the subversion package, which can build subversion=subversion and subversion=subversion-static forks. Here both forks share the same version and source files. Packages that parse the forked name are also possible, e.g. with gcc=gcc41-cross-<arch>, gcc.conf reads <arch> from the name and builds a cross-compiler for that architecture. Greetings, Stefan Fiedler