On 1/12/11 8:21 PM, Paul Wratt wrote:
The short version... Find a spec file. The spec is the build instructions for rpm. spec goes in /usr/src/REDHAT/SPECS. Run rpm -ba <filename>.spec to begin the build process. If the process completes successfully a binary rpm will be generated under /usr/src/RPMS/m68kmint (or noarch for a package that is all scripts or docs, etc). The rules for the spec are fairly loose but one rule is to make sure you have a line that says "Vendor: Sparemint"On Thu, Jan 13, 2011 at 11:22 AM, Bernd Mueller<ragnar76@googlemail.com> wrote:Am Donnerstag, den 13.01.2011, 09:55 +1100 schrieb Paul Wratt:On Thu, Jan 13, 2011 at 2:38 AM, Helmut Karlowski <helmut.karlowski@ish.de> wrote:Paul: What do all these (?) mean? I think a tutorial should not ask but answer questions. Also there is a strange character: '''XCONTROL'''´s Would be nice if you could fix these. -Helmut(?) need to be verified, or is missing details (like the authors name for GEMView) the weird character is supposed to be ' but it cam straight after 3 of them (for bold) Paulhi, i've just converted the rpm faq into the wiki format ( http://wiki.sparemint.org/index.php/RPM_FAQ ). Can you move that in the "about sparemint" box? bye BerndThere is an "how to contribute rpm pages to sparemint" doc somewhere on the sparemint server too. maybe Mark can also verify its contents with current procedure, as I remember there being issues/changes with new version or rpm I would be good to have that doc as the 1st Item in the new RPM_FAQ (it was relatively short if I remember rightly, sorry I can not provide a url atm) Paul
The ideal way to understand the build process is to download an existing srpm from sparemint.org. Install it on your system with rpm -Uvh <filename>.src.rpm. The resulting files will all be under /usr/src/REDHAT folders.
On the http://dev.sparemint.org website which is not yet in public use, you would then upload the generated src.rpm file. The fact that you have this file to upload is "proof" that the source builds properly. Once it is uploaded, the file is submitted to several aranym build farm instances who then extract and rebuild the package optimized for 060, coldfire, etc. That's the theory anyway, and mostly it is working.
Hope this helps someone get started. It's really not that complicated if you understand the overall structure of the unix hiearchy and what each package does (generally). On our platform it is annoying because we have bugs that kill the build. To get around this we have been short circuiting the build. That is we begin the build process, pause the process, and change the filesystem with a build that is already done and then allow rpm to finish. You only need to resort to this on large packages like gcc, etc. Usually anything else you can just reboot and the build will succeed.
Thanks, Mark