Re: RPM packaging with SCons

On February 20, 2022 12:33:15 PM MST, Werner Reisberger <[email protected]> wrote:
>On 2022-02-20 16:51, Mats Wichmann wrote:
>
>> SCons includes two rpm things - the packager and the tool. The 
>> packager,
>> called through the Package builder, which attempts to construct a
>> specfile first, indeed seems a little flimsy, but the tool, which sets
>> up the Rpm builder, doesn't seem to do much more than your suggestion -
>> call rpmbuild given a .tar.gz.   For reasons not clear to me, there is
>> no documentation entry for the Rpm builder - nor are there any direct
>> tests of it. Odd (note: we should do something about that).
>> 
>> Anyway - have you looked at the Rpm builder?  Would it work in your
>> decoupled scenario?
>
>Yes, I looked at it but it seems to be rather inflexible. E.g. you 
>cannot easily change the way a RPM should be build. It expects that 
>rpmbuild gets a tar file with an included spec file. It then builds a 
>source and a binary RPM because it runs with the "-ta" option. I may be 
>able to alter this to "-tb" to build only a binary package but it looks 
>that it limits the processing to this two options because the expected 
>source is defined by the get_cmd(source, env) function. It also 
>specifies where the temporary path for rpmbuild is created.
>
>Basically I have no idea how I could call this tool and interface with 
>it out of a SConscript.
>
>--Werner
>_______________________________________________
>Scons-users mailing list
>[email protected]
>https://pairlist4.pair.net/mailman/listinfo/scons-users

Ok, fair enough. I may try to do some prospecting on that to find the intent. Any eventual tool will have to do some of what that does, like define RPMCOM & RPMSOMSTR for consistency, so good to know if it can be salvaged.
-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.