Re: rpm.RPMTAG_SOURCERPM vs rpm.RPMTAG_SOURCEPACKAGE

On Mon, 26 Jan 2009, Ville Skyttä wrote:

> On Friday 23 January 2009, Panu Matilainen wrote:
>
>> Yup, what was there before was correct (and identical to what rpm itself
>> uses) - the existence of RPMTAG_SOURCERPM tag means it's a binary,
>> otherwise it's a source.
>
> I just remembered something that was the cause of a long since resolved
> rpmlint bug: RPMTAG_SOURCERPM is also empty for installed gpg-pubkey
> packages, so they would fall into the "it's a source" category above.

Oh, yuck. Thanks for pointing that out, "fixed" now upstream by adding 
RPMTAG_SOURCERPM to gpg-pubkeys on import. The real fix is to get these 
non-packages out of rpmdb for good (rpm 4.6.x has some bits towards that 
but not quite there yet).

> There's a headerIsSource() in rpm's lib/header.h, is that exposed to Python
> somewhere?  I suppose yum is not the only package wanting to offload this
> issue to rpm ;)

Not at the moment, but trivial to add of course (I'll try to remember to 
do this sooner than later :)

 	- Panu -