Jeff Johnson wrote: >> Though I did think that only the function vectors >> (fmtFunction, tagFunction) were in the actual API ? >> > > This is a still feeble avstraction for the XML/TAML/JSON spewage. > > The API I was referring to is with header tag extensions/formats. > That API is too limiting atm, which is why you added an integer arg.
Ah, OK. Yeah, I added new formats and new tags instead. (such as the :isodate format or the json/database tags) >>> The --queryformat code -- while quite general -- was never >>> intended for (but is handling) the spewage load that --qf is being >>> repurposed for. Can't be helped, can clean up the trash any time >>> its clearer what spewage is _REALLY_ needed, and a commitment to UUID >>> and RPMTAG_HDRID and RPMTAG_PKGID is a huge step in the right direction. >> >> I'm not sure if there are any consumers of the current >> metadata, so I'll just go ahead and change it I think ? >> > > Go fer it. There is interest from UP in rpmrepo, and (iirc) cAos > was using for a while. Otherewise its just thee and me. Should be mostly fixed now, did some tricks where --qf got in the way (such as the DBINSTANCE inside the loop) >> The yum sqlite looks a bit off in terms of table fields, >> and the wnh mongo shouldn't need to store dbinstances... >> > > There is "make check" in place for popt packages that should > detect the changes. Likely "off" because yum has changed, not > the sqlite chema. The python pickle for dirs has never been done > correctly, could be done with embedded python, otherwise RPM > is written in C, not python, and the yium sqlite schema is useless > for RPM and has an impossible to resolve "incompatibility" because > of the choice to put a python object directly into the sqlite database. Hmm, trying to find where this python pickle enters... Did you have an example, I _think_ it's all fixed now ? i.e the generated SQL does load with the Yum schemas, as created by the "yum-metadata-parser" libxml2 code: http://yum.baseurl.org/gitweb?p=yum-metadata-parser.git;a=blob;f=db.c If it was about the "filelists", that was more about dirnames vs. basenames than about Python vs. C afaik. There's still some "file vs. stream" checksum issues lurking in the repomd.xml creation, needing reproducer. --anders ______________________________________________________________________ RPM Package Manager http://rpm5.org Developer Communication List rpm-devel@rpm5.org