[Rpm-maint] Rpm Database musings

[Panu Matilainen]

On 03/04/2013 01:23 PM, Michael Schroeder wrote:
> On Mon, Mar 04, 2013 at 12:19:34PM +0100, Ales Kozumplik wrote:
>> On 03/04/2013 11:21 AM, Michael Schroeder wrote:
>>> Actually libsolv can do a "incremental" update if it has an old
>>> solv file available, i.e. it takes the unchanged content from the
>>> old solv file and only queries new headers from the rpm database.
>>> Ales doesn't yet use this in method in hawkey.
>>
>> Didn't know about it but it will become relevant once optimizing DNF load
>> time is relevant. Also, can libsolv do the same for the XML metadata?
>
> No, but there's no way to do "random access" for the xml data
> anyway, we always have to parse it completely.

The .sqlite db format (which yum prefers) does support random access. 
OTOH the .sqlite files tend to be somewhat bigger than the compressed 
.xml variants etc.

	- Panu -