[Yum-devel] yum on an olpc machine (slooooooooooow)

[seth vidal]

On Sat, 2006-12-16 at 11:25 -0500, Jay Soffian wrote:
> On Dec 16, 2006, at 10:41 AM, seth vidal wrote:
> 
> > That's a time that cripples usability.
> 
> Sanity check.
> 
> We're talking about
> 
> - extracting the headers from RPMs
> - converting those to XML
> - converting the XML back to binary (in the form of sqlite db)
> - transferring the binary db to the end system
> - so that yum can:
> - use the binary db
> - to download binary headers (since the extracting information is not  
> sufficient on its own)
> - to use RPM to solve a transaction



> 
> Um, wouldn't it be a lot more sane to generate an rpmdb in the first  
> place (/usr/lib/rpmdb/...), transfer that to the end-system, and use  
> rpm's built-in dep solver?
> 
> I'm just saying...

1. a full rpmdb is massive in comparison to the amount of data we put
into the xml metadata.

2. the built in depsolver doesn't do multiple repositories to any detail

3. the bdb's aren't portable across arch's or endianess.

-sv