[Yum] Re: Recreating headers yields different files

[seth vidal]

On Mon, 2003-04-28 at 18:18, Axel Thimm wrote:
> On Mon, Apr 28, 2003 at 04:58:40PM -0400, seth vidal wrote:
> > > it uses the current time. If you ask google for
> > > http://www.google.com/search?q=gzip+modification+time
> > > you get as the first document
> > > http://www.ietf.org/rfc/rfc1952.txt
> > > which describes the gzip format:
> > 
> > Right, so I looked at the gzip module in python and the zlib module and
> > I came shockingly to the conclusion that controlling or disabling time
> > stamping on the gzip call is not possible in those libraries.
> 
> Maybe writing first the file, timestamping it and then gzipping it?

That's a whole lot more disk writes. and for people doing yum-arch over
nfs that's expensive.



> That's correct, only that you need to pull all of these 17MB each time
> anything changed (like adding a new package, which for my repo has a mean size
> of 1.3MB). Even so, it isn't that critical, I just thought I spotted something
> that can be optimized (thinking here more of web caching than rsyncing).


it can be and if it can be done nicely in the code I will, but I don't
think it can be done that cleanly at the moment.

-sv