-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 The Monday 2006-05-01 at 22:01 +0200, Michael Schroeder wrote:
Because the bzip2 de/re-compression eats up lots of cpu cycles...
Ah! That explains it. I knew that it handles a lot of data, but it didn't somehow seem reasonable the time it used. Couldn't it use some other faster compress method? Less compression, perhaps? I didn't see an option in the man page for that.
No, it can't. The rpm payloads are compressed with bzip2, applydeltaiso has to use the same algorithm to be able to recreate an exact copy.
Ah, so that the md5sum matches... It could be uncompressed, the differences applied, then recompressed with something else - but then the checksum would be different. I suppose, I don't really know how it works.
Failing that, I could try recompiling the bzip binary optimizing for pentium 4 instead of the default :-?
I doubt that this will help, as memory latency is what's hurting bzip2, but you can try nevertheless.
Next time I'm bored O:-) - -- Cheers, Carlos Robinson -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.0 (GNU/Linux) Comment: Made with pgp4pine 1.76 iD8DBQFEVot/tTMYHG2NR9URAix1AJ9jp3DwA3nIarjen3BcKXYATlz9vACgjw9A GSvZcsEQEX7WH62Im/kOmUQ= =8w2V -----END PGP SIGNATURE-----