Lachlan Andrew wrote:
On 30/10/2007, John Heffner <jheffner@xxxxxxx> wrote:
Lachlan Andrew wrote:
we were thinking of just taking byte-wise differences of the entire data
structure.
Would anyone else find it useful to improve the compressibility of web100 logs?
John, if we implement this, would you support it in future releases of
web100_userland?
Sounds like a decent idea. How are you going to compress the deltas?
I have no problem applying a patch if it works and you find it useful. :)
Thanks for your prompty reply, John.
So far we're just compressing them off-line using bzip, partly so
that the CPU load of compressing didn't interfere with experiments and
partly for simplicity. Do you think the patch should have an option
to pipe it directly through bzip?
Using libbz2 would probably work. It would need some tweaking of the
autoconf scripts, too.
-John
_______________________________________________
Discussion mailing list
Discussion@xxxxxxxxxx
http://internal.web100.org/mailman/listinfo/discussion
|