Miles Bader <[EMAIL PROTECTED]> writes: > The strong point of abrowse in my experience is that it was much, > much, faster than rbrowse in some cases (with remote repositories), > because it could avoid fetching every revision.
It was mostly more fluid, because it fetched the log for a full version before displaying anything. After benchmarking and "strace"-ing a little, I understand the problem you're refering to: rbrowse will fetch all the revisions (not all the log files, but the .listing file for the version) and filter them afterwards. This results in an overhead for tla 1.0 archives (I didn't check for baz archives. We may be downloading the same .listing file several times, but this will be solved when using the flattened namespace). For regexp search, we can't do otherwise. For string search, it would probably result in an awfull hack, with rather low benefit since the problem will be solved as people migrate to baz archive format. I'm now convinced we have to let abrowse in 1.5. It's currently obsoleted in the sense "not in the output of baz help, and displays a warning stating that it will be removed". Newcomers will learn to use browse, which is the way of the future, while long time users will almost not see the difference. -- Matthieu _______________________________________________ Gnu-arch-users mailing list [email protected] http://lists.gnu.org/mailman/listinfo/gnu-arch-users GNU arch home page: http://savannah.gnu.org/projects/gnu-arch/
