You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by "William A. Rowe, Jr." <wr...@rowe-clan.net> on 2003/02/20 17:26:49 UTC
Re: Patch : small optimization in request.c -
ap_directory_walk()
At 10:19 AM 2/20/2003, Jean-Jacques Clar wrote:
>Just bypassing a call to apr_filepath_merge by using the canonical_filename.
-1 (veto, not vote.)
You have no way of knowing that 3p modules stacked one-on-top-of-another
are respecting canonical_filename. Especially beasts like rewrite, et al.
module 1: maps /someuri to /foo/bar/somefile, sets filename and canon_file
module 2: rewrites filename /foo/bar/somefile to /boo/dar/otherfile, doesn't
unset or update the canonical_filename.
dir_walk passes.
canonical_filename was an early design. In the future 2.2 generation, it will
be the product of dir_walk'ers for whatever backend store the server is using.
But wait, that's the filename. Effectively it was a good idea that really proved
to be unviable as long as all have direct access to r->filename.
Good thought. Sorry.
Bill