On 11/08/2010 02:30 AM, Hrvoje Niksic wrote:
> Micah Cowan <[email protected]> writes:
> 
>> Yeah, that's always been the case. The question has always been: from
>> where did we get our md5 implementation? builtin, openssl, or ...? We
>> used to prefer openssl's and then fallback on a builtin one (which
>> probably came from libiberty, and shares parentage with gnulib's). In
>> some circumstances we also used another source (one provided on Sun OS
>> or OpenSolaris). This tag's purpose was to identify which one was
>> being used, so if anything went wrong, we'd know whose md5
>> implementation to blame :)
> 
> Has that ever happened for anyone?  It seems that even when MD5 "fails",
> it will be a failed compilation or at worst failed dynamic linking, not
> incorrect run-time operation.

Not to my knowledge.

The main point of these build-info tags was to provide as much
information as possible as to how wget was built, and to illuminate the
choices that were made by such build scripts as "configure"; I may have
been overstating it a tad by saying it'd be useful if md5 breaks, though
I do feel that just because they've never been broken before doesn't
mean they won't later. And I have personally witnessed dynamic link
problems produce errors in an MD5 sum routine (in mod_php, not wget), in
which case this information would've been useful.

But yeah, if there's no longer a choice being made, then illuminating
that non-choice isn't helpful.

-- 
Micah J. Cowan
http://micah.cowan.name/

Reply via email to