This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Email Archive Search Mystery
- To: Jeffrey Oldham <oldham at codesourcery dot com>
- Subject: Re: Email Archive Search Mystery
- From: Hans-Peter Nilsson <hp at bitrange dot com>
- Date: Thu, 2 Aug 2001 19:23:08 -0400 (EDT)
- cc: <gcc at gcc dot gnu dot org>, <overseers at gcc dot gnu dot org>
On Thu, 2 Aug 2001, Jeffrey Oldham wrote:
> Kaveh Ghazi's change to libU77/configure.in is described in
> http://gcc.gnu.org/ml/gcc-patches/2001-07/msg00921.html. Since I knew
> he changed the code in Jul, I went to the gcc-patches archive for
> 2001Jul and searched in this time period. Searches (all words) for
>
> libU77 configure fails
> libU77 configure.in fails
> libu77 configure fails
> libu77 configure.in fails
>
> libu77/configure.in succeeds
> libu77/configure succeeds
>
> The email message's first line is
>
> The recent change to libU77/configure.in to always define
>
> so all these queries should match. What's happening?
All these boil down to two different phrases, as words are
case-insensitive and the "." is a delimiter that splits off the
"in" which is dropped because it's too short (and a common
word).
So the different phrases are "libu77 configure" and
"libu77/configure", and that difference is because the htdig
configuration says:
extra_word_characters: _/
which as you might imagine, adds those two characters to the
characters that make up a search-word.
The underscore is there so you can search for variable names.
I did some archeology but I couldn't find why slash was added to
be considered part of a search word. I've CC:ed overseers to
see if anyone remembers a valid reason. If none can be found,
I'll remove the slash for the gcc side. Thanks for reporting
this.
brgds, H-P