Optimizing searches on short_desc

[Myk Melez]

Joel Peshkin wrote:

> I noticed that most searches for strings in short_desc do not take 
> advantage of the fulltext index.

None do on b.m.o, but after the upgrade we'll be in a position to find 
out how many people take advantage of that search option.

> 3 questions:
>
> 1) myk - Do you have any data showing what fraction of all BMO 
> searches are "contains all words/strings" on the summrary?

Here's what the current access log says:

[root at mecha httpd]# grep 'GET \/buglist\.cgi' access_log | wc; grep 'GET 
\/buglist\.cgi' access_log | grep 'short_desc=[^&]' | wc; grep 'GET 
\/buglist\.cgi' access_log | grep 'short_desc=[^&]' | grep 
allwordssubstr | wc;
  40244  790696 26393924
   5649  113366 4929319
   5509  110491 4832808

(i.e. of 40K searches, only 5K or so are summary searches, but most of 
those are "contains all words/strings" searches)

-myk