154004 - QuickSearch should be able to find a bug by its alias

Status: RESOLVED FIXED

(Bugzilla :: Query/Bug List, defect)

Description

[Andreas Franke (gone)]

On b.m.o, the bug lookup text box in the footer supports lookup by alias.
QuickSearch doesn't do this. It should, in some form.

Here is some discussion from the webtools newsgroup:

=======================
[jesse:]
Btw, we should remove the "alias" field before it becomes 
impossible to change the bug id boxes to search boxes.  
The alias field is a workaround for searches being slow,  
and searches will soon become faster.
=======================
[afranke:]
There is an alternative solution: Enable quicksearch to find
bugs by their aliases (or substrings of their aliases). 
Currently this can only be implemented as a special
prefix, e.g. alias:bmo-regressions ,which would defeat
most of the usability win. But when the perl port to
quicksearch is checked, then it could first check whether one 
of the non-prefixed word-fragments entered in the quicksearch 
box occurs as a substring in some alias for a bug, and display 
the matches at the top of the result page, before the "normal" 
query results. And if you enter only a single string and there 
exists a bug with exactly that alias, then the "normal" query 
does not need to be executed anymore because the bug is probably 
the one the user is looking for. In this case, there can be a 
link on the result page that executes the normal quicksearch
query, without looking for aliases again.
=====================
[gerv:]
A simpler alternative to your alternative - restrict aliases to the set 
[A-Za-z-_] (if they aren't already) and then do an alias lookup first if 
the search is a single word.
=====================
[afranke:]
Yes, but you still need the perl port of quicksearch 
for this, because from JS you can't find out whether
the returned result from show_bug.cgi?id=foo is a 
bug view or a not-found error message, and fall back
to the normal quicksearch in the latter case. Or can
you do this? How? Btw, the perl-port checkin thing is
in bug 70907.

(I still think substring search in the aliases is better
than just exact lookup because e.g. you will want to
hit bmo-regressions if you type in bmo, or regressions,
or even regression.)

Comment 1

[Jouni Heikniemi]

Should this be blocked by bug 70907?

Comment 2

[Andreas Franke (gone)]

Yes, I think so. Marking dependency.