Hi Samphan,

On Sun, Jan 22, 2006 at 10:27:53AM +0700, Samphan Raruenrom wrote:
> [...] 
> We submit many duplicates! :-)

Well - at least you try to avoid it :-)

> We always do a lot of search before submitting an issue.
> However, searching in IssueZilla to see whether the bug
> has been reported is not very effective. The term that was
> used to report the bug may be different from the one we
> use to search, though we always try many terms.

When searching for duplicates/whether the issue has already been
reported, it's usually best to query for
"Closed Duplicate" issues with only one or two search-terms in the
summary-fiels (if you use two or more, make sure to switch from "case
insensitive substring" to "all words/strings"!)

If that returns too much hits, then add some more terms to the
description (again make sure to switch to "all words" - you should thing
about modifying your default query so that it starts with all words
right from the beginning)

Then open one of the duplicates to find the issue all of those have been
marked duplicate of. This usually is the quickest and easiest way.

> The reporter may report different aspect of the problem,
> seeing the problem from different angle, may not put
> keyword in the subject or the word may have many
> synonyms.

See above. Finding the one that is left is harder than finding one of
the widespread duplicates.

Just make sure to start with a general term and only when that returns
too much hits start adding more terms to fine-tune the selection. Don't
start with more than four terms. (and never put all of those into the
summary)

> I suspect many have encounter the very same situation.
> There are 9600 duplicates per all 60000+ issues.
> This is 16% which is very high?

No, I don't think that this is very high.. Sure it could be less, but
OTOH, release-cycles are long and thus many people are focused with bugs
for a long period of time → many duplicates.

> Is it possible to enhance the search issue feature?
> Maybe having a choice to use Google to search
> just issues, or better search ranking system.

I always have my issue-List sorted by IssueNumber. This gives a fairly
good feeling whether an issue can fit the criteria or not.

I doubt that a google-search over the IZ-data will help you to identify
the issue more easily than a "clever" IZ-query.

Don't bother to include unconfirmed issues in a query to check whether
that issue has already been reported. (It would be great if you would
(and try to confirm the one you find afterwards, but one cannot expect
you to do so)
And you usually don't need to include new/started/reopened either. If
you cannot find a duplicate that already was closed as duplicate,
chances are low that you find the one in the opened ones. (again it
would be great if you could do a query for those as well when you cannot
find the issue otherwise, but again one cannot expect you to do so - at
least not currently, given the bad performance of IZ in the last weeks)

ciao
Christian
-- 
NP: Metallica - Fuel

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to