allow clues to match "duplicate count > X"; e.g. to list top-crashers

[Alexander Sack]

If we extract duplicate count for a bug, we could allow clues to match bugs with duplicate count greater than X.

Use-Case: For firefox there are lots of crashes. In order to see really important ones, it would be beneficial to produce a list of "top-crashers" by matching bugs with more than 10 duplicates.

Any suggestsions on the 'op' element syntax/semantics for such a feature?

[Daniel Holbach (dholbach)]

This will be implemented with bug 79140. I'm not convinced we need that in the clue files. Once we transition to the 'bughelper server', it will be easy to add a
   bughelper --nd '>10' -p firefox -A
query.

I'm inclined to close the bug as duplicate.

[Alexander Sack (asac)]

On Wed, Apr 25, 2007 at 06:22:05AM -0000, Daniel Holbach wrote:
> This will be implemented with bug 79140. I'm not convinced we need that in the clue files. Once we transition to the 'bughelper server', it will be easy to add a
> bughelper --nd '>10' -p firefox -A
> query.
>
> I'm inclined to close the bug as duplicate.
>

What are the reasons that you don't want this rule not to be available
for clue files? IMO, all mechanisms should be available in clue
files, as we can easily integrate them in bughelper-data.

Further, we could combine it with different ops, like:

 * find all top-crashes of firefox
 * find bugs that are not MASTER, but have duplicates (-> triaging bug)

Later integer expressions would allow me to match for other measures,
like

 * time since last activity

... again in combination with certain state/tag 'op' combinations this
can have different meaning:
 a) are we waiting for a report? -> Close bug
 b) are we waiting for developer to process bug -> Show up in "Needs
 Processing" task page.

 - Alexander

[Daniel Holbach (dholbach)]

I don't feel strongly about it, it's more of a design decision.
 * Clue files contain raw clues, which can be used in different queries.
 * Task lists should be generated by using arguments in the command line.