Optimize ticket.query.execute

[anonymous]

The ticket.query.execute function could be optimised somewhat…

     if self.max == 0:
            self.has_more_pages = False
            self.offset = 0
        else:
            self.has_more_pages = True
            self.offset = self.max * (self.page - 1)

source:trunk/trac/ticket/query.py?rev=13234#L110​

            self.num_items = self._count(sql, args)

            if self.num_items <= self.max:
                self.has_more_pages = False

            if self.has_more_pages:

source:trunk/trac/ticket/query.py?rev=13234#L295​

if self.max = 0 this: self.num_items = self._count(sql, args) is useless since if self.has_more_pages: will always be False.

So I suggest adding something like:

if self.has_more_pages: // This line
     self.num_items = self._count(sql, args)

            if self.num_items <= self.max:
                self.has_more_pages = False

            if self.has_more_pages:

That would save one sql-query which in worst case scenario could double(?) the execution time I suppose

[Christian Boos]

Well, we compute .num_items not just to set the .has_more_pages flag, but also because we need that information. The question is if in this particular case (which I admit we don't use or test much, because it's probably a bad idea to not paginate the results), we could retrieve the .num_items from the main query. It looks like we could easily do that.

[anonymous]

Then it might have been my bad, since it just returns the result i didn't think of that it could be used later…

However it might be better/more optimized to calculate num_items/has_more_pages on demand.

I noticed this when I used the xmlrpcplugin with max=0, so the count query was a bit .. unexpected…

[Jun Omae]

I've had the same issue while developing ​th:ExcelDownloadPlugin. The plugin provides download in excel format from query page and converts always all tickets from query (max parameter is ignored) to excel format.

Currently, the plugin overrides _count instance method of Query instance to prevent SELECT COUNT(*) query. See ​source:exceldownloadplugin/0.12/tracexceldownload/ticket.py@13866:180-188#L163.

[anonymous]

I do understand that Query can't work perfect for everyone, but its never nice when you have to hack around it like that =/

I propose adding a separate method has_more_pages instead, so it can be calculated on demand instead

[Ryan J Ollos]

Retargeting tickets in milestone:1.1.3 that have no owner. Please retarget as appropriate.

[figaro]

Added keyword 'performance'. Upped priority to normal.

[Ryan J Ollos]

Narrowing focus for milestone:1.2. Please move ticket to milestone:1.2 if you intend to fix it.

[Ryan J Ollos]

Milestone renamed