Improving performance

[alessio.spadaro@…]

The plugin, during the post-process phase, calls check_ticket_access for each ticket in the hdf. This is extremely slow as the number of the tickets grows. Maybe a better solution could be:

• build a list of id from the hdf
• select all the ticket in a single call returning the relevant fields (id, reporter, owner and cc list)
• use this list to filter out the hdf, removing tickets that doesn't match

I hope i'll be able to prepare a patch implementing this.

[Noah Kantrowitz]

This can't be done in 0.10 without breaking the encapsulation of the ticket API. If you need a faster database, try Postgres.

[alessio.spadaro@…]

I'm actually using Postgres, and with as little as a thousand ticket the load on the machine for, say, the active tickets standard report is quite high (for each query we have at least 1+n queries, where n is the outcome of the first query).

I thought it was possible obtaining such list with a single call on the query object, but double-checking it seems that this is not possible :(

I'll open a RFE on trac (Btw, thanks for your work)

[Noah Kantrowitz]

This is already possible with trunk, as they recently re-added the ability to use ids in queries. However I have no plans to port this plugin to trunk in the near future.