grostoon wrote:

> Hi Pavel,
>
> Ok, I understand that garbage collection has work to do but : - why
> does it takes so much time for just 50Mo database file ? - why does
> FB2.0 give the result in a few seconds ???

FB 2.0 has new, reworked index structure that handles GC in indices with
a lot of duplicates more efficient. It has also improved GC strategy
and GC thread scheduling (but that apply only to Super Server, Classic
still uses cooperative GC only, so if you have tested Classic, it all
goes to new index structure). See 2.0 release notes for details.

> So garbage collection is the sole problem, may I have much better
> times if I set the sweep interval to 0 ?

No, sweep has nothing to do with this.

> BTW, I'm using isql with direct access to the database file (i.e I'm
> not using the fb_inet_server but direct connect
> '/home/toon/fbtest.fdb'). Does this make a difference ?

No.

best regards
Pavel Cisar
IBPhoenix