[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: AttributeDescription caching (was: some problems after your commits)



Howard Chu wrote:
> 
> One more comment here - since there is no way to promote an existing
> read lock to a write lock, you have to release the read lock and try
> to reacquire the write lock. As such, the cross-check is still needed.

Correct. But if you use a binary tree (or any sorted storage) the check 
is a side effect of adding the entity to the tree, and the chances of
having the same AttributeDescription created simultaneously by two
threads
are very small, as you previously said. But this is pure speculation.

However, I agree that in common use there's little need of keeping 
the variations of an AttributeDescription sorted for fast access.

Pierangelo.


-- 
Dr. Pierangelo Masarati               | voice: +39 02 2399 8309
Dip. Ing. Aerospaziale                | fax:   +39 02 2399 8334
Politecnico di Milano                 | mailto:masarati@aero.polimi.it
via La Masa 34, 20156 Milano, Italy   |
http://www.aero.polimi.it/~masarati