thanks Jorge.

My manage is pretty keen for us to give Velocity a go at the moment
but if it doesn't make the cut, i'll definitely take a look at
NHRedis :-)

Definitely interested to see how you go implementing the new
concurrency strategy too.

cheers
Fatal.

On Nov 28, 12:07 am, Aaron Boxer <[email protected]> wrote:
> Fatal,
>
> Glad to hear you sorted that out.
>
> If you are looking into distributed caches for NH,
> you should check out my provider for Redis:  
> https://github.com/boxerab/NHRedis
>
> Redis is a blazing fast, distributed cache that supports lists,
> hashes, and transactional
> semantics.
>
> NHRedis does support distributed hard locks. Also, I am planning to
> write a copy on write
> concurrency strategy for NH, based on NHRedis, which will remove the
> locks on gets,
> and dramatically simplify the design.
>
> Cheers,
> Jorge
>
>
>
>
>
>
>
> On Fri, Nov 26, 2010 at 10:24 PM, Fatal <[email protected]> wrote:
> > hi Jorge,
>
> > Yes, I am using the read-write concurrency strategy. It turns out that
> > the code from that stackoverflow link didnt work too well wrt locking/
> > unlocking. It basically wasnt working properly and the exceptions were
> > being swallowed without any form of logging.
>
> > As it turns out, NH is only calling lock once on the ICache
> > implementation for each unlock.
>
> > I did some reading on the velocity today and have updated the
> > NH.Velocity cache to work a little better; patch contributed as
> > NHCH-31 (http://216.121.112.228/browse/NHCH-31).
>
> > cheers
> > Fatal
>
> > On Nov 27, 3:39 am, Aaron Boxer <[email protected]> wrote:
> >> Hi Fatal,
>
> >> I take it you are using read-write concurrency strategy.
>
> >> I've delved into the caching code a little, and from what I can
> >> gather, multiple Lock
> >> calls simply increment the lock count on the same soft lock object. The 
> >> lock
> >> handle should not be changing.
>
> >> I'm curious to know how you work this one out, as I am developing a new
> >> cache strategy class for read committed and repeatable read isolation.
>
> >> Cheers,
> >> Jorge
>
> >> On Fri, Nov 26, 2010 at 2:37 AM, Fatal <[email protected]> wrote:
> >> > hi all,
>
> >> > I've begun using the NHibernate.Caches.Velocity implementation today.
> >> > I'm updating the code to be compatible with the newer version (aka
> >> > AppFabric Caching Services). To do so, I used the VelocityClient code
> >> > from this stackoverflow question as a 
> >> > basis:http://stackoverflow.com/questions/3233792
>
> >> > I seem to be having some trouble with errors coming out of the locking
> >> > implementation and I just wanted to clarify the behaviour of
> >> > nhibernate caching as it relates to locking.
>
> >> > What i'm seeing is that NHibernate is calling Lock() several times for
> >> > the same key before it calls Unlock(). With the implementation of the
> >> > code from the link above, this causes exceptions because its trying to
> >> > add a different lock handle with same key to a dictionary (i.e.
> >> > because it is already locked)
>
> >> > None of the the NHContrib caches apart from this one seem to implement
> >> > locking yet so i'm curious as to whether its a bug with NHibernate or
> >> > if its intended behaviour and it is expected that locking
> >> > implementations ignore the case where the item is already locked.
>
> >> > Is anybody able to provide any clarity around this locking
> >> > functionality?
>
> >> > thanks,
> >> > Fatal
>
> >> > --
> >> > You received this message because you are subscribed to the Google 
> >> > Groups "NHibernate Contrib - Development Group" group.
> >> > To post to this group, send email to [email protected].
> >> > To unsubscribe from this group, send email to 
> >> > [email protected].
> >> > For more options, visit this group 
> >> > athttp://groups.google.com/group/nhcdevs?hl=en.
>
> > --
> > You received this message because you are subscribed to the Google Groups 
> > "NHibernate Contrib - Development Group" group.
> > To post to this group, send email to [email protected].
> > To unsubscribe from this group, send email to 
> > [email protected].
> > For more options, visit this group 
> > athttp://groups.google.com/group/nhcdevs?hl=en.

-- 
You received this message because you are subscribed to the Google Groups 
"NHibernate Contrib - Development Group" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to 
[email protected].
For more options, visit this group at 
http://groups.google.com/group/nhcdevs?hl=en.

Reply via email to