[ZODB-Dev] ZODB 3.9.0 released

[Jim Fulton]

On Wed, Sep 9, 2009 at 5:42 AM, Chris Withers<chris at simplistix.co.uk> wrote:
> Jim Fulton wrote:
...
>>> - when using `with` blocks and a ConflictError is raised, what happens?
>>
>> The transaction is aborted.
>
> That might be surprising for old-time Zope users... Still, I assume the
> ConflictError propagates through so they'll see it raised, right?

Yes.

>>> - if using cache-size-bytes, is the parameter that limits the number of
>>> objects by number ignored?
>>
>> No.
>
> Is there any way to get this to happen?

There might of if someone had spoken up in the many months since this
was added. Now, it is a wildly low priority.

> Would seeing the number limit extremely high have the right effect?

For some definition of desired.

>>> - where can I find out more about invalidation-age?
>
> http://svn.zope.org/ZODB/trunk/src/CHANGES.txt?rev=103676&r1=103674&r2=103676
>
> There's a "that" missing from the first of the new lines ;-)
>
>> src/ZEO/tests/invalidation-age.txt
>
> Ah cool, now I get it :-)
>
> typo: verificatioin -> verification

Feel free to fix typos yourself.

> I take it this is irrelevant if you're not using persistent client caches?

No, even "non-persistent" caches persist for the length of the client
process and have to be checked on reconnection.

> Are those safe to use now? [1]

They are much safer than they used to be.  We're using them.  We've
had a couple of problems that I need to investigate.

Jim

-- 
Jim Fulton