[Seaside-dev] Re: [Seaside Commits] Seaside 3.0: Seaside-Core-jf.682.mcz

[Philippe Marschall]

2011/1/5 Dale Henrichs <dhenrich at vmware.com>:
> On 01/04/2011 10:17 PM, Philippe Marschall wrote:
>>
>> 2011/1/5 Julian Fitzell<jfitzell at gmail.com>:
>>>
>>> On Tue, Jan 4, 2011 at 1:31 PM, Dale Henrichs<dhenrich at vmware.com>
>>>  wrote:
>>>>
>>>> On 01/04/2011 11:48 AM, Philippe Marschall wrote:
>>>>>
>>>>> 2011/1/4 Dale Henrichs<dhenrich at vmware.com>:
>>>>>>
>>>>>> GLASS doesn't reap sessions using the normal Seaside mechanism, so I
>>>>>> created
>>>>>> a WANoReapingStrategy class that did nothing and then modified the
>>>>>> code
>>>>>> in
>>>>>> WACache to use that as the reaping strategy.
>>>>>>
>>>>>> GLASS runs with multiple vms and those vms are all sharing the session
>>>>>> state, so it doesn't make sense to try to coordinate session reaping
>>>>>> across
>>>>>> multiple vms ... it is easier to run a separate vm that does the
>>>>>> session
>>>>>> reaping (once a minute by default) ...
>>>>>
>>>>> Yeah, we get that. The question at least the way I understood it
>>>>> couldn't you just use the plain WACacheReapingStrategy instead of
>>>>> WANoReapingStrategy/WANullReapingStrategy?
>>>>>
>>>>> Cheers
>>>>> Philippe
>>>>
>>>> Philippe,
>>>>
>>>> If you are asking _would_ I use WACacheReapingStrategy, the answer is
>>>> probably no:)
>>>>
>>>>  - I prefer to be explicit that the reaping strategy is a noop, not a
>>>>    side effect of the implementation ... GLASS users shouldn't be
>>>>    tempted to change it in their applications
>>>>  - I want to isolate myself from any future changes to
>>>>    WACacheReapingStrategy that may effect it's noop-ness
>>>>  - The class WANoReapingStrategy happens to have made it into the wild
>>>>    amongst the current GLASS users so I'm disinclined to remove it
>>>
>>> No, I agree with Dale: having an explicit Null Pattern reaping
>>> strategy is much more explicit. The implementation of #reap seemed
>>> unnecessary since it is not the job of the ReapingStrategy to
>>> determine how to reap, but rather *when* to reap (the method is just a
>>> private helper method that delegates to the cache).
>>>
>>> My *real* question was: why do we need
>>> WANullReapingStrategyConfiguration?
>>
>> One of these days I'll have to learn how to read. Yeah, good question.
>>
>> Cheers
>> Philippe
>> _______________________________________________
>> seaside-dev mailing list
>> seaside-dev at lists.squeakfoundation.org
>> http://lists.squeakfoundation.org/mailman/listinfo/seaside-dev
>
> That one we _can_ do without ... At the time I added the
> WANoReapingStrategy, I did not spend a lot of time trying to figure out what
> was and was not necessary and just followed the pattern that other
> strategies used which was to add a configuration, too:)

Done.

Cheers
Philippe