Re: [C1][General][Patches] Implement monitor cleaning

[Robin Green <gr...@hotmail.com>]

"Tagunov Anthony" <at...@nnt.ru> wrote:

>This mechanism should solve a problem I can see no non-radical
>way to solve: our monitors (in processors) generally never get >cleaned.
>
>One solution is to clean the monitor in the beginning of the >process() 
>method,
>but what if the same processor (xslt, xinclude) get invoked twice (3 
> >times..)
>per request? For this solution we're likely to loose information >about the 
>request's
>dependencies.
>
>This is what I propose: let us implement some new interface (I called it
>Invalidatable) and make the processors implement it (an alternative is
>to add one method to Changeble interface).

A third alternative is to completely redefine the caching system so that you 
don't have the kind of Monitors that need clearing at all! That's what I've 
done - it's not in CVS yet, it needs cleaning up and voting on. I hope to 
post a link soon but I've also got 1.8.1 to think about (this is far too big 
a change to go into 1.8.1 so it'll have to go in later).



_________________________________________________________________________
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.

Re: [C1][General][Patches] Implement monitor cleaning

[Tagunov Anthony <at...@nnt.ru>]

Hello, Robin!

Robin Green wrote:

>"Tagunov Anthony" <at...@nnt.ru> wrote:
>
>>This mechanism should solve a problem I can see no non-radical
>>way to solve: our monitors (in processors) generally never get >cleaned.
>>
>>One solution is to clean the monitor in the beginning of the >process() 
>>method,
>>but what if the same processor (xslt, xinclude) get invoked twice (3 
>> >times..)
>>per request? For this solution we're likely to loose information >about the 
>>request's
>>dependencies.
>>
>>This is what I propose: let us implement some new interface (I called it
>>Invalidatable) and make the processors implement it (an alternative is
>>to add one method to Changeble interface).
>
>A third alternative is to completely redefine the caching system so that you 
>don't have the kind of Monitors that need clearing at all! That's what I've 
>done - it's not in CVS yet, it needs cleaning up and voting on. 

Very pleased to hear that! I've been thinking about something like that too, 
but did not find time for any large job! Glad you've done it!

Is your solution storing dependency information somewhere in the Page object?
It would be great! The dependecies would vanish right when the cached object
would vanish!

>I hope to 
>post a link soon but I've also got 1.8.1 to think about (this is far too big 
>a change to go into 1.8.1 so it'll have to go in later).
>