Hello,
my code is not changed =). its safe in branch.
I was asking why the resetExpiration() method permanently disables the idle
timeout check interval, forcing the selector to run it for every select.
2009/1/27 Oleksiy Stashok <Oleksiy.Stashok_at_sun.com>
> Hi,
>
>
> I want to understand the purpose of the reset call:
>
> + ((SelectorThreadKeyHandler) selectorHandler.
> getSelectionKeyHandler()).resetExpiration();
> key.attach(response.getResponseAttachment());
>
>
> in trunk that resetExpiration() permanently disables the use of expiration
> check interval,
> so the check is performed on every select.
>
> in branch i changed it to reset the nextexpirationchecktime to 0,
> hence the next select will perform idle check, and after that
> its back to the interval delay per check.
>
> Actually it wasn't me, who made changes to your code. If you'll take a look
> at diff
> -
> selectorThread.getSelectorThreadKeyHandler().resetExpiration();
> + ((SelectorThreadKeyHandler)
> selectorHandler.getSelectionKeyHandler()).resetExpiration();
>
> you see, that I just changed the code, which tooks SelectorThreadKeyHandler
> from SelectorThread to take it from ProcessorTask instead.
>
> I have no idea, why your code was changed :(
>
> WBR,
> Alexey.
>
>
>
> btw. comet is using the main grizzly selector logic now,
> hence i had to enforce idlecheck once per second to keep the precision that
> comet allows for.
>
> --
> regards
> gustav trede
>
>
>
>
--
regards
gustav trede