[DGD] Callout hack

[Felix A. Croes]

Shentino <shentino at gmail.com> wrote:

> As far as I know, DGD's statedumps only use a 32 bit word to encode
> all callout delays, so even if I could extend it, I'd need a way to
> pack the bits back again so that my superlong callouts could persist
> across a statedump.  I've done a trial run of eliminating the 60.0
> second top limit.  The delays appear to be "unwrapping" in intervals
> of 256 seconds, but as long as DGD is running, they still go off at
> roughly the same time as a second based callout for the same time.
>
> It just gets all screwy if I dump state.

It seems you still have a few bugs to squash.


> Wouldn't that wreak havoc with thread concurrency once MP comes around
> the block?  I'm leery about centralizing this way...

Not necessarily.  See:

    http://pattern.dworkin.nl/pipermail/dgd/2005-April/000064.html

Regards,
Dworkin