On 05/20/2012 10:17 AM, James Turner wrote:
> This is interesting - as far as I know, the current GC does not
> include a maximum delay and restart facility. If it did, that would
> entirely satisfy the current issues.  At least by my understanding.
>
> Equally, I've looked at the current GC code and didn't notice any
> code to support this feature. Does anyone else have any further
> information about this? Since it would be far simpler and more
> robust than any other solution thus suggested.

I was lucky enough to notice this come by.  I wouldn't hold your
breath. :)

This was an experiment, and honestly I have no idea why I put it in
the docs.

The idea was to do the GC normally, check timestamps periodically, and
then longtmp() out of it past some threshold, leaving the intermediate
sweep stuff in place.  But that's not enough, because now you need to
track all mutated reference-storing objects in a separate list so they
can be swept again.  And you need to have some kind of heuristic for
when it's OK to restart the sweep.

I have a vague memory of being sure I'd cleverly solved this, but I
never got it working and at this point, frankly, I suspect I was
wrong.

In my advancing age, I've come to believe that low-latency GC is just
a pipe dream.  You can have a realtime GC or you can have a production
system, but you can't have both at the same time.  Every managed
runtime in the modern world has latency bugs in some application or
another, every one of them.

Andy


------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Flightgear-devel mailing list
Flightgear-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/flightgear-devel

Reply via email to