Re: [foo-devel] more info concerning hack

[martin r. <fo...@ru...>]

dear ramon,

On Thu, Jun 09, 2005 at 05:59:24PM +0200, Ramon Gonzalez-Arroyo wrote:
> Contrary to my first impression that the problem was bypassed with your=
=20
> hack, actually either it's another problem of memory leakage, or it's=20
> the same one but at really a different order of maginitude.

yes, there were some problems with memory leaks, all of them more or
less related to the new reference counting memory management style
(this is NeXTStep vs. OpenStep programming style).  there are some
design issues in conjunction with that which could be discussed for
the further development of foo, but for the moment i *hope* they are
fixed.

> What happens is that the computer starts to swap like crazy after some=20
> time, but until now my processes have managed to end. When it's done,=20
> and I stop foo, some memory is released, but, of course, something of=20
> the swap keeps. This means that even stopping foo after every=20
> process-run, I need to restart the computer every two/three runs. So it=
=20

that is a bit strange.  as soon as foo is stopped, *all* the memory
occupied by it should be released by the operating system.  that's why
it's strange that you have to reboot your machine.  of course, after
terminating a fat memory-leaked foo process, most of the other active
pages are swapped out to disk, which means, that the system will
behave a bit sluggish for some time until all the necessary pages are
back in memory -- but it shouldn't require a reboot...

anyway, i hope these problems are gone now...  please let me know
whether it works for you now or whether i introduced the next couple
of bugs with fixing...

bests,

martin