Problems with r40628

Will Coleda will at coleda.com
Mon Aug 24 03:09:32 UTC 2009


On Sun, Aug 23, 2009 at 10:45 PM, Austin
Hastings<Austin_Hastings at yahoo.com> wrote:
> Howdy,
>
> I spent today paying the svn update tax for parrot, going from 40290 to
> latest, then using svn-bisect to figure out what was breaking my stuff.
>
> Presently, my Close compiler doesn't work. That's my fault. But it doesn't
> work in a particular way that gives me hope for light at the end of the
> tunnel.
>
> When I updated, though, it started to not work in a different way. It would
> consume all the available memory (700m or so) and then I would get a
> "Killed" written out.

See also: http://code.google.com/p/partcl/source/detail?r=619 ; where
another 9 partcl spec tests are now failing due to memory issues. (I'm
seeing explicit memory related PANICs from parrot.)

I haven't yet done a bisect. The tests were running to completion at
parrot 40633/tcl r601, and are now failing at parrot r40744/tcl r618

> That struck me as suspicious - it seems more like a kernel type Killed than
> a shell type "Killed due to resource limit blah blah".
>
> At any rate, svn-bisect tells me that the problem comes from r40628, which
> naturally has a whole bunch of files associated with it. (It's never the
> one-line change that gets me... :( )
>
> With no evidence whatsoever, I suspect that hte problem might lay with
> resizable pmc arrays, because I create a *bunch* of them in my debugging
> code -- pretty much M per function call, where 1 <= M <= 5.
>
> Like the yellow-bellied coward that I am, I'm going to update to 627, and
> hope that some of the work I've seen going on today will fix this. If not,
> I'll put some more time into it when I get my Close refactor done.
>
> =Austin
>
> _______________________________________________
> http://lists.parrot.org/mailman/listinfo/parrot-dev
>



-- 
Will "Coke" Coleda


More information about the parrot-dev mailing list