[BRLTTY] Solaris memory leak, maybe windows too

[Willie Walker]

Just to follow up: it's been running on my machine for well over 24hrs 
and I've started/stopped Orca many times in that period.  It seems very 
stable.  I'm looking forward to a new tarball release that incorporates 
this patch.

Many thanks!

Will

Willie Walker wrote:
> Thanks Dave and Samuel!
> 
> It seems to function well.  I ran BrlTTY and monitored the pmap -x 
> output while I used Orca.  The stack remained at a steady 12K and the 
> total size remained at a steady 5.5M.
> 
> I've had it running for the past hour and have started/stopped Orca many 
> times during that period.  It seems quite stable.
> 
> Many thanks!  This was a tricky one to find, and I admire Samuel's 
> ability to track it down and Dave's ability to flip over a patch so 
> quickly.
> 
> Will
> 
> Dave Mielke wrote:
>> [quoted lines by Samuel Thibault on 2009/08/20 at 00:57 +0200]
>>
>>> So for variable-length arrays, the standard doesn't say that its
>>> lifetime extends only to the end of the block, thus allowing compilers
>>> to be lazy and just allocate them on the stack and not free them.  The
>>> only portable way to free them is thus to return from the function
>>> containing the block.  For brltty, that means making the content of the
>>> while(1) loop a function called by runProgram, else its variable-length
>>> arrays keep getting allocated and never freed.
>>
>> Thanks.
>>
>> Will: Please try the attached patch (stack-1.patch). If it works, I'll 
>> clean it up a bit and commit it.
>>
>>
>>
>> ------------------------------------------------------------------------
>>
>> _______________________________________________
>> This message was sent via the BRLTTY mailing list.
>> To post a message, send an e-mail to: BRLTTY at mielke.cc
>> For general information, go to: http://mielke.cc/mailman/listinfo/brltty
> 
>