lol... the reason I said ASAP is that I always forget to turn it  
on.... and having it done automatically for debug/devel runs would help.

As my applications stand right now.... I don't properly destroy _lots_  
of reference counted objects... I'd like that to change in the future  
and one way to get some visibility would be to get a nag message every  
time I run in debug/devel modes ;-)

I would rather it not be on by default for opt mode.  Opt should be  
opt... as in... it should have everything not necessary for  
calculation turned off.  People can still configure it to be on in opt  
mode... but by default I expect opt to be no nonsense.

Derek

On Dec 11, 2008, at 2:15 PM, Roy Stogner wrote:

> John Peterson wrote:
>> On Thu, Dec 11, 2008 at 2:38 PM, Roy Stogner <[email protected] 
>> > wrote:
>>> While explaining the reference counting base class to Paul the other
>>> day, it struck me: why is this configure option independent of
>>> compilation mode?  It's basically only used for debugging.   
>>> Instead of
>>> just enabled or disabled, would it make more sense to add a third  
>>> (and
>>> new default) option, that enables reference counting for dbg and  
>>> devel
>>> modes but disables it for opt/prof?
>>
>> Is this a performance issue?  Not every class is reference counted,
>> and it's not like a reference-counted pointer which requires a lot of
>> bookkeeping and junk to make sure all the counts are up-to-date.  I
>> don't object to turning it off during opt, just wondering if there  
>> was
>> a big slowdown for your code.
>
> No perceptible slowdown that I've noticed; it just struck me as a
> "debugging" sort of feature that we were using even when not  
> debugging.
>  I'd like to hear from Derek, though; I wasn't expecting an answer of
> "ASAP"!
> ---
> Roy
>
> ------------------------------------------------------------------------------
> SF.Net email is Sponsored by MIX09, March 18-20, 2009 in Las Vegas,  
> Nevada.
> The future of the web can't happen without you.  Join us at MIX09 to  
> help
> pave the way to the Next Web now. Learn more and register at
> http://ad.doubleclick.net/clk;208669438;13503038;i?http://2009.visitmix.com/
> _______________________________________________
> Libmesh-devel mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/libmesh-devel


------------------------------------------------------------------------------
SF.Net email is Sponsored by MIX09, March 18-20, 2009 in Las Vegas, Nevada.
The future of the web can't happen without you.  Join us at MIX09 to help
pave the way to the Next Web now. Learn more and register at
http://ad.doubleclick.net/clk;208669438;13503038;i?http://2009.visitmix.com/
_______________________________________________
Libmesh-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-devel

Reply via email to