Re: First stage of Quesa changes now complete

[Roger H. <rog...@mi...>]

On Thursday, January 27, 2005, at 04:10  pm, Joseph J. Strout wrote:

>
> That is surprising.  This may be a silly question, but are you sure 
> Sampler is showing you the time spent only in the parent function, and 
> not the time spent in the parent function or any of its children 
> (functions it calls)?

Yes, you have to check the box 'Invert call tree' , then the left hand 
column displays just the time in a each function.
We call Q3Object_Is_Type from nearly every Q3 routine and some of the 
E3 routines too.

>
>> I think in the release version we are checking out inputs too often, 
>> for instance in E3Shared_Acquire we check the parameter is of type 
>> shared. We then call Q3Shared_GetReference which again checks its 
>> parameter is of type shared.
>> It also calls E3System_Bottleneck even though we must have come into
>> E3Shared_Acquire from a Q3 routine which will have already called
>> E3System_Bottleneck.
>
> That's a good point.  I would think that once we're in a E3 routine, 
> we should be safe to directly call another E3 routine rather than 
> going through the Q3 version.  As a general principle, we may trust 
> ourselves, even if we don't trust the user.

Good. I'm not going to do a wholesale change, but when I find a time 
critical area I will apply that principal.
As for not trusting the user, in a debugging environment I agree 
wholeheartedly, in production if we detect
an error then most of the time an application will merely fail in 
another way. I use Quesa through C++ wrappers
so that I would actually have to go out of my way to call Quesa with an 
object parameter which was not
actually an object. Maybe I could compile a special version of Quesa 
where Q3Require etc are defined to be
nothing, just like Q3Debug is in the release version.

Roger.