[ILUG] Re: [linux-edu] war games?

[Kenn Humborg]

[previous discussion of using nested ifs for resource
allocation/deallocation]

> Because in any of the coding standards I've used and based on the reading
> I've done, 3 layers is enough embedding to go to.   Anything else requires
> something more awkward.   I'd be tempted to go with the extra
> level but I'd
> run the risk of the code not getting a clean review and having to
> rewrite it
> anyhow.   I don't like having to break the code down into anything smaller
> because you have to do extra checking.   In this case, because I'm a 'doze
> programmer, I'd rsort to using events and use the
> waitForMultipleObjects API
> to handle that.  I have to say I _REALLY_ would hate having to go
> that far.

I think you're missing the point.  We're talking about acquiring
resources in general, not waiting for things to happen.  Think
malloc() rather than WaitForSingleObject.

Very often a function has to allocate (or create, or open, or
connect to) a resource and be sure to deallocate (or delete, or
close, or disconnect from) this resource before returning.

For example:

   open input file
   open output file
   malloc I/O buffer
   do stuff
   free buffer
   close output file
   close input file

You are talking about waiting for multiple events to happen.
In general, there is no analogue of the "resource deallocation"
step there.

Later,
Kenn