Re: GNAT for MS Visual Studio

[Simon Wright <simon.j.wright@mac.com>]

"Randy Brukardt" <randy@rrsoftware.com> writes:

> It would be easier to correctly use "protected" (task-safe)
> containers in concurrent uses than ones that only allowed a small
> amount of concurrency.  Not everyone is an expert in concurrency,
> after all.

Clearly one can protect a single container, either by wrapping it in
(for example) a protected object; or by providing an internal
mechanism.

But it's been my experience that applications that involve containers
at all usually involve more than one container, in the same way that a
database application usually involves more than one table. With either
implementation technology, you have to preserve integrity. And in the
Ada case, that means locking more than one container object at the
same time.

The BCs had a form where each container object was constrained by (an
access to) a sharable mutex object; so you could form a group of
container objects with a common mutual exclusion mechanism.

But the only way I could see to make this work was to grab the mutex
using an open protocol:

   grab the mutex;
   manipulate the containers;
   release the mutex;

and once you've done that, there's no point in having the containers
mutex-aware -- just use the plain sort, with external mutex. Which is
why the BCs no longer have support for concurrency.