Email List: Xaustin-group-lX
[All Lists]

RE: New cancelation points -- "compatibility" issue?

To: yyyyyyyyyyyyyy@xxxxxxxxxxxxx
Subject: RE: New cancelation points -- "compatibility" issue?
From: "Alexander Terekhov" <yyyyyyyy@xxxxxxxxxx>
Date: Mon, 22 Sep 2003 21:18:00 +0200
Jason Zions wrote:
[...]
> I'm getting tired of this ++-ify thing, Alexander. Let me tell 
> you how this world works, because I don't think you understand 
> yet.

I appreciate your wiliness to explain to me how this world works,
but you really need not to repeat it. I have really good memory.
Fire a sophocles search on... uhmm "lambda" (in the message body, 
and on this list, followups to offtopic list aside for a moment).

> So. Please let me know when you build, or find, a full-up C++ 
> language binding implementation for POSIX. 

It's far from "full-up" (covers only a few bits of threading... 
and it somewhat sucks implementation and interface wise), but 
you might want to take a look at: 

http://www.boost.org/libs/thread/doc

They also have some "filesystem" bits and so forth. The only
problem is that boost folks don't participate here and folks 
from here don't seem to care a bit about ++.

> Until then, though, I'm not sure how strongly C++ issues can 
> drive the C-language binding for POSIX.

The code (and it makes no difference whether it's C or C++) 
"above" cancelation point need to ensure that no leaks will 
occur and also take care of invariants in the event of cancel 
delivery. You put that code in place for concelation points, 
and don't care about cancel delivery for calls that are NOT
cancelation points. Changing interface to allow cancel for
any previously "no cancel" call is a major "compatibility"
problem, as I see it.

regards,
alexander.

<Prev in Thread] Current Thread [Next in Thread>