Fwd: Re: [Caml-list] Bigarray question

[David Thomas <david_hd@yahoo.com>]

Actually, the language being experimental is perhaps a
better reason to stick to the documented
functionality. Something *might* change, and you're a
lot more likely to hear about it if they  break
documented functionality than if they happen to change
something that's supposed to be interal that just
happens to break your code.
 
--- Thomas Fischbacher
<Thomas.Fischbacher@Physik.Uni-Muenchen.DE> wrote:
> 
> On Mon, 28 Nov 2005, Dmitry Bely wrote:
> 
> > 1. For the current implementation it's safe (use
> > Data_bigarray_val(v) to change the data pointer).
I don't see 
> > any reason why it could become unsafe.
> > 2. It is up to you whether to use this feature or
not.
> 
> I will neither use undocumented programming
techniques, nor 
> recommend doing so to other programmers, no matter
what kind
> of language. That would - quite in general - be a
very bad habit
> that eventually must lead to bugs that are hard to
impossible to
> diagnose and debug.
> 
> Assigning to Data_bigarray_val is not explicitly
permitted.
> 
> Nevertheless, being allowed to do so would be
helpful. So, a 
> simple change to the documentation would suffice to
make me 
> happy.



		
__________________________________ 
Yahoo! Music Unlimited 
Access over 1 million songs. Try it free. 
http://music.yahoo.com/unlimited/