Re: Ada & Scripting

[Martin Krischik <krischik@users.sourceforge.net>]

sk wrote:

> Hi,

>   > I have some C sources for 'popen' which show how this is
>   > supposed to work. Now I "only" have to map the function to Ada.

>   > I saw them. As I said 'popen' ony returns one of the
>   > pipes - so I have to program my own version of 'popen'.
> 
> ... and as I said, the method is 2 PIPES (not 'popen')

I understand that much
 
> The routine (algorithm) goes, roughly, like so ...
> 
> <algorithm>
> 
[...]
> 
> </algorithm>

Which is precicly what popen does - only popen then drops one of the pipes
and returns only the other one. Personaly I think that was very wastefull
of the original designer of popen.

> Use the "select" or "poll" system calls in the parent
> process to determine if the pipes are open, open but
> blocked or closed.
> 
> There is no difficulty in implementing this algorithm
> in Ada, but it is a very "C" way to do it. A more Ada
> solution is to use Ada tasks or protected objects, but
> there is still a 'fork' and 'exec' call involved since
> this is the way a Linux system (and Windows I think)
> implements the "pipe" concept. So, I tend to stick
> to the parent/child fork and exec method retaining the
> "C" way to do things.

I wonder if I should base my implentation use GNAT.Expect instead. 
GNAT.Expect does all of the above and I don't like to reinvent the weel
just for the fun of it.

Only I was planing for line IO and not pattern matching. Well, I'll look
into it.

Thank you for your help. It enables me to make an informed desicion - which
ist probably the most important part in programming.

With Regards

Martin

--
mailto://krischik@users.sourceforge.net
http://www.ada.krischik.com