Re: [psh-dev] wishlist

[Omer S. <osh...@in...>]

On Tue, May 02, 2000 3:35 PM Richard P. Groenewegen wrote:
> According to the mailing list archives on April 5, Daniel McSorley
> asked if it would be desirable (if not already present) to have a
> bash-style
>
> cmd1 && cmd2
> and
> cmd1 || cmd2
>
> behaviour.  He also offered to implement this himself.  As far as I
> can see his remarks were ignored on the list.  I for one would like to
> see these constructs present in psh and I hope the silence on the
> list hasn't scared of Daniel McSorley.

I'd like to apologize to Daniel McSorley on behalf of all of us for ignoring
him. I think we each have our own idea of what code we maintain, and each
assume that the code in question is always "someone else's job". I promise
that I will at minimum reply to every ignored message on this list;
apologizing on other's behalfs is one thing, but I won't presume to commit
their time for them. :)

> Still, it would be even better
> if there was a file stating that &&/|| constructs are present /
> desirable / undesirable / worked on / not for now / whatever.  This
> would also make it easier for the casual programmer such as myself to
> make a contribution without interfering with design choices made by
> core programmers.

Yes, I agree. This would make it easier for people to write big patches
without worrying about them being rejected over some decision made months
before. If everyone would please send me (or the list) everything they can
think of, I'll compile the entries and offer them to the list for dicussion.

> It seems to me that the former is more expected shell behaviour.  But
> is this something that will ever be?  Or are there fundamental
> obstructions that leave us no choice to accept the Psh::evl-situation?
>
> I'm very enthusiastic about the psh and I guess a list of intended
> features will only stimulate my enthusiasm and satisfy my curiosity.
>
> cheers,
> Richard

There are some basic and unavaidable problems. When I type
$ chmod u+x file
do I want Perl's builtin chmod() or system("chmod", ...)? This namespace
collision has no easy solution. (Note that in this case, making the wrong
choice would result in an error, because Perl's chmod isn't symbolic.) Psh,
as you may have discovered if you read through the manpage, has a list of
evaluation strategies to run a command through, which though they are
customizable, don't provide a magic bullet to kill namespace collisions
without the user specifying (e.g. the "! command" hack) the namespace they
want. There was some talk a while back about replacing this code with a more
flexible OO version, but I don't think that's been implemented. It's still
on my personal TODO list for when I have some time.

--
Omer Shenker                     osh...@in...