dynamic named directories clobber $reply

dynamic named directories clobber $reply

[Mikael Magnusson]

I noticed that after a successful ~[something] substitution, ~[] or
~[invalid] would return the same substitution, because $reply leaks to
the parent shell, and my function doesn't unset $reply when it doesn't
find anything valid. If I make it local in the function it doesn't
work at all, is there some way it can be local in the hook somehow?

-- 
Mikael Magnusson

Re: dynamic named directories clobber $reply

[Bart Schaefer]

On Jun 4,  2:24pm, Mikael Magnusson wrote:
} Subject: dynamic named directories clobber $reply
}
} I noticed that after a successful ~[something] substitution, ~[] or
} ~[invalid] would return the same substitution, because $reply leaks to
} the parent shell, and my function doesn't unset $reply when it doesn't
} find anything valid. If I make it local in the function it doesn't
} work at all, is there some way it can be local in the hook somehow?

I'd say the short answer to that is "no".  Nothing else that uses
reply/REPLY internally to the shell invokes any kind of local context
for them.

However, the hook conceivably could explicitly unset reply before the
call to the shell function.  This would have to happen in Src/utils.c
subst_string_by_hook(), and presumably only if a function or variable
named zsh_directory_name was actually found.

I'm undecided whether that's the correct thing to do, or whether it
really should be up to the shell function to make sure reply is correct
regardless of whether it found a match or not.  It's unclear because in
the case of an array of zsh_directory_name_hook functions, you may not
want each to go clobbering the results of the one before.

Re: dynamic named directories clobber $reply

[Peter Stephenson]

On Sat, 4 Jun 2011 14:24:14 +0200
Mikael Magnusson <mikachu@gmail.com> wrote:
> I noticed that after a successful ~[something] substitution, ~[] or
> ~[invalid] would return the same substitution, because $reply leaks to
> the parent shell, and my function doesn't unset $reply when it doesn't
> find anything valid.

In addition to Bart's reply, this would only be an actual bug if it was
happening when your function returned non-zero, which is what it's supposed to
do if it doesn't have a match.

-- 
Peter Stephenson <p.w.stephenson@ntlworld.com>
Web page now at http://homepage.ntlworld.com/p.w.stephenson/

Re: dynamic named directories clobber $reply

[Mikael Magnusson]

On 4 June 2011 23:24, Peter Stephenson <p.w.stephenson@ntlworld.com> wrote:
> On Sat, 4 Jun 2011 14:24:14 +0200
> Mikael Magnusson <mikachu@gmail.com> wrote:
>> I noticed that after a successful ~[something] substitution, ~[] or
>> ~[invalid] would return the same substitution, because $reply leaks to
>> the parent shell, and my function doesn't unset $reply when it doesn't
>> find anything valid.
>
> In addition to Bart's reply, this would only be an actual bug if it was
> happening when your function returned non-zero, which is what it's supposed to
> do if it doesn't have a match.

Yeah, it turned out I forgot to return 1 on failure in my function. I
was worried about the hooks not having any way to unset reply so they
wouldn't clobber eachothers' results, but then I realized I was
confusing it with the completion case where they all get to add stuff.
In the n case though, as soon as one hook returns true no more hooks
are called so it should be fine.

-- 
Mikael Magnusson