>
>On Mon, 3 Nov 2008 13:09:33 -0600 Nicolas Williams wrote:
>> On Mon, Nov 03, 2008 at 02:00:34PM -0500, Glenn Fowler wrote:
>> > varargs functions are only useful when you have all of the
>> > arguments in place at compile time
>> >
>> > a portable post-compile-time interpreter wishing to use the varargs
>> > interface
>> > would be out of luck
>
>> But posix_spawn() already has a vectored interface. Are you arguing
>> in favor of the original proposal with a single string and some argument
>> boundary escape scheme?
>
>just my 2 cents on varargs functions
>thinking (erroneously) that only a varargs api was proposed
The initial proposal has one interface which uses "string we could
try to parse into words".
In we're saying that you really should have a varargs function because
we're trying to do something other than posix_spawn does.
I think that we have a problem with a "parse them strings into them
words".
We can live with execv() vs execl(), e.g.,:
systemlist(arg, arg1, argv2, ..., NULL)
systemarray(argv);
But we already have the second function and that is why we think we just
needed the first function.
Casper
