Re: POSIX misunderstanding

[Albert Cahalan]

On Wed, 2004-08-18 at 23:34, Paul Eggert wrote:
> Albert Cahalan <address@hidden> writes:
> 
> > On Wed, 2004-08-18 at 13:49, Paul Eggert wrote:
> >> Albert Cahalan <address@hidden> writes:
> >> 
> >> > Well, so does the --lines option.
> >> 
> >> No, that uses an allowed extension.  It's not prohibited, the way that
> >> multi-digit options are prohibited.
> >
> > Where?
> 
> Guideline 3 says "Multi-digit options should not be allowed."
> That's an explicit prohibition.

I meant, where is --lines allowed, and how does this
not apply to -12345 as well? Look:

"Each option name should be a single alphanumeric character"

So the second "-" is a violation, along with the
ordering distinction between --lines and -lsn-ei.

> > the long options violate guideline 3.
> 
> I don't see why.  Guideline 3 is about short options.

I don't see any other kind of option allowed.

> > Why would the standard warn that some systems might not be standard?
> 
> The standard contains several warnings of that sort.  It's a pragmatic
> document.  It hasn't been taken over by lawyers (yet :-).
> 
> > the standard does indeed mention that the utility syntax guidelines
> > might be violated by some standard-conforming implementation
> 
> No, it doesn't say that.  It doesn't say that such an implementation
> conforms to the standard.

Quite frankly, if non-conflicting historical behaviors
are prohibited, then the standard isn't worth the paper
it's printed on. How likely do you think it is that Sun
and HP will drop the "head -42 ..." syntax? I'm quite
sure they'd manage to get certified in spite of this.
I do believe the "warning" is a loophole.

In any case, the important thing is that nobody gets
misled into choosing this pedantic config. You could
require that POSIXLY_CORRECT or POSIX_ME_HARDER be set
before disabling the historical options. You could
require that somebody passes "--pointlessly-pedantic"
to the configure script. Whatever, as long as nobody
trips on this. Stuff is needlessly breaking.

Perhaps a correction to the standard is in order,
if indeed you are correct about the lack of an
allowance for supporting historical behavior that
does not conflict with standard behavior.