On 8/6/07, Keith Bierman <[EMAIL PROTECTED]> wrote:
>
> On Aug 6, 2007, at 10:15 AM, Josh Hurst wrote:
>
> >>
> >> I hope Indiana is going to set this build flag to 'ksh' by default.
> >
> > +1
> >
> > Shipping ksh88 as /bin/ksh in Indiana would be a disaster - first
> > Opensolaris promises to replace the old ksh88 with a far superior
> > ksh93, invests a year of development and then we neglect the promise
> > and work. We'd be in deep dooo-doo if this happens. Neither would it
> > be a good omen for other Opensolaris projects.
> >
>
> I don't think anyone suggested that. The question, as I understood
> it, was would the thing called ksh be the Solaris traditional one or
> not (with ksh93 already in as ksh93).
>
> We seldom get a chance to break compatibility.

Sun already broke compatibility between Solaris 7 and Solaris 10. We
have firsthand experience with non-trivial breakage in our products
due to changes in Solaris ksh where Sun refused to fix the problems,
referring to 'standard conformance' as justification for these changes
(which no other Unix vendor implemented).

Fortunately ksh93 has an own test suite (which was successfully
defended by the project team against several attempts from Sun to keep
this part out of the project) which will hopefully (together with the
POSIX test suites) prevent such incompatible vendor derivatives in the
future.

> This is our one big
> shot at making some disruptive changes, so I'd vote to make them all
> now.

I wholeheartedly agree. I think the compatibility concerns at Sun are
a paper tiger - most ksh88 scripts work with ksh93 and you need to
search hard to find any scripts which will break.

Irek
_______________________________________________
indiana-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/indiana-discuss

Reply via email to