Don, we do not intend to make all closed utilities we replace builtins in ksh93. Making AST sed a builtin was just an easy thing because the code is well designed and easy to make re-entrant.
Olga On Fri, Mar 12, 2010 at 8:21 PM, Don Cragun <dcragun at sonic.net> wrote: > > On Mar 12, 2010, at 10:39 AM, Garrett D'Amore wrote: > >> On 03/12/10 10:32 AM, Don Cragun wrote: >>> On Mar 12, 2010, at 14:49:39 +0100, olga.kryzhanovska at gmail.com wrote: >>> >>> >>>> AT&T AST sed from the ast-open package. It is a direct descendant of >>>> the AT&T sed in UNIX. The same package contains ksh93, has the same >>>> license (CPL 1.0) as ksh93 and passes VSC. >>>> >>> While it is nice (and correct) to say that ksh93 passes the VSC tests >>> for sh, that is not an indication that the sed functionality built >>> into ksh93 will pass the VSC tests for sed. >>> >>> I assume that the Oracle sponsor you get for your putback for this case >>> will verify that this version of sed passes the VSC tests before you >>> putback. I also assume that you (with help from David and Glenn) will >>> fix any failures detected (either by changing the code or by getting >>> approved TSDs from TOG for test suite failures) if sed does not pass >>> the VSC sed tests. This is what has been done with all of the other >>> ksh93 project putbacks. >>> >> >> I don't think there was any indication anywhere that the bits would *not* >> pass POSIX conformance tests. Obviously a change of this nature >> (specifically to /usr/xpg4/bin/sed) requires such conformance testing to be >> done. Such testing, however, is not architectural in scope. (Conversely, >> if the team were going to abandon POSIX conformance, then that would be >> architectural in scope. But that is not what has been proposed.) >> >> - Garrett > > Garrett, > Alan asked if this version of sed passed VSC sed tests. Olga said > ksh93 passed VSC sh tests. I was just pointing out that Olga didn't > answer Alan's question. If I wasn't clear, I also thought I was > pointing out that the project team has always done the right thing > with respect to VSC. > > Furthermore, I am saying that I think it is implicit in any of these > cases that replace "standard" utilities with ksh93 built-ins passing > the appropriate VSC tests is an assumed checklist item before putback > will be allowed (and, therefore, doesn't need to be explicitly stated > in the ARC case). > > Alan, > In a later message, you asked: "And what about the POSIX conformance > testing?" > > POSIX conformance is a subset of UNIX certification. The UNIX > certification VSC sed tests are all that is needed for POSIX > certification as long as Oracle actually maintains the UNIX branding > that has been in place for more than 20 years. Note, however, that > since Sun laid off the UNIX standards conformance group there has been > no formal statement from Sun or Oracle that there is any intent to > renew the brand as part of the next Solaris major or minor release. > (Patch releases of Solaris 10 are covered under the current branding > documents.) > > Cheers, > Don > _______________________________________________ > shell-discuss mailing list > shell-discuss at opensolaris.org > http://mail.opensolaris.org/mailman/listinfo/shell-discuss > -- , _ _ , { \/`o;====- Olga Kryzhanovska -====;o`\/ } .----'-/`-/ olga.kryzhanovska at gmail.com \-`\-'----. `'-..-| / Solaris/BSD//C/C++ programmer \ |-..-'` /\/\ /\/\ `--` `--`
