Roland Mainz wrote:
I don't want this to be a flamewar. The fact that I'm happy to make
these changes is no reflection what so ever on the ksh93 project and the
great work you are doing.
No, but if you want to maintain compatibilty to pdksh then we either
cannot use any newer POSIX or ksh93 constructs (like associative arrays)
in that code or we have to maintain to seperate sources - one for pdksh
(or port it to plain bourne shell and disallow any usage of newer shell
constructs to make sure the script becomes a real pain) and one for
POSIX/ksh93 shells.
Given that no such constructs are used in webrev I don't see a this
being a problem.
Personally if we are going to rewrite webrev from scratch I'd seriously
consider doing it in something other than shell script.
I have the fixes from the original submitted included in my version of
webrev that I'm adding Mercurial support to and it works just fine.
Now personally I like the changes this fix proposed because it puts the
code into the style I would have used originally, ie while do < rather
than cat | while do.
Did you actually look at what the fixes are ?
Yes, but my comment was of a more generic nature. And at some point I am
wondering more and more why I am doing all the work with ksh93 on
Solaris when some of the more interesting (potential) consumers are kept
at a strict "pdksh-compatibilty" level. Or short: It's frustrating.
The most significant changes in that bug are actually about things other
than ksh and are about finding and using an appropriate awk program
dealing with aspects of the "portability" to other SCM systems (like no
SCCS on Linux).
--
Darren J Moffat
_______________________________________________
tools-discuss mailing list
tools-discuss@opensolaris.org
