John Plocher wrote: > Roland Mainz wrote: > > Michael Shapiro wrote: > ... lots ... > > I just don't see the "why" here. > > Sure, one /could/ do all sorts of spiffy things at build time iff one > could just use ksh93. > > But, I don't see anything in the ON tree today that /requires/ the use > of ksh93.
Erm, the problem we have is that all new Makefiles in the ksh93-integration prototype004 tree use SHELL=/bin/ksh to simplify some things like include file generation and l10n support (e.g. message catalog generation) and someone pointed out during pre-pre-review that this is currently not allowed. The whole stuff could be re-written to use /bin/sh at the expense of readabilty/mainablity and at that point I am simply wondering whether it could be possible to allow an alternative (with /bin/ksh as intermediate step). This would also help removing existing limitations (e.g. argument size, string/line length limitations we hit (and worked around) in the past in usr/src/lib/libast/Makefile.com). > In the absence of a problem that forces the adoption of ksh93 /now/, > I'd suggest that we keep our focus on getting the current ksh93 project > finished and integrated before we invent more work. Right. But this wasn't thought as "... ordering another large chunk of tunnel..." just because we saw a light in one direction (either tunnel exit or the 22.05h express train) ... I was only trying to scratch carefully at the door of the gatekeepers and ask if this is "Ok" (note that ksh93 is used during the message catalog build phase (which forces us already to a two-stage commit, first with message catalog generaetion "off" and then later enable this part (short: xx@@@@@-"flag day" (and I am trying to be nice and already prepared binary packages (see http://www.opensolaris.org/os/project/ksh93-integration/downloads/2007-03-18/) for build machine adminsto lower the pain for them)))). > If needed, get a > limited scope exception that allows the use of ksh93 as part of the > ksh93-specific build process, but treat it as an implementation detail > private to the ksh93 project. Erm... I'd like to avoid the "limited scope" part if possible. If gatekeepers allow this it may be nice to allow this for others, too. > Being "done" with this phase of ksh93 > /now/ is infinitely better than some theoretical "perfect" that is never > achieved. Right (and we're hopefully near that goal... I am currently testing ast-ksh.2007-03-22 which fixes more or less all issues which came up during the pre-pre-review discussion (and more warning fixes, code cleanup, C99 fixes, "-xalias_level"/"-fstrict-aliasing" optimisations now working, multibyte/CSI fixes (e.g. you can now use non-ASCII charcters in variable/function/etc. names (even in ja_JA.PCK and zh_CN.GB18030)), security fixes to the "restricted shell mode" cage, etc. etc. ... we're certainly not sitting around and do nothing...)). ---- Bye, Roland -- __ . . __ (o.\ \/ /.o) [EMAIL PROTECTED] \__\/\/__/ MPEG specialist, C&&JAVA&&Sun&&Unix programmer /O /==\ O\ TEL +49 641 7950090 (;O/ \/ \O;) _______________________________________________ opensolaris-code mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/opensolaris-code
