Garrett D'Amore wrote:
> Roland Mainz wrote:
> > Richard Lowe wrote:
[snip]
> As to this particular case, I feel limited use in some applications is a
> different proposition than globally searching ON and converting most of
> it wholesale.

Erm... that was not my intention. I am NOT planning to take over OS/Net
and/or convert every application in OS/Net which can't climb up a tree
fast enough to use libast or turn it into a ksh93 shell script... :-)

> In the former case, limited library use makes sense, in
> the latter case maybe this functionality really should be part of libc.

Please take a look at my other email for that...

> As to whether there should be a giant paradigm shift towards some common
> stdio from what we are doing, I agree with the original poster that
> "justification is required".  Why would we want to change?

Possible options/justifications (in the case of using libast in OS/Net)
include (but not limited to, just what I still remember at 6:03am on
Sunday):
- Improved performance (for example in almost all cases where we did
benchmarks sfio was faster than Solaris's stdio implementation)
- Better functionality (Ok, that's debateable since some of the benefits
may be the result of the enforced usage of XPG6/C99)
- Reduction in code size
- Reduction in memory footprint
- Simplification of code

> What are the
> benefits?  And the costs?  From what I've seen so far, you have a
> tendency to see costs vs. benefits with a certain subjective view, that
> doesn't necessarily concern all use cases (e.g. not all Solaris users
> are writing ksh scripts that need to work with terabytes of data.)
> Examination of the tradeoffs for major changes is certainly, IMO, warranted.

For ksh93 (and only for ksh93) my tentency is too look at the current
and _future_ usage and even cover the one or other "edge usage" which is
still important like support for the zh_CN.GB18030 or ja_JP.PCK locales
(which are both "pet projects" of the respective governments) or making
sure not to establish artificial barriers like adding a new shell to
Solaris which is not capable of dealing with 64bits (this is a minority
right now but I expect that this may play a more important role in the
future. There are already complains about the 32bit-only "perl" in
Solaris and the "solution" of NIH was to move the applications over to a
platform which has a 64bit "perl" by default (suprise suprise, this was
not Solaris...). Why should a shell which has support for large arrays
and trees NOT have 64bit support ? 64bit is one of the MAJOR selling
points of Solaris...) by default.

----

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

Reply via email to