Martin Bochnig wrote:
On Wed, Aug 5, 2009 at 4:16 PM, Alexander Eremin<[email protected]> wrote:
Also we can divided these scripts for x86/64 and sparc versions - now
all stuff on one
Doing this for entire ramdisk images (IA32 vs. AMD64 and sun4u vs.
sun4v) is certainly a good idea, as we discussed before.
But for scripts it would be against the design of Solaris / SVR4 /
POSIX / everything good.
I don't agree. The scripts that Roland is talking about are ones
delivered as part of the operating system. The fact that they are a
script is an implementation detail that the end user or admin is not
supposed to know about or care about and more importantly they aren't
supposed to take advantage of it either.
"No user serviceable parts inside"
We can do do change the implementation of programs that have committed
output between shell, perl, C, Java.
Providing that shcomp can guarantee 100% compatibility of execution I
see no reason not to use it for core system things like /lib/svc/method
if it can save us space and improve execution time.
However I don't think this should be something that IPS is responsible
for in the specific case of shell scripts. We should IMO either never
use shcomp for a given path or always use it.
--
Darren J Moffat
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
