>Uh, just to be pedantic, _why_ are we still carrying this baggage?  Is
>there an actual, genuine need to retain K&R compatibility going forward
>forever?  I'd have thought at some point Solaris could just say "K&R no
>longer supported", and rip out the K&R headers, along with the busted
>/usr/ucbinclude and (most especially!) /usr/ucb/cc.

I recently noticed that someone used /usr/ucbinclude (Specifially to
use the extremely mis-designed BSD signal functions)

>Are there real customers who still insist on K&R?  Are people still
>wanting to compile their old SunOS 3 or 4 applications from source?  I
>know the compatibility guarantees are nice and all, but at some point,
>it seems that it can be taken too far.  (I'm not proposing breaking
>binary compatibility, only source compatibility.  If people want to
>update their applications for Solaris, then they should really update
>and quit screwing around trying to run them in "compatibility" mode. 
>Heck, I imagine that it would not be terribly hard to write a tool to
>help developers identify -- and suggest alternatives -- code that has to
>be changed for Solaris.)

Since the compilers still have a mode which accepts prototypes
but also non-prototyped code, we could rip out the non-prototyped
headers.

>Certainly it seems that every other OS on the planet that is still being
>actively developed has dropped any pretense at supporting pre-ANSI
>compilers.

Indeed; so perhaps we should.

>I realize that this would probably be a nightmare at ARC, but I think
>that eliminating pre-ANSI support could do huge wonders for improving
>maintainability and readability of Solaris sources.  And the old SunOS
>/usr/ucb/cc is probably the cause of more pain and suffering amongst
>ISVs (especially those that support source code distributions) than
>nearly any other "feature" of Solaris.

I'm not sure why you'd think this is a nightmare at ARC; we would
need to look at which standards we (claim to) support.  If there are
non which require K&R, then it's a no-brainer, ARC-wise.

We do have a bit of non-ANSI code, I think, we stil compile, the
SunOS 4.x binary compatibility bits.  Not sure which compile mode we
use, but I think it still needs the old CPP.

Casper
_______________________________________________
opensolaris-code mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/opensolaris-code

Reply via email to