Craig Mohrman wrote:
> 
> April is back but very behind in emails.
> 
> I think we are closing in on a start of codereviews so I want
> to make a laundry list of last minute items and make sure we are good
> to go.
> 
> 1) build stability - Craig/Roland
>         I have yet to get a full x86 build to go through due to ksh93 tests
>         failing.  I believe I can now isolate it down to 1 or 2 tests
>         that fail at random times.

Could you post the matching test output (just the test output, not the
whole build log) here, please ?

>         sparc builds run all the way through but still unstable at times.
>    1a) whether the tests should be in (and run) the ON source tree vs
>        a separate tree.

Seperation of the test suite sources is not possible as they are
delivered to usr/demo/ksh/tests/ , too. And there are the usual problems
that 1) the test gate is not open (yet) and 2) the test gate and the
OS/Net sources must be 100% in sync (which appears to be a tricky and
on-going problem as far as I heared in the past).

>        I'll let the code review process deal with that.

What about this: We disable the test run in the default build but keep
the "testshell" target around ? AFAIM that should be acceptable for
gatekeepers as the test suite would no longer disturb the build...

> 2) ksh93 getconf - Glenn/Roland
>         Sounds like Glenn is getting close here and we have to wait for this.
> 
> 3) ctfconvert:pics/wordexp.o: Couldn't parse stab "#vla0:(0, 14)" - Roland
>         I believe this is a performance issue which Roland could
>         bug fix later?

It's not exactly a performance issue since the code is doing a
|fork()|,|exec()|-sequence later... it was originally thought to test
the C99 VLAs in OS/Net code. Originally it worked perfectly (with Sun
Studio 10) but now it bites back badly with Sun Studio 11. And a fix for
this problem is not available and it doesn't seem to be available in the
next month. And even if there would be a fix we could need a "flag day"
to deal with the problem that the ctf* tools need to updated to reflect
the change.
Or short: This is hopeless (and I am close to develop depressions here
because yet another feature in the tree has been slaughted... ;-( ).

I just rip this stuff out and replace it with non-VLA code... ;-(

> 4) build cleanup - Roland/Craig
>         I gave Roland a bunch of make clean/clobber stuff to go do.
>         Still one output popping out of the build that should not.
>         [attached]

Attached... where ?

----

Bye,
ROland

-- 
  __ .  . __
 (o.\ \/ /.o) roland.mainz at nrubsig.org
  \__\/\/__/  MPEG specialist, C&&JAVA&&Sun&&Unix programmer
  /O /==\ O\  TEL +49 641 7950090
 (;O/ \/ \O;)

Reply via email to