[csw-dev] Problem with cdrecord on Solaris 8 x86

[Dennis Clarke]

>>
>> I'll let it die when I see it work. Which I have not.
>>
>> Dennis
>>
> I think this falls into the category of the 6th standard deviation,
> where there is probably 10-20 people still running Solaris 8/X86 and
> trying to use cdrecord.

You think there are that many ?

> If someone hasn't logged a bug about it not working, then I think it's
> probably best just to drop it, given the amount of wasted effort it will
> be if noone really needs to see it working.

I have a response to that.

Simply put, if I run a test run of some package on this baseline lowest
common denominator machine and the package does not function as expected
then that is the *valid* result of the test. It is a data point and I can
not erase it. It points to a failure somewhere.

If I test Ken May's GNOME packages on this exact same machine I eventually
get a good working set of packages. Does that mean they can be trusted? If
you are telling me that there is no reason at all to test on Solaris 8 x86
then by all means let me know and we can throw away the whole Solaris 8
build server. Why bother to build it on Solaris 8 if no one anywhere will
ever run it on Solaris 8 ?

No, I think that the ABI is a good concept and if something works well on
Solaris 8 x86 then it will work flawlessly on Solaris 9 and higher revs of
Solaris. If it does not run on Solaris 8 x86 then why would I assume it runs
anywhere else?

Either we do the testing or we don't.

Because we have William Bonnet working on XFCE and IF that will only ever be
tested on Solaris 10 x86 then hurrah for me and I'm a happy man. We can put
a bullet into this old test server. The same goes for my Sparc 20 here.

There has to be some baseline test criterion.  A firm statement that says
"it was tested and seen to be good on Solaris 8 x86 and on Solaris 8 Sparc".
If we can not say that .. then what are we doing here for the past five
years?

Please keep in mind that I can knee jerk respond as fast as anyone and if we
ditch the whole Solaris 8 test phase then great. Let's move forwards to
Solaris 10 where it is all working fine anyways.

Dennis Clarke