cvs commit: src/sys/compat/linux linux_socket.c

[Brooks Davis]

On Thu, Mar 10, 2005 at 06:19:04PM +0000, Paul Richards wrote:
> On Thu, Mar 10, 2005 at 09:36:46AM -0800, Brooks Davis wrote:
> > On Thu, Mar 10, 2005 at 05:20:50PM +0000, Paul Richards wrote:
> > > On Thu, Mar 10, 2005 at 11:55:10AM -0500, Brian Fundakowski Feldman wrote:
> > > > 
> > > > How is this even a problem?  The company targets 5.3-RELEASE, period,
> > > > for all of 5.x.  They do not need to upgrade their build computer to
> > > > anything past RELENG_5_3 even in the face of security issues.
> > > 
> > > That really irritating bug that causes their development tools to
> > > crash is fixed in 5.5. Clearly they'll want to run with that later
> > > version.
> > 
> > Then they should tell their customers that they only support 5.5 or they
> > should routinly test on 5.3 to make sure that keeps working.  For that
> 
> That's the crux of FreeBSD's problem really. We can't tell the
> commercial world to change the way they work to suit what we want
> to do. If we want them to take FreeBSD seriously we have to fit
> with the expectations they have.
> 
> > matter, in the case we are discussing, as long as they don't use the new
> > flags they would not have any problems and if they failed to test on the
> > old version they claimed to support that's their problem.
> 
> Across a stable branch I think it's unrealistic to expect commercial
> developers to keep track of all these nuances. These are the sorts
> of development overheads that make costs prohibitive and thereby
> rule out support for particular platforms.

If the company doesn't test their software on all the releases they
claim to support or track changes between, how can they possibly expect
it to work?  Even making the assumption that no new features were added,
how can they be sure they aren't using a feature that caused kernel
panics in the previous release?  As far as I can tell, there is no
difference between adding a feature and fixing a bug that prevented a
feature from working before.

-- Brooks

-- 
Any statement of the form "X is the one, true Y" is FALSE.
PGP fingerprint 655D 519C 26A7 82E7 2529  9BF0 5D8E 8BE9 F238 1AD4
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/cvs-src/attachments/20050310/65b82835/attachment.bin