svn commit: r209622 - head/sys/dev/ixgbe

[Gleb Smirnoff]

  Jack,

On Wed, Jun 30, 2010 at 10:09:22PM -0700, Jack Vogel wrote:
J> I went to a LOT of trouble setting up a mirror at home and just as I went to
J> commit
J> the change its updated.
J> 
J> While I understand the good intentions, please don't do this again. If I had
J> been
J> unresponsive for days or something I understand, but its been hours, and I
J> was
J> fixing it.

We can't forecast future. Yesterday, I couldn't tell whether you will
be unresponsive for days or for hours. Do you propose a policy to wait for
days before putting an obvious fix? Live for days with a broken build?
Waste peoples time, receive mails from tinderbox, receive mails on current@
from real people?

Every time you^Wsomeone breaks build, a lot of manhours are wasted. Dozens of
developers encounter breakage and spend their time on fixing it in their
working trees. Not mentioning common FreeBSD users, who are brave enough to
run CURRENT.

If the first committer, who encountered the breakage, commits a fix, then
a lot of manhours are saved, avoiding all others doing this job for their
own trees. And the breakage passes by almost unnoticed.

So, anytime I update sources and encounter an obvious error, I WILL FIX IT,
to save time of other people, since I respect time of other people. And
you do not, because you never do even a simple test build of changes you
commit, not speaking about run tests. Skipping these important steps, you
save your own time, for the price of other peoples time in the order of
magnitude.

Look at the commit history of your drivers: every second check in, or even
more often, either build is broken, or some important change is backed out,
or functionality of the driver is broken. Don't you think you should do
more testing before checking in?

-- 
Totus tuus, Glebius.