svn commit: r220755 - in head: . contrib/gcc/doc contrib/gcc/objc contrib/libobjc etc/mtree gnu/lib gnu/lib/libobjc gnu/usr.bin/cc gnu/usr.bin/cc/cc1obj gnu/usr.bin/cc/cc_tools gnu/usr.bin/cc/doc s...

[Warner Losh]

On Apr 19, 2011, at 9:29 AM, John Baldwin wrote:

> On Tuesday, April 19, 2011 10:28:23 am mdf at freebsd.org wrote:
>> Trimming since I have a mostly-unrelated question...
>> 
>> On Tue, Apr 19, 2011 at 5:40 AM, John Baldwin <jhb at freebsd.org> wrote:
>>> On Monday, April 18, 2011 3:59:45 pm Warner Losh wrote:
>>>> In this case, there was a new kernel thing just after, so it turned out OK.
>>>> But let's not gratuitously bump the version since the granularity we have
>>>> already allows the ports to make good choices on when to leave something in or
>>>> out.
>>> 
>>> Except that that directly contradicts our previously established policy that
>>> these version bumps are cheap and that we should do more of them (this came up
>>> a few years ago when we changed the policy so that the new "stable" branch
>>> after a release starts at N + 500 (e.g. 802500) rather than N + 100 to give
>>> more room for version bumps on current).
>> 
>> I thought I remembered reading (within the past 2 years) that
>> __FreeBSD_version should not be incremented more than once a day,
>> since there was a limit of 100 before the version minor number was
>> affected.  Did I get the polarity backwards and that was the old
>> policy?
> 
> Well, I would avoid more than once a day still, but the 100 limit is now 500
> in 8.0 and later (we had more than 100 bumps during 8.0-current which resulted
> in a discussion where we chose to raise the limit to 500 rather than
> discourage bumps in current).

There were times in the 8.x release train when I got hit by this problem a lot.  I'd update to get a fix in some other part of the tree, and there's be another bump even though I had compiled a kernel just hours before.  While I can live it it from time to time, there was a stretch where it happened to me all the time and it wound up costing me a substantial portion of a working week.

It is all about windows.  If there's a small window since the last bump, please biggy back on it.  If there isn't, by all means bump.  I'd tune small measured in days rather than a single day, since it is rare that ports need to know with such precision when something happened and small windows tend to impact fewer people than the bumps do.

Warner