svn commit: r266760 - head/contrib/bmake

[Warner Losh]

On Jun 4, 2014, at 9:18 PM, Julian Elischer <julian at FreeBSD.org> wrote:

> On 6/4/14, 11:19 PM, Drewery, Bryan wrote:
>> On 6/4/14, 2:26 AM, Eitan Adler wrote:
>>> On 2 June 2014 23:35, Baptiste Daroussin <bapt at freebsd.org> wrote:
>>>> On Mon, Jun 02, 2014 at 09:46:38PM -0700, Simon J. Gerraty wrote:
>>>>> 
>>>>> On Mon, 2 Jun 2014 19:58:31 -0500, Bryan Drewery writes:
>>>>>> Not sure if anyone cares, but this change breaks all ports tree
>>>>>> checkouts from before 2014-05-05 on src head with this revision.
>>>>> 
>>>>> Yes, an older ports tree would need an older bmake (or fmake).
>>>>> Are we saying ports is *not* ready for that hack to be removed?
>>>>> 
>>>> 
>>>> IMHO it is, just people has to be aware of that :)
>>> 
>>> ports-announce@ is the correct location for this announcement :)
>>> 
>> 
>> I think that would just confuse people. I don't think this is worthy of such a big announcement. I can write up something to current@ and ports@ though. It's really only an issue if you are trying to use older ports tree and why would you anyway in most cases?
>> 
> We have a scenario where we check out a ports tree at one revision, but then need to slide parts of it back and forth to get to a specific revision of a port that we need. We can not affort to re-verrify every port revision every month, so it stays at an old revision generally but individual ports my upgrade if there is a security risk or may remain on anold revision if a newer version breaks thins for us. (it happens).
> 
> If that breaks we will not be happy

I plan on reverting this because it is premature.  Too many people have their own forked ports trees that haven’t necessarily been updated. The update is simple, true enough, but it is surprising. There’s no harm in leaving that code there for a while longer since it is opt-in now. I may add a warning to highlight its use...

Warner