blanket portmgr approval vs. non-fixing changes

[Baptiste Daroussin]

On Tue, Jun 28, 2016 at 11:15:56PM +0200, Matthias Andree wrote:
> Am 28.06.2016 um 11:17 schrieb Baptiste Daroussin:
> 
> > What you are asking is part of the blanket in particular when changing things in
> > individual ports, we expect committers to have a look at pending PR (yes I know
> > I have been guilty of individual port change without sometime checking about
> > pending PR which was wrong from my side)
> > 
> > For sweeping changes this is a bit different as when a change touches a large
> > portion of the tree we can not expect the committer to have a look at each
> > individual ports.
> 
> Baptiste,
> 
> to give you a provoking counter example:
> 
>   By that logic, I would not have been expected to notice that the
> bitcoin garbage insisted on db48, I could just have killed it off and
> moved the bitcoin ports onto db5.  (That's stretching it a bit because
> there was Peter Wemm's objection to the DEPRECATED= tag on record already.)
> 
> 
> Meaning that, in this thread: I beg to differ on sweeping changes.
> 
> These do need a thorough review, and often a series of -exp runs, to
> keep the number of casualties low.  If I had gone by this policy of
> sweeping changes, we'd nuked all DB2, DB3 and DB4 ports and had force
> moved all the bitcoin and openldap ports and whatnot onto db5 without
> consulting anyone, and I guess we'd heard a lot more screaming than with
> the approach I chose, meaning look at several dozen of ports before
> committing the breaking and sweeping changes.
> 
> And I do think we should, opposite to what you are proposing, make the
> committer spend extra time for high-profile ports that entail sweeping
> changes to chase down the breaking change to, say, a library port.
> 

I might have been not explicit enough, of course any changes should be tested,
and of course high profile ports breaking means special attention and prevent
the sweeping change to actually happen.

Read the context... here we were speaking of changes that adds extra works for 
maintainers on individual ports and the fact that someone should check for PR on
everysingle port that are going to be touched. In that case on a sweeping change
like what mat did for removal of ${PORTSDIR} you cannot expect the committer to
check everysingle ports for pending PR, but of course you can expect that for
every single change sweep or not the committer has to test all the ports touched
by the change to ensure they do not break.

Bapt
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: not available
URL: <http://lists.freebsd.org/pipermail/freebsd-ports/attachments/20160628/87ed0c15/attachment.sig>