Re: git: 8ecb1f814114 - main - Revert "lang/rust: Update to 1.63.0"

On Thu, Aug 25, 2022 at 03:42:09PM +0200, Piotr Kubaj wrote:
> On 22-08-25 15:24:12, Baptiste Daroussin wrote:
> > On Thu, Aug 25, 2022 at 02:21:28PM +0200, Piotr Kubaj wrote:
> > > On 22-08-25 14:08:26, Baptiste Daroussin wrote:
> > > > On Thu, Aug 25, 2022 at 10:50:30AM +0000, Piotr Kubaj wrote:
> > > > > The branch main has been updated by pkubaj:
> > > > > 
> > > > > URL: https://cgit.FreeBSD.org/ports/commit/?id=8ecb1f8141144c1603eb4026122d2e60eeaccd64
> > > > > 
> > > > > commit 8ecb1f8141144c1603eb4026122d2e60eeaccd64
> > > > > Author:     Piotr Kubaj <pkubaj@FreeBSD.org>
> > > > > AuthorDate: 2022-08-25 10:29:24 +0000
> > > > > Commit:     Piotr Kubaj <pkubaj@FreeBSD.org>
> > > > > CommitDate: 2022-08-25 10:49:36 +0000
> > > > > 
> > > > >     Revert "lang/rust: Update to 1.63.0"
> > > > >     
> > > > >     This reverts commit f36cc819accadae1fefc14c203e5fbe9c74e1492.
> > > > >     
> > > > >     I asked for an exp-run for powerpc and powerpc64 on a recently provided
> > > > >     cluster machine but got no reply.
> > > > >     
> > > > >     PR:     265915
> > > > 
> > > > You cannot take the decision to revert on your own for many reasons:
> > > > 
> > > > 1. etiquete: we ask people to actually revert if needed and escalate if they
> > > > don't, we never revert directly ourself! the only one allowed to do "rude" revert
> > > > are people taking care of the package building when something break the delivery
> > > > of packages for Tier 1
> > > I did it myself so that users don't upgrade in the meantime, since that
> > > would create further confusion. If I let it wait, then PORTEPOCH should
> > > have been bumped for all the affected ports, which I think we all agree
> > > would be troublesome.
> > 
> > No what will happen is given it works on Tier1, the Tier2 will fix one by one
> > the issues that raised if any without reverting, meaning no PORTEPOCH would be
> > needed.
> 
> Isn't it better to fix issues BEFORE committing?
> 
> > > 
> > > > 
> > > > 2. an exp-run happened and was granted by portmgr, exp-run are costly we cannot
> > > > expect exp-runs for non tier 1 architecture for every changes (beside the fact
> > > > :x
> > > 
> > > I do not ask for exp-runs "for every change". In fact, this was the 1st
> > > time I asked for an exp-run on powerpc*.
> > > 
> > > If you can't expect exp-runs on non-Tier 1, why do we have powerpc*
> > > exp-run machines?
> > 
> > We do not have those machine functionnal! so we cannot do those exp-run, and
> > even if we had, it will be at the discression of the people doing the exp-run to
> > see if they have the resources or not to add an extra target for exp-run.
> The attached email says the contrary.

I can see your level of trust here in what I am saying, I am part of both portmgr
and clusteradm, so if I say portmgr do not have powerpc machines available for
exp-run, you can be sure that I double checked.

> 
> > > 
> > > This also wouldn't have happened if antoine@ had explained earlier he's
> > > not able to do exp-runs on powerpc*.
> > 
> > While I agree that antoine should have answered (maybe he missed the messages)
> > it does not mean that you are allowed to make a driven by revert.
> > > 
> > > >  
> > > > 3. powerpc* is not yet a Tier 1 so there is no promise on the delivery of
> > > > packages.
> > > 
> > > Yes, but everyone can still build from ports.
> > 
> > yes but again this is tier 2! same lack of warrant
> If you talk about "warranty", is there a "warranty" that all the ports
> build on Tier 1?
> 
> Warranty implies some paid support and compensation when things don't
> work. So there's actually no warranty, we are not RHEL.

Common you perfectly know what it means!

Bapt