Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 19 Dec 2017 19:23:20 -0600
From:      Mark Linimon <linimon@lonesome.com>
To:        Jan Beich <jbeich@FreeBSD.org>
Cc:        Mathieu Arnold <mat@FreeBSD.org>, Mark Linimon <linimon@FreeBSD.org>, svn-ports-head@freebsd.org, svn-ports-all@freebsd.org, ports-committers@freebsd.org
Subject:   Re: svn commit: r456719 - in head: biology/bedtools comms/gnuradio editors/libreoffice emulators/stella games/armagetron graphics/gsculpt net/linknx net/rsplib science/afni security/botan2 security/hig...
Message-ID:  <20171220012320.GA28486@lonesome.com>
In-Reply-To: <bmiu-jjv4-wny@FreeBSD.org>
References:  <201712191441.vBJEfTLf054114@repo.freebsd.org> <8tdy-a671-wny@FreeBSD.org> <7ae87444-6657-7506-f987-14ec50f648f6@FreeBSD.org> <bmiu-jjv4-wny@FreeBSD.org>

next in thread | previous in thread | raw e-mail | index | archive | help
Let me address the overall concerns first, and then give my personal
opinions.

Now that I've finished a sweep through all the aarch64 errorlogs, I
intend to go through things one-at-a-time to catch the cases I've missed.
In particular, it seems some things should be labeled "newer clang error"
than "aarch64 error."  It's sometimes hard to tell at first glance.

I will probably wait until the results are in from my armv6 -trybroken
run here at the house to do so.  (I have also been going through the
armv6 errorlogs, and doing preliminary armv7 runs, also with -trybroken).

My personal theory is that if I'm going to be looking at hundreds if not
thousands of errorlogs, I'm going to miss things.  And yes, I assert that
I have gone over more than a thousand errorlogs this year, if you count
all the tier-2s on various buildenvs.

When I started looking at arm* last year, we had nearly a thousand failed
ports indicated on poudriere, that merely showed as "build error".  We were
building these ports on the cluster over and over again, pointlessly.  Soon
we won't be.

Soon, anyone(TM) can go to the poudriere result pages and browse the Ignored
section and quickly see what the errors are.  I am sure that a little work
on data reduction will show common errors that may be more easily fixed in
groups.

It's more than one person's (e.g. myself's) worth of work.  But the intention
is to get us to the point where most of the ports on tier-2 "Just Work".

I need a lot of collaborators.

I see what I've been doing as necessary brute-force work to get armv6
(really, should be armv7) and aarch64 up to near-parity with amd64, so
that we can then open up the question of "are these archs tier-1, and
if so, do we expect maintainers to fix PRs filed against these archs."

That policy decision, and the related question of "what resources do we
provide to maintainers if they don't have them", is a portmgr decision.
When I left portmgr it was entirely to a) no longer be involved in such
decisions and b) avoid the continual, prolonged, drama and meta-discussions.

And by the latter, I *specifically* mean this thread, and the dozens of
similar ones that have preceded it over the years.

And those last two paragraphs are about all I am going to say about my
own opinions on the matter.

tl;dr: yeah, I've broken some eggs, but I'm trying to make an omelette.

mcl



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20171220012320.GA28486>