Ok, I am changing my vote to:
[0] Abstain
See below for details.
On Sun, Oct 25, 2015 at 10:10 AM, Andrea Pescetti<pesce...@apache.org>
wrote:
Damjan Jovanovic wrote:
[-1] Disapprove
That's bad! Let's see if we can manage to explain it and at least get a 0,
since I don't see anything really blocking in your reports (I mean, we do
not require that the release has no issues at all, and we agree that what
we are releasing is better than 4.1.1 was).
The deal breakers:
1. The source tarball (tar.bz2 at least) contains main/MacOSXX64Env.Set
and
main/MacOSXX64Env.Set.sh. I don't believe they belong in there (they're
not
in trunk), and if they do, they're missing ASLv2 licenses and cause
unapproved license errors in the RAT report.
This is true but it is cosmetic, meaning that this is simply a generated
file that was archived by mistake. If you configure the sources, you will
get this or the corresponding file for other platforms. It is not in our
tagged snapshot at
https://svn.apache.org/viewvc/openoffice/branches/AOO410/main/ and while
it would be possible to simply remove it, I still see it as simply cosmetic
(but I'm available to discuss whether we should remove those files; this
affects only the sources, it is a merely cosmetic change since those files
are unused or overwritten during the build, and it has zero effect on the
binaries so we needn't new binaries; and it does not even require a commit,
since it is a problem with assembling the .tar.gz file).
I suppose since the file is generated, unused, and was present in previous
releases, it shouldn't stop this release.
Apache release policy may require those files to be removed or copyright
notices added because "Every source file must contain the appropriate ASF
License text" (http://www.apache.org/dev/release.html#license) and "Every
ASF release *must* comply with ASF licensing policy. This requirement is of
utmost importance and an audit should be performed before any full release
is created. In particular, every artifact distributed must contain only
appropriately licensed code" (http://www.apache.org/dev/release.html). But
since the file is a generated file and not a source file, that may not
apply?
2. The source fails to build on 32 bit Xubuntu 14.04 both on VMs and
physical hardware (details later).
We officially support Windows, Mac OS X and Linux. In terms of "baseline",
the reference for Linux is, for historical reasons, CentOS 5 (a still
maintained, but quite old, version; this means that our binaries can run on
virtually all distributions); build works there. There are hundreds of
other distributions: with the new bugfixes, OpenOffice will build on recent
versions of Fedora and Ubuntu (64-bit). Xubuntu 32-bit might be one of the
platforms where it is fine to describe how to fix the build (more below).
Ok, good that we use CentOS 5 for binaries. My concern is that if it
doesn't build on 32-bit Xubuntu, it won't build on 32-bit Ubuntu either, as
they are very similar. But I'll check that at some stage. Since most people
don't run Linux, and most Linux users won't be building from source, and of
those building not all will be on 32-bit (X)ubuntu, and we can document the
workaround, I guess it's ok.
None of the below should stop the release; I was just commenting generally
and explaining why some tests weren't done.
Please can we see links to RAT reports, changelog, test results, code
quality metrics, and other useful info in the emails proposing a release.
Yes, sorry for not providing them. Here we are:
- RAT reports: https://ci.apache.org/builders/openoffice-linux64-rat
- Changelog: we traditionally have not had a CHANGELOG file. This link
provided by Dennis
https://bz.apache.org/ooo/buglist.cgi?bug_status=CONFIRMED&bug_status=ACCEPTED&bug_status=REOPENED&bug_status=RESOLVED&bug_status=VERIFIED&bug_status=CLOSED&f1=flagtypes.name&list_id=170870&o1=substring&query_format=advanced&resolution=---&resolution=FIXED&v1=4.1.2_release_blocker%2B
will show you what changed. I'll update our Release Notes, a subpage of
https://cwiki.apache.org/confluence/display/OOOUSERS/AOO+4.1.2 , with a
more readable version of it.
- Test results and code quality metrics are not something we are required
to provide (and we actually never did). Does this mean they are not
important? No, they are! We may want to make them part of the process for
trunk, where we now have a better situation thanks to your work.
Neither Google Test nor the unit test changes themselves were backported
from trunk to 4.1.2, and many old cppunit tests would fail and break
This is known. An important thing to keep in mind is that 4.1.2 is closer
to 4.1.1 than to trunk: we can't backport all the nice things we have on
trunk. In many situations, 4.1.2 is an improvement to 4.1.1, and also in
this respect 4.1.2 is no worse than 4.1.1.
================
Xubuntu 14.04, 32 bit
================
The binary package installs. Java 7 is automatically detected and used.
The source reproducibly fails to build (on both VMs and physical hardware)
in main/svl with this error I reported on the dev list months ago
This seems the typical situation where we would add a comment in the
"known issues" section of the Release Notes saying what to expect, and how
to fix it. A similar situation is the Java 8 compatibility when building
the SDK. In both cases we provide something that people can build on a
variety of easily available systems; and we refer to the Release Notes for
less common cases.
All these fail on trunk as well, and trunk has 15 additional fvt test
failures and several errors not present in 4.1.2 :-(.
OK, then I recommend that focus in on trunk for tests. We can't backport
everything and we have to live with the fact that 4.1.2 will be exactly
like 4.1.1 in some respect. For sure, it is no worse than 4.1.1.
Otherwise thank you everyone for this RC, and hopefully the next one will
be released.
I hope we will not need a next one! I'm available, as said, to discuss
recreating the source archive and removing the two "temporary" files that
were archived and that are not in SVN. I'll open a thread for this as soon
as I have thoroughly checked it.
The rest of your comments can probably be addressed in Release Notes or
(in the case of tests) on trunk since those improvements are out of scope
for 4.1.2. I don't see any of them as regressions or as blocking 4.1.2.
Thank you for a terrific, very detailed, review. Even though we won't be
able to accommodate all of it into 4.1.2, nothing of it will be wasted. For
sure, you provided plenty of good information to add to the Release Notes,
so you get credit for helping me improve the Release Notes too!