cloudstack-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Noah Slater <nsla...@tumbolia.org>
Subject Re: [AFSCS40] Release status for CS 4.0
Date Tue, 02 Oct 2012 20:50:56 GMT
Cool, thanks Chip.

As a follow up, Apache httpd do "pre-release" versions for testing on the
dev@ list.

"This directory contains pre-release test versions of the Apache HTTP
Server and are not officially released tarballs. Please use only for
testing."

http://httpd.apache.org/dev/dist/

I am not involved with the httpd project, but I am guessing they use this
as a staging area for builds that people want the community to test, and
for hosting artefacts that are being voted on. (For a point of reference,
in CouchDB Landâ„¢ we call an RFC on the proposed release to catch any
concerns before the length release and voting process.)

On Mon, Oct 1, 2012 at 4:29 PM, Chip Childers <chip.childers@sungard.com>wrote:

> On Sun, Sep 30, 2012 at 1:09 PM, Noah Slater <nslater@tumbolia.org> wrote:
> > Minor correction, "build" is official ASF nomenclature.
> >
> > "A Build is a package which is not suitable for distribution to the
> general
> > public. They are works-in-progress, and as such the only people builds
> > should be offered to are other people working on product development at
> the
> > foundation."
> >
> > "Release candidate" is not, however, and is best avoided. We don't really
> > do release candidates at apache, so if you use it to talk about voting
> > artefacts, you risk confusing people who assume you're using the more
> > common sense of the word. If something is alpha, call it alpha.
>
> I like the distinction you made here, specifically that we stop using
> the term release candidate until we are talking about something that
> we will / are voting on.  To me, a release "candidate" is exactly
> that:  a candidate being voted on.
>
> Let's start using those terms from now on.  The weekly builds that
> I've been doing (source only) are just that - builds.
>
> -chip
>
> > On Sun, Sep 30, 2012 at 12:23 PM, Noah Slater <nslater@tumbolia.org>
> wrote:
> >
> >> Chip,
> >>
> >> Can you point me to where you're hosting these builds?
> >>
> >> We need to be super careful about the distinction between the following
> >> items:
> >>
> >>    - A build
> >>       - A source or binary package that will note be voted on
> >>    - A release candidate
> >>       - A source package that is being voted on
> >>    - A release
> >>       - A source package that has been voted on
> >>
> >> (Please note that "build" and "release candidate" are not official ASF
> >> nomenclature. You could call a build a "package" or a "nightly" or a
> >> "tarball" or whatever it happens to be. Build kinda works for most
> things.
> >> It's a preparation of the source. And release candidate might just be
> >> called "the voting artefact" or what have you. It's the thing we're
> voting
> >> on for a release.)
> >>
> >> A binary package will never be anything more than a build that is
> prepared
> >> by an individual for the convenience of others. So let's just get that
> out
> >> of the way. A binary package can never graduate to anything other than a
> >> build.
> >>
> >> A source package can do many things though.
> >>
> >> On the one hand, as an individual, we can prepare source packages as
> >> snapshots, or nightlies. A committer can upload them to
> people.apache.org,
> >> and advertise them to developers. (But we must not advertise them to
> users
> >> without heavy caveating.) Generally, these are used by people working on
> >> the project itself, not with the project. Though, we may want to
> highlight
> >> a particular build before starting the official release process, to get
> >> feedback on things.
> >>
> >> If we think we're ready for a release, we can prepare a build to vote
> on.
> >> We upload this to people.apache.org, and we start a VOTE thread. At
> this
> >> point, the build becomes a release candidate. And only at this point.
> (Also
> >> note that because a release candidate must progress to a release without
> >> any modification, we cannot include "RC" in the version number.)
> >>
> >> If the vote passes, the source package becomes a release, and we upload
> it
> >> to our dist dir and tell users about it.
> >>
> >> In this context, language like this concerns me:
> >>
> >> "each RC build should come with release notes"
> >>
> >> These are not release candidates unless we're voting on them, and they
> >> certainly must not come with release notes.
> >>
> >> If an individual is providing nightlies, let's call them nighties, and
> >> let's attach "QA notes".
> >>
> >>
> >> On Tue, Sep 25, 2012 at 8:13 PM, Chip Childers <
> chip.childers@sungard.com>wrote:
> >>
> >>> On Tue, Sep 25, 2012 at 3:08 PM, Alex Huang <Alex.Huang@citrix.com>
> >>> wrote:
> >>> > Chip,
> >>> >
> >>> > In the future, we should.  If we were doing this right (which we
> aren't
> >>> today), each RC build should come with release notes about what QA has
> >>> found to be problems.  I think what you're putting up right now are
> more
> >>> closer to nightly unstable builds than RC builds.
> >>>
> >>> Agreed on that front.  Really though, I'm not doing a "build".  I'm
> >>> packaging the code only.
> >>>
> >>> We're in a weird state right now, since we won't be able to pass a
> >>> vote yet.  The way I see other projects doing this, is that even
> >>> unstable builds / source packages can be considered a release.  They
> >>> just get labeled something like "alpha" or whatever.  The projects do
> >>> vote on them though (which we're not ready for).
> >>>
> >>> I guess I'll just keep incrementing for now - so that those people
> >>> looking at the source package know that it's a new version (vs. Citrix
> >>> QA, which I believe is pulling unofficial builds from Jenkins for
> >>> functional testing).
> >>>
> >>> -chip
> >>>
> >>> > The good news is that the quality has been pretty good so even the
> >>> nightly unstable builds are good.  Today, that's mostly due to the
> majority
> >>> features in this release came from Citrix and were already tested by
> >>> Citrix.  For future releases, we should follow a process of QA
> completes
> >>> 100% testing and that's a RC build while there's nightly builds for
> people
> >>> who are actually developing if they're interested.
> >>> >
> >>> > --Alex
> >>> >
> >>> >> -----Original Message-----
> >>> >> From: Chip Childers [mailto:chip.childers@sungard.com]
> >>> >> Sent: Monday, September 24, 2012 8:14 PM
> >>> >> To: <cloudstack-users@incubator.apache.org>
> >>> >> Cc: cloudstack-dev@incubator.apache.org
> >>> >> Subject: Re: [AFSCS40] Release status for CS 4.0
> >>> >>
> >>> >> Alex,
> >>> >>
> >>> >> I've been cutting "RC" source code bundles, and have been numbering
> >>> them
> >>> >> as RCx (Wednesday will be RC3).
> >>> >>
> >>> >> Do you think I should switch to a more informal scheme until we
have
> >>> >> something to vote on officially?
> >>> >>
> >>> >> - chip
> >>> >>
> >>> >> Sent from my iPhone.
> >>> >>
> >>> >> On Sep 24, 2012, at 10:27 PM, Alex Huang <Alex.Huang@citrix.com>
> >>> wrote:
> >>> >>
> >>> >> > Hi All,
> >>> >> >
> >>> >> > I've been reminded that I've neglected to update the community
on
> the
> >>> >> current status for CloudStack 4.0.  I apologize for that oversight.
> >>>  From now til
> >>> >> the actual release, I will give a daily update on the status. 
If
> you
> >>> feel anything
> >>> >> is missing, please let me know and I'll try to include them on
the
> >>> next update.
> >>> >> >
> >>> >> > Summary
> >>> >> > As of 9/24/2012, CloudStack 4.0 release has past code freeze
stage
> >>> (over
> >>> >> three weeks ago).  A source code branch has been forked and is
> called
> >>> 4.0.
> >>> >> Nightly build is running on Jenkins on the 4.0 build.
> >>> >> >
> >>> >> > Feature List
> >>> >> > There are two features that missed the 4.0 release.  Auto-Scaling
> and
> >>> >> Brocade Plugin.  Both are due to having significant code changes
due
> >>> past the
> >>> >> code freeze date.
> >>> >> >
> >>> >> > Code Readiness
> >>> >> > - There are ~5 code related reviews on the review board scheduled
> >>> for 4.0.
> >>> >> Most of them are waiting for review and commit.
> >>> >> > - There are <10 bugs on Jira for the first cut of the release.
> >>> >> > - Upgrade from previous versions is currently being worked
on.
> >>>  Scheduled
> >>> >> to be done by the end of the week.
> >>> >> >
> >>> >> > License Readiness
> >>> >> > - Majority of the VM configuration issues have been resolved.
>  There
> >>> is one
> >>> >> remaining wrt rsyslog.conf.  Thanks to Chiradeep and Chip.
> >>> >> > - Technology export issues are still be worked on by David
Nalley
> >>> and AFS
> >>> >> legal.  This may be a blocking issue.
> >>> >> > - Licenses need auditing.
> >>> >> >
> >>> >> > Doc Readiness
> >>> >> > The current plan for docs is to write an INSTALL.TXT to give
> >>> instructions on
> >>> >> how to install from source.  All docs will be generated to a living
> >>> document
> >>> >> that continues to improve past the release.  The link to this living
> >>> document is
> >>> >> to be determined.
> >>> >> >
> >>> >> > TODO:  Docs need help on the new features in the 4.0 release.
> >>>  Specifically
> >>> >> we need help with Niciria Integration and Caringo documentation.
> >>> >> >
> >>> >> > QA Status
> >>> >> > QA is proceeding through the test cycle.  It is currently
at 45%
> of
> >>> completion.
> >>> >> The number of bugs generated from the tests have been minimum so
the
> >>> >> quality of the release currently looks pretty good.
> >>> >> >
> >>> >> > Release Plan
> >>> >> > - The current plan is for QA to complete its test cycle between
> >>> 9/26-9/28.
> >>> >> > - When QA decides the test cycle is read, we will cut a RC1
> release.
> >>> >> > - We are currently pushing to clear bugs generated from the
test
> >>> cycle asap.
> >>> >> > - After all P1 and P2 bugs are cleared, 4.0 release will be
> >>> submitted for
> >>> >> approval to announce.
> >>> >> >
> >>> >> > --Alex
> >>> >> >
> >>> >> >
> >>> >
> >>>
> >>
> >>
> >>
> >> --
> >> NS
> >>
> >
> >
> >
> > --
> > NS
>



-- 
NS

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message