On 07/02/2012 02:04 PM, Jasha Joachimsthal wrote:
On 2 July 2012 10:53, Ate Douma <[email protected]> wrote:
On 07/02/2012 10:00 AM, Jasha Joachimsthal wrote:
It's a good idea to postpone the release and work on fixing the issues
first.
While I agree on the latter, I'm not sure we should postpone the release
(0.13).
As we're striving for a release every month I'd rather cancel/skip the
0.13 release and focus on fixing the trunk (0.14-SNAPSHOT) instead so we
hopefully have a more stable and reliable 0.14 release end of this month.
Postponing 0.13 would mean creating a separate branch based on the 0.13
tag and working towards a 0.13.1 release candidate (note that a 0.13-RC1 as
Chris proposed isn't really an option anymore as 0.13 already has been
tagged).
All this seems like unnecessary extra work to me, unless some major other
changes were planned/targeted for trunk which cannot wait.
Instead, I propose canceling the 0.13 release and work towards a better
0.14 release instead.
Ate
I'm okay with skipping the release and continue towards a stable 0.14 at
the end of the month. Whoever wants a sticky revision can refer to the 0.13
tag and create the artifacts themselves.
It looks like the discussion about the 0.13 release candidate has dried up and
nobody voted for it either.
My conclusion then is that everybody (lazily) agrees with canceling this release
and that we'll focus on improving and fixing the trunk instead.
The next planned release candidate then will be for version 0.14 end of this
month.
I will drop the release candidate staging repository and remove the binary
artifacts from the temporary rave builds folder on people.apache.org.
Thanks, Ate
On 2 July 2012 00:09, Chris Geer <[email protected]> wrote:
On Sun, Jul 1, 2012 at 2:54 PM, Ate Douma <[email protected]> wrote:
On 06/30/2012 05:25 PM, Franklin, Matthew B. wrote:
Unfortunately, I found a pretty big bug. When we cut over to the new
interface model, the rave-shindig classes began using the username as
the
opensocial id (similar to igoogle,etc) rather than the arbitrary
database
entity id. Unfortunately, when I made those changes, I didn't update
the
security token classes in rave portal. This means that any code in
shindig
that checks the security token id against the passed in userid will
fail.
This primarily affects appdata; which, IMO is a pretty big deal..
Apologies, but when you consider this with Ate's potential bug, I am
not
sure we should ship the release...
Matt, thanks for finding and reporting this. I agree this seems like a
rather serious bug.
I haven't had time yet over the weekend to dive deeper into RAVE-708 but
will try to find time for it coming days.
The merge of the model interfaces changes, the upgrade to OpenJPA 2.2.0,
and on top of that, the upgrade to shindig 2.5.0-beta2, all happened in
the
last week. Overall this release gives me a bit uneasy feeling of being
(too) unstable/unreliable and certainly as not enough tested.
I'd like to hear others opinion on it, but I'm currently inclined to say
we should hold off/cancel shipping this release.
Maybe we should take the coming weeks to better validate and fix/improve
the quality and reliability instead of keep rushing in more major
changes.
As well as JIRA could use a bit of scrubbing and cleaning up of
old/outstanding issues I think.
We are also entering the summer holiday period (I myself will be 3 weeks
away after next week) so maybe we should anticipate a bit slower
progress
anyway or at least lesser time or eyes available for properly review and
test major changes.
All in all, I'm hesitant to push out a lesser tested/validated 0.13
(unlucky?) version out.
WDYT?
I don't have a problem delaying this release. If there are some features
people really want from a non-SNAPSHOT version we could always tag this
as
0.13-RC1 or something like that since it does have a lot of good stuff in
it. I guess it depends on the length of the delay. Two things I'd like to
see happening are expanding both Jasmine and integration tests. It would
be
nice if we could get automated test coverage for the issues we found this
go around.
On that note, I've been looking into the Jasmine tests and have hit a
snag.
I'd like to have more of the jQuery capabilities available in the tests,
in
case we want to use more of jQuery in Rave. Right now we are stubbing out
the $ variable with just the bare minimum of what we need. Is there a way
we can use the real jQuery library and just overwrite the things we need
to
tweak?
Chris
-----Original Message-----
From: Ate Douma [mailto:[email protected]]
Sent: Friday, June 29, 2012 7:43 PM
To: [email protected]
Subject: [DISCUSS] Apache Rave 0.13 Release Candidate
Discussion thread for vote on 0.13 release candidate.
For more information on the release process, checkout -
http://www.apache.org/dev/****release.html<http://www.apache.org/dev/**release.html>
<
http://www.apache.org/dev/**release.html<http://www.apache.org/dev/release.html>
Some of the things to check before voting are:
- can you run the demo binaries
- can you build the contents of source-release.zip and svn tag
- do all of the staged jars/zips contain the required LICENSE and
NOTICE
files
- are all of the staged artifacts signed and the signature verifiable
- is the signing key in the project's KEYS file and on a public server
