Hi,
I went through the list* of eligible change sets for the 0.9.x
branch to get closer to the 0.9.1 release.
I blocked or merged all of my commits, and all new stuff
from Damien, Chris, Adam, Paul and Noah. I'm sorry if
I stomped on anyone's feet here, but I figured I might as
well go through the whole list while I'm in there.
I blocked anything that has remotely to do with "new"
or "refactor" and only merged a few fixes.
We still do have some eligible change sets open and I
do believe all of them should be merged into 0.9.x, but
I don't feel qualified enough to comment on the stability.
These change set are:
------------------------------------------------------------------------
r758093 | damien | 2009-03-25 00:48:33 +0100 (Wed, 25 Mar 2009) | 1 line
Fix for crash when compacting an empty database
------------------------------------------------------------------------
------------------------------------------------------------------------
r763858 | damien | 2009-04-10 04:21:37 +0200 (Fri, 10 Apr 2009) | 1 line
Fixes for leaked file handles, with test.
------------------------------------------------------------------------
------------------------------------------------------------------------
r765364 | kocolosk | 2009-04-15 23:21:23 +0200 (Wed, 15 Apr 2009) | 2
lines
URL-encode attachment paths during replication
------------------------------------------------------------------------
------------------------------------------------------------------------
r766338 | nslater | 2009-04-18 17:20:00 +0200 (Sat, 18 Apr 2009) | 1
line
create /var/run/couchdb during init script
------------------------------------------------------------------------
------------------------------------------------------------------------
r766883 | damien | 2009-04-20 23:34:46 +0200 (Mon, 20 Apr 2009) | 1 line
Fix for process leaks with retrying compactions.
------------------------------------------------------------------------
------------------------------------------------------------------------
r771474 | kocolosk | 2009-05-05 00:25:23 +0200 (Tue, 05 May 2009) | 2
lines
standalone attachment GETs should respect "rev" qs param
------------------------------------------------------------------------
------------------------------------------------------------------------
r771480 | kocolosk | 2009-05-05 00:36:46 +0200 (Tue, 05 May 2009) | 2
lines
use revisions when replicating attachments. Closes COUCHDB-337
------------------------------------------------------------------------
Adam, Damien, Noah: Can you comment on your changes
and whether they should go into the 0.9.x branch?
When we're through with these, I believe we're ready to start
the 0.9.1 release.
--
Finally, and I'm the first to follow, it'd be great if we all could
get into the habit of either merging or blocking commits to
trunk for the 0.9.x branch.
As a reminder.
Blocking REVISION:
svn merge -c REVISION --record-only trunk branches/0.9.x
svn commit branches/Y.Y.x -m "blocked REVISION from trunk"
Merging REVISION:
svn merge -c REVISION trunk branches/0.9.x
svn commit branches/Y.Y.x -m "merged REVISION from trunk"
Cheers
Jan
--
* for i in `svn mergeinfo trunk branches/0.9.x --show-revs eligible`; do
svn log -r $i;
done
On 6 May 2009, at 19:38, Damien Katz wrote:
On May 5, 2009, at 9:12 AM, Adam Kocoloski wrote:
On May 4, 2009, at 9:32 PM, Chris Anderson wrote:
Devs,
Are we ready for 0.9.1? My pet patch is in and backported, how
about yours?
I wonder if we should try to shore up the JIRA records of what's in
0.9.1. Currently I see
COUCHDB-306 Wacky error responses to malformed documents
COUCHDB-310 Fix hardcoded redirect to "/_utils/"
COUCHDB-311 Wrong encoded _external error message
COUCHDB-322 Specifying reduce=true on a view with no reduce does
not cause an error.
COUCHDB-334 With deferred commits and 100+ active dbs, CouchDB can
lose uncommitted changes
COUCHDB-342 url-encode attachment paths during replication
as well as one open ticket targeted for 0.9.1:
COUCHDB-328 [patch] allow futon reduce textareas to contain
accidental white spaces.
I'm sure there are other resolved/closed tickets missing from that
list. As far as my work is concerned, I think
COUCHDB-337 attachments from old/conflict revisions are not
accessible via standalone API
would be a decent candidate for backporting. It's a simple fix,
but the fact that we ran like that for months without anyone
noticing probably means it's low priority. Cheers,
I think the fix is an important one. The big problem isn't the
failures, it's when it doesn't error out that's the problem. It can
put the wrong attachment data into another revision, which is a form
of data corruption.
Adam
