hadoop-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Allen Wittenauer <awittena...@linkedin.com>
Subject Re: Hadoop support for hbase
Date Mon, 10 May 2010 16:58:50 GMT

Let me understand this:

a) the hbase folks have been required to patch hadoop due to bugs
b) they have been doing this for X months now
c) we finally have momentum on getting 0.21 out the door
d) hey, let's make their life easier and take resources out of 0.21 by creating a branch

Are we willing to create a branch for everyone that asks? What happens if mahout/lucene/solr/joes-random-application
asks?  Is it a good idea to create a branch that hopefully has a very short life time?   Is
it the latest fashion amongst the PMC to request random branches?  

trunk
0.21
0.20
0.20-append
0.20-security-sure-hope-this-turns-into-1.0
0.20-lifecycle-management-stuff-that-seems-to-never-get-committed
...

I'm not PMC, but can I make requests too, since I have to patch what we use?

0.20-capacity-scheduler-is-broken-but-we-shipped-it-anyway
0.20-stop-using-bsd-and-gnu-only-extensions-during-forks
0.20-hadoop-devs-cant-spell

I'd rather have the HBase folks put pressure on the Hadoop PMC to get 0.21 out the door than
create a custom-to-be-thrown-away-in-less-than-a-year branch.  This way we *all* benefit rather
than a select, but vocal group.
Mime
View raw message