[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: more about Zookeeper

Hi Ted,

thank you a lot for your email! I really mean that. It would be wonderful to 
get more communication going between Apache and Debian. I've learned that 
these two large and important free software projects are nevertheless 
different in many ways.

I also appreciate very much the nice conversations we had last week. I could 
learn a lot from you.

About the ZooKeeper package and partly also Hadoop/HBase which I also 

a) The first and most important reason for orphaning these packages is, that I 
did quit the job where I used them. It's neither possible nor adequate to 
maintain a package that you don't use yourself. - So as long as there isn't 
anybody else who'd like to maintain these packages, we can stop the discussion 

b) I think that there is actually very little interest for 
Hadoop/HBase/ZooKeeper packages _in_ Debian. ( There is interest for packages 
_from_ Cloudera or Apache _for_ Debian, but that's something else.) Even the 
company I worked for switched to use the Cloudera distribution, because they 
felt more confident about it. Now there'll be the Bigtop project and people 
will rather use the "official" apache releases which are tested by 
Yahoo/Facebook/Cloudera then the Debian packages.
So without any discussion about ZooKeepers Code quality, it might just not be 
worth to do the packaging work if nobody wants to use it.

[1] http://mail-archives.apache.org/mod_mbox/incubator-

c) A Debian stable release with ZooKeeper (/Hadoop/HBase) is not possible 
without a long term support commitment from upstream. I don't see, how Debian 
could fix bugs in ZooKeeper and test those without upstream helping with 
insights and test infrastructure.

d) From what I've heard, read and experienced about the history and culture of 
Apache and Debian, quality has different priority in these two projects. 
Debian has a "Culture of technical excellence"[2] and is very keen about its 
independence from companies. I experienced the Hadoop projects as very company 
driven where quality is not nearly as important as in Debian.
Now we have different priorities and standards. I propose we don't enter a 
discussion about those as long as points a)-c) aren't resolved.

[2] Stefano Zacchiroli (Debian Project Leader): "Debian: 18 years and 
counting" http://conferences.ellak.gr/2011/files/2011/05/debian.pdf

I'll have a university paper to deliver until Thursday and will do the 
orphaning afterwards.

Best regards,

Thomas Koch, http://www.koch.ro

Reply to: