hadoop-hdfs-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Evgeny Turnaev <turnae...@gmail.com>
Subject Cyhdfs announce
Date Sat, 15 Sep 2012 14:11:29 GMT
Hi all.
   This an announce of a new python wrapper around libhdfs called
cyhdfs. cy stands for cython.

   PyPI:  http://pypi.python.org/pypi/cyhdfs/0.1.2
   bitbucket: https://bitbucket.org/turnaev/cyhdfs/src

   License is bsd.
   Code development status is beta, but we are using it for 3 month in
production in a daemon that receives zmq messages and writes it to
hdfs using this lib. So the other api except writing files not so
tested in production.

   The original intention to write it was because i did not found
anything that simple to write/read from hdfs from python.
   Pydoop have hdfs api but it uses boost and we didn't managed to
compile it on FreeBSD. Also i wanted something lightweight and simple.
   I compiled it on freebsd/ubuntu with cloudera hadoop-0.20.2-cdh3u3,
hadoop-0.20.2-cdh3u5 and hadoop-1.0.3. To compile it you need cython
   But if you hadoop version higher than 0.20.2-cdh3u5 there are
chances it will compile without cython as cydhfs ships with
pregenerated cpp file for those versions.
   Compilation with hadoop 2.X.X  untested as we are not planning to
use it. If any person wants to add support of hadoop 2.X.X or just
maintain this lib i would gladly give all rights to repository and
PyPI to him.

   I'll be glad if anyone find this lib useful or it will be included
in hadoop distribution.

Турнаев Евгений Викторович

View raw message