avro-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Doug Cutting <cutt...@apache.org>
Subject Re: [VOTE] Avro release 1.4.0 (rc1)
Date Mon, 30 Aug 2010 22:07:42 GMT
On 08/30/2010 02:48 PM, Scott Carey wrote:
> Is it expected that one can unpack the src tarball and run build.sh?
> the 'dist' target does an svn export, so it breaks in an unversioned
> directory.  I suppose this is ok if the rest succeeds.

Perhaps we should make that fail with a more friendly error message, and 
perhaps even make it work with git, but we currently rely on the source 
control system to determine the set of source files.

> On Java, tests are not passing for me (from the tarball or trunk) :
> https://issues.apache.org/jira/browse/AVRO-644

Ugh.  I think Patrick is on vacation this week.  Philip?

> I don't have the proper requirements installed for php and have not
> completed a build.  We need some documentation on build
> requirements.

I added the list of packages I needed to install to the BUILD.txt in the 
src tarball.  It was just php5, phpunit and php5-gmp.  Perhaps this 
information should be duplicated in PHP's README too.

> I'm unfamiliar with Ruby, and it is complaining about yajl meanwhile
> 'gem install' doesn't know what yajl is and I'm stopping at that
> point for now.  Again, we might want to document the build
> requirements  (so far for me: echoe and yajl).

The packages I had to install on Ubuntu to build Ruby are listed in 

> Since I'm rather Ruby and PHP ignorant, I'll look past those.  I
> would like all the java tests to pass though.  Unfortunately, the one
> that is failing above works fine from eclipse or if I run it in
> isolation, and only fails when in batch with the others.  I won't
> have time to look into it in more detail today.

I have also seen that test fail sporadically but not consistently.  This 
is new code that no one is using yet, so I'm not convinced that possible 
bugs in it should stop the release, but I also don't like to ship a 
release that doesn't pass unit tests.  So, if we can't figure this out, 
perhaps we should temporarily disable these tests?

Thanks for reviewing this, Scott!


View raw message