The reason we maintained CHANGES.txt in Hadoop was to bifurcate new features, incompatible changes etc.
The same can be done by coming up with a convention for commit msgs; also having 'git log' local makes it much more palatable than 'svn log'. In svn land CHANGES.txt is much more critical for developer productivity. Arun On Aug 28, 2012, at 8:13 PM, Roman Shaposhnik wrote: > On Tue, Aug 28, 2012 at 8:06 PM, Konstantin Boudnik <[email protected]> wrote: >> I am coming from Hadoop's land where a normal practice is to accompany any >> commit with an appropriate record in the CHANGES.txt. It seems to be much >> easier not to miss anything this way around, doesn't it? > > Understood. Personally, I'm a big fan of autogenerating CHANGES.txt > from the Git/SVN history. Not sure what others think. > > Thanks, > Roman. -- Arun C. Murthy Hortonworks Inc. http://hortonworks.com/
