pre-commit hook for git about gensim HOT 4 CLOSED

I just put this in my .vimrc:

autocmd BufWritePre *.py  :%s/\s\+$//e

If you make github alter commits you push to it (if that is even possible?), you'll run into all kinds of issues.
With a simple editor configuration like above (or a local hook) you can keep your commits clean locally, this is simpler and less confusing (i.e. commit hashes won't change)

Either way, you should always review diffs when merging changes or picking patches. And when you do, it's easy to spot newly introduced whitespace (if you have git coloring enabled)

from gensim.

Enforcing checks on GitHub doesn't seem to be possible. On the other hand, adding the check (pre-commit hook) locally is trivial, I added the instruction to https://github.com/piskvorky/gensim/wiki .

Contributors will now be advised to read that page first, before modifying the code in any way.

from gensim.

The suggested sed line on that page doesn't work when lines have ^M at the end.
If you use \s instead of :space: it will clean up trailing tabs, spaces and the ^M's as well.
Also I don't get why you supply an empty argument to sed before the statement, it will interpret this as an empty statement and the statement as a filename

from gensim.

Dieter, the [:space:] class in sed contains all whitespace characters (not just space, the name is perhaps misleading).

Also, in OS X (and other BSD systems I guess), the -i parameters needs a mandatory extension: providing empty '' means no backup will be created.

If your system behaves differently, you can add your version to the wiki (with some explanation).

from gensim.