Check for Python projects
Closed this issue · 14 comments
First of all, I must say that julython is an awesome idea and thank you for it.
I've run across an interesting thing in the leaderboard: http://www.julython.org/projects/gh-FNNDSC-chrisreloaded/
There would be nothing wrong with that project except for the fact that welll, it's not Python.
And that's against the rules.
I'm not sure how to fix this but perhaps we could have a look at GitHub API and see if the project is mostly writen in Python?
Thanks for julython anyway!
Boo, I hate cheaters :) Yeah, it is a tough one as it is on your honor. I don't wan't to get too strict, cause Julython.org itself has a lot of Javascript to make it go. The POST commit hook tells the files that are in the commit, so we could guess by the extension. I wanted to do more with the information. But the lead up to January was spent porting the site from Appengine. My Idea is to rank the commits a little bit. IE how many files are changes, how many lines, what types of files, etc.
Well, we don't need to guess the language. Github already does that. If you wouldn't mind calling Github API it might be fairly easy to see what is the most used language in the project (see http://developer.github.com/v3/repos/#list-languages).
I like the idea of changing the way commits are ranked. I believe that commits that change just one line should be treated the same way as total rewrites.
Yes, for github it should be simple. Although they still could guess wrong, so I would just like to flag the project in some way to signal we think it is not Python. Rather than rejecting it.
Some people are using bitbucket tho, have to check on their API. Not sure if they have that.
@mrshu I have removed the points from this project in the backend so they wont show on the leader board anymore. Thanks for spotting this!
@rmyers Even though that was not what I was aiming at, Thanks!
AFIK bitbucket doesn't have such an API. But I totally agree on flagging the project rather than removing it. The idea of julython is awesome and the language should not matter.
As it never should =).
hmmm, good point. Maybe during the next one we could relax the rules. I still think the best is to look at the extentions of the commits. That way we could tell how many commits were for each language and score accordingly. We could have Python vs Javascript vs Ruby vs language X. Good thing we have 6 months to decide ;)
Good idea with the competition between languages (although we'll have to carefully set the ranking rules to reflect the amount of good core produced)!
What also surprised me was the fact that not many people in the Python community knew about julython. I believe it deserves more publicity. But that should be discussed somewhere else.
I like the idea of competing languages. On the other hand, I committed some documentation changes in some of my commits which were in .rst files for python projects. How would those commits have been affected under the attempt to discern a project's language?
Well, 'rst' and 'txt' will count as documentation. But still will score for the project. So it won't matter what the majority of the project is you'll score points for the actual work done during the month.
It might be interesting to see if more points for documentation would result in more documentation =).
What do you think?
Docs? We don't need no stinkin' docs! Just read the source. :-P
Negative points for adding features without docs ;)
Quarter-points. Then give them the full point + 1 for documenting it.
multiple languages are now supported.