4 ms·
On testing culture in GitHub projects
- aeden 14y agoCan anyone actually get to the site right now? I was getting database connection errors but now the site is not even responding.
- lsinger 14y agoTotally embarrassing. I'm trying to get it running again.
- psylence519 14y agoThat is the site, demonstrating what a lack of testing gets you. It's brilliant.
- lsinger 14y agoNice one. :)
- RossM 14y agoCache link: http://webcache.googleusercontent.com/search?q=cache:blog.leif.me/2012/09/github-testing/&hl=en&client=ubuntu&hs=3Ve&channel=cs&prmd=imvns&strip=1 http://webcache.googleusercontent.com/search?q=cache:blog.le... and link to the paper concerned: http://se.uni-hannover.de/pub/File/pdfpapers/Pham2012.pdf http://se.uni-hannover.de/pub/File/pdfpapers/Pham2012.pdf
- lsinger 14y agoEither I fixed the problem or traffic has simply gone down enough.
- RossM 14y agoI'm a huge fan of drive-by commits. GitHub's web editor makes it really easy to correct a typo or add a two-line fix - and for many people that might be their first contribution to open-source, without the faff of downloading git etc.
- dmethvin 14y agoMaybe for documentation or source-comment fixes, but most code fixes should have one or more unit tests to accompany them. To me, a pull request represents something that should be pretty close to finished work. There may be a few style issues to clean up but it should be functionally done. If someone wants to discuss how to approach or solve a problem in general without working and tested code, I'd prefer to see that happen in the bug/issue rather than a pull request. Otherwise it can lead to wasted effort and/or hurt feelings on the part of the contributor.