On Thu, Feb 13, 2014 at 03:00:21AM -0500, Francesco Romani wrote:
Recently, we've had a bugs http://gerrit.ovirt.org/#/c/24242/ that could have been avoided had we used pylint to check our code before its usage. Two other bugs-in-waiting can be quickly removed http://gerrit.ovirt.org/#/c/24313/ http://gerrit.ovirt.org/#/c/24412/
I would like to suggest http://gerrit.ovirt.org/#/c/24382/ "make check: add a pylint check"
We have a long way to go before pylint is happy with our code, but I believe that pylinting some of our modules is a good start.
The down sides are many: it's slow, it's another dependency, it has false negatives, and I do not yet understand how it behaves (particularly, the interdependency between checked modules).
What do you think? Should we add it?
+1
My take: I am for it, or at very least to give it a try, following the path you suggested.
I use pylint in a few other projects and it saved me from some bugs and led to improvements. IMO it is worth its price.
While I'm waiting for more acks/nacks, could infra add "pylint" to our Jenkins slaves?
Regarding Nir's on-gerrit comment: I'd like to have pylint run by default in order to avoid adding new pylint-detectable bugs. If we do not add a pylint_blacklist now, I do not believe we'd ever make it empty.