[Numpy-discussion] Issue Tracking

Fernando Perez fperez.net at gmail.com
Tue May 1 17:44:41 EDT 2012


On Tue, May 1, 2012 at 1:36 PM, Jason Grout <jason-sage at creativetrax.com> wrote:
> This example indicates that basing  your decision on what it is like
> *today* may not be valid either.  You'd hope that they won't do

Very true ;)

> Anyway, like everyone else has said, Ralf, Pauli, et. al. are really the
> ones to vote in this.  Given Fernando's responses, I realize why GHI
> still works for us---our small project has me and 2-4 students, and we
> all pretty much meet each week to triage issues together, and there are
> only about 40 open issues.  It's a simple enough project that we need

I just reread my reply with a full stomach, and I wanted to add
something, because I think it may appear a bit too negative.  *In
practice*, the system does work very fluidly, and other than the
no-labels-on-PRs, it just gets out of your way.  Being able to simply
type #NN in any comment or git commit ('closes #NN') and get
everything auto-linked, closed if needed, etc., has major value in
practice, and shouldn't be underestimated.

Using two separate tools adds real friction to the everyday workflow,
and if GH has taught me one thing, it's that the very fluid workflow
they enable leads to massive productivity improvements.  For IPython,
the change from Launchpad/bzr to GH/git has been truly night and day
in terms of productivity.  We process a volume of code today that
would be unthinkable before, and I think a big part of that is that
the tools simply get out of our way and let us just work.

So, as much as I do complain about real problems with GHI, I also
think it's important to evaluate carefully the cost of a dual-system
solution.  Sometimes the lesser tool you know how to use is better
than the fancier one that creates friction.  Put another way: no
matter how fancy your new $400 racket is, you'll never beat Pete
Sampras on a tennis court even if he's using a wood board to play :)

Cheers,

f



More information about the NumPy-Discussion mailing list