[bip] Reproducible research

[James Casbon]

2009/3/4 James Taylor <james at jamestaylor.org>:
> I think the bigger problem is that for reproducibility, sometimes a
> ten year old version is exactly what you want. If you are trying to
> reproduce an analysis that you did in the past.
>
> This is one of the biggest problems we (Galaxy) face with
> reproducibility in workflows -- none of our data providers yet allow
> us to go back and re-issue the exact same query with a guarantee of
> the same results.

A good argument for queries to exist as idempotent urls?   This would
support reproducibility and other nice things like caching.