@luispedrocoelho @rbuels There's much effort involved in getting code up from "works fine in the lab" to "good enough to put on sourceforge"- @luispedrocoelho @rbuels But I'm not supposed to spend my lab's research time on handling input contingencies and user documentation.
- @luispedrocoelho @rbuels If no resources exist to write hardened code, a journal mandate is hollow.
- @luispedrocoelho @rbuels IMHO bringing code to distro level == 0.2-0.4 FTE. ~$50K/yr. That's means losing 1 postdoc (benefits inc.)
- @luispedrocoelho @rbuels Until funders mandate & fund code distro, research(labs writing good codebase) << research(labs that don't).
- @luispedrocoelho @rbuels Most labs do not have $$$ to produce distributed level code. Most scientific code == pipeline hacks.
- @luispedrocoelho @rbuels Same reason you should trust clinical results produced from non-distributable patients.
- @rbuels @luispedrocoelho Lab that does that ends up with rep for distributing messy code & not supporting it. Liability issues aside.
- @luispedrocoelho @rbuels because in the lab you do not need INSTALL README and doc/ files to the extent you need on sourceforge.
iddux
Iddo Friedberg
Computational biologist. Interested in gene and protein function prediction, prediction of function from structure, metagenomics http://iddo-friedberg.net
Share & notify
Notify
140
