Closed ihodes closed 8 years ago
I am having second thoughts about biopam
, in particular as a source for Biokepi
tools. In particular:
biopam
doesn't have the volume of tools necessary to compete with other options (ex. bioconda
).That makes sense; are the problems you've seen with Bioconda unrelated to conda/related to how much of a pain it is to package some of these tools?
The problems that I've experienced with bioconda
in particular stem from trying to fight it. Conda follows the python
development practice of resolving things when needed, so adding things to a PATH
variable. Fighting against that by trying to find a specific library is onerous.
The packaging pain is just a constant factor worse than biopam; onerous documentation and more baroque fields to fill out. But on the plus side you can just add for a specific version of c++
to be installed too, if you have a weird dependency.
In that case, is the recommended install path just regular opam?
No, opam is better suited for projects in OCaml.
I am reticent to push tools that have such a unique utility into the public domain/repositories. My current preference would be bioconda, but for biokepi purposes I think we should maintain our own repo. Whether it should be opam
or conda
based is an open question; how many forks and of what projects does the lab want to maintain.
Pushed to opam
to get going on this for now.
Would be nice to use the biopam infrastructure in Biokepi to get this in!