Closed sc13-bioinf closed 4 years ago
Is this really the only way to get this into conda? Via precompiled binaries? Is there no way I can make a stack install
pipeline within bioconda or something? @maxibor, can you confirm this as well?
Creating a conda package needs two things:
A meta.yaml
file for recipe metadata
A build.sh
that contains the compilation instructions, which will be compiled by bioconda.
Here are the official Bioconda instructions for building a conda package for Haskell: bioconda.github.io/contributor/guidelines.html#haskell
Here is a bioconda recipe of a Haskell package github.com/bioconda/bioconda-recipes/tree/master/recipes/ngless
Thanks a lot. I'll have a look and see whether I can get that organised.
So the reason why it is so hard is because conda recipes target centos6. It turns out that haskell is too modern to build on this os. The only way to do would be to build an entire modern os from scratch in your recipe so you can then build haskell. Static binaries are therefore the only way to go until conda moves to something more recent.
OK, I've decided to unassign myself from this and not pursue it further myself. It has low priority for me. There exists a simple and cross-platform way to install this, and making it even more convenient with conda is nice to have but not necessary. If someone wants to take this on, let me know, I'm happy to grant temporary write access to this repo if need be.
The bionconda recipe for sequenceTools will give another way for people to install it easily. It will also make it easy to incorporate into eager2. The only way to get sequenceTools running in a bioconda recipe is by installing a pre-compiled static binary. To make this process easier, Travis CI can be used to build and publish this static binary automatically. Therefore there are the following things to do: