Code in containers

[annshress]

Since different developers are working with different operating systems, I think we should consider using docker for development so that two developers will have a same experience.

Example: bioformats2raw (an application used for converting .czi files to .zarr) doesn't work with M2 chip, but works on linux.

[blowekamp]

I agree having a Docker file to build a reproducible environment would be very useful, and make it easier for developer so set up the correct environment.

To be a little more specific, bioformats2raw works in a linux docker container on with the "linux/amd64" platform option on the Mac ARMs.

[mbopfNIH]

I have bioformats2raw running on my M1 chip using Intel-based Python and Java installations via Rosetta. I don't believe the M2 introduced new architecture changes, so the above solution should still work.

[annshress]

Yes. It's working in a linux docker container. I am still suggesting that we use Docker for the purpose of reproducible environment.

[mbopfNIH]

Yes, I agree that a common Docker dev environment would be helpful. One drawback to doing this could be performance, especially when testing out resource-intensive routines, which bioformats2raw can be. So I think it is useful to have the application installed directly on your machine without having to go through Docker. Although, setting up the intel-based environment is a pain.

[blowekamp]

It looks like we didn't add the--max_workers=$nproc options to bioformats2raw before. That will make that command run nearly nproc-X faster so hopefully the performance difference will be less of an issue.

[annshress]

If you are running locally and want to test the workflows, we recommend using docker container. Look into Docker instructions in the documentation.