Re: [Request for review] rna-star

[Michael Crusoe <michael.crusoe@gmail.com>]

Hello Nilesh!

Thank you for noticing this.

Please add entries to debian/changelog in the future. I use `dch 'changed a thing'` followed by `debcommit -a` instead of using `git commit`, to ensure I document my changes.

In your patch you removed `-O3`, I've put that back to simplify the patch. It doesn't hurt to specify that twice.

I've pushed my cleanups. Can you also add logic to compile rna-star multiple times using -mavx2 on down for amd64 along with a wrapper script?

https://salsa.debian.org/med-team/mmseqs2/-/blob/32b4e057b5d0e205f32b0b5184433ac155405198/debian/rules#L20 is a good example (along with https://salsa.debian.org/med-team/mmseqs2/-/blob/32b4e057b5d0e205f32b0b5184433ac155405198/debian/bin/simd-dispatch )

On Fri, 26 Feb 2021 at 13:44, Nilesh Patra <nilesh@debian.org> wrote:

Hi Michael and other folks,

The version 2.7.8a of rna-star which Steffen uploaded a couple of days ago seems to use intel intrinsics which leads unfortunately builds on non-amd64

Since simde is a good way to mitigate this, I tried using this -- on testing on a arm64 porter box, build goes green and autopkgtests (I manually an them) look good as well.

Can I please get a review of my changes?

Also, I'm not sure if it'd be good to be uploaded during the soft freeze, and need suggestions here. Note that it also uses a "Built-Using" on simde due to it's license being GPL3+

Nilesh

--

Michael R. Crusoe
Co-founder & Lead,
Common Workflow Language project
https://orcid.org/0000-0002-2961-9670
mrc@commonwl.org