Re: New version of muscle with changed options compared to previous versions [muscle_5.1-1_source.changes ACCEPTED into unstable]

[Andrius Merkys <merkys@debian.org>]

Hi,

On 2022-01-17 16:55, Andreas Tille wrote:
> I've just realised there is a new version of muscle (now on Github)
> which has not only a changed algorithm but also changed command line
> options (I had to adapt autopkgtest which is also a bit weak - some
> educated person should check the result and drop a proper md5sum for
> the expected result).
> 
> I think we should follow upstream development but given that this quite
> wide spread tool is used in several other packages (like biopython or
> bioperl) I expect failing tests of those in the near future.  This opens
> the question whether we should possibly maintain a muscle3 package - but
> than we also need to patch those tools to spot the changed name of the
> binary.
> 
> Any opinions of real users of this package are welcome.

I consider myself a muscle user, albeit mostly for teaching. Integration
with biopython is quite important aspect to me as well.

Although stability is a desired quality, algorithm improvement is
probably more important. However, I did not compare the new muscle with
its predecessors thus I cannot say which change outweighs which.

I do not think that maintaining two parallel muscle packages is really
worth the work. Having muscle 5.1 and biopython supporting only muscle 3
will cause confusion. As for failing tests of other packages, it is
probably best to gently prod their upstreams to implement support for
new muscle.

But all of this is just my opinion.

Best,
Andrius