add libsemigroups package

[dimpase]

​libsemigroups is a C++ library for computing in semigroups, and is used by semigroups GAP package, which is to be included in Sage, cf. #27295

So this automatically gives it rights to become an optional package.

tarball: ​http://users.ox.ac.uk/~coml0531/sage/libsemigroups-0.6.7.tar.gz

To the release manager, merge simultaneous with #27295.

[dimpase]

Ready for review. On another ticket we should also add its Python bindings: ​https://github.com/james-d-mitchell/libsemigroups-python-bindings/

---

New commits:

​15bf22f

add libsemigroups package

[tscrim]

It installed and package tests were okay for me (Ubuntu 18.04 LTS). However, I think this is mostly useless package for Sage without the Python bindings (or GAP's semigroups package). So I think we should add the Python bindings as part of this ticket too.

[dimpase]

Replying to tscrim:

It installed and package tests were okay for me (Ubuntu 18.04 LTS). However, I think this is mostly useless package for Sage without the Python bindings (or GAP's semigroups package).

I don't quite understand the point you are trying to make here. This ticket is a dependency of #27295, which is for exactly this, adding GAP's semigroups package.

So I think we should add the Python bindings as part of this ticket too.

No, not here. Adding python bindings should be on another ticket.

[tscrim]

Replying to dimpase:

Replying to tscrim:

It installed and package tests were okay for me (Ubuntu 18.04 LTS). However, I think this is mostly useless package for Sage without the Python bindings (or GAP's semigroups package).

I don't quite understand the point you are trying to make here. This ticket is a dependency of #27295, which is for exactly this, adding GAP's semigroups package.

This does not make sense to me as a standalone (at least, it seems like the Cython code would be complicated to generated examples, but maybe I didn't look closely enough). Based on that, I feel either this should be as part of #27295 or have the Python bindings added to this ticket.

[dimpase]

Now it is a part of #27295.

[dimpase]

A ticket should not be trying to do too much, and get too hard to review due to this.

[tscrim]

I agree, but I also think tickets also should have some standalone meaning. Different commits can help granularize changes similar to splitting tickets. I just don't think this quite passes the standalone test.

[dimpase]

We have added standalone C++ libraries without any bindings and a clear plan why they are needed in the past, cf e.g. mprfcx. This one is much better in this respect.

[dimpase]

Different projects have different commits per ticket policies. We don't have any, and I seldom see people reviewing tickets per commit, either.

[tscrim]

I am not one to generally follow something I do not agree with just because there is prescient. I am only setting a positive review here under the condition that this is merged with #27295, but I feel it would be better for this to be included as part of #27295.

[embray]

Ticket retargeted after milestone closed (if you don't believe this ticket is appropriate for the Sage 8.8 release please retarget manually)

[embray]

Updated link in the description to the current source code home of libsemigroups.

[embray]

Oops, pasted the wrong link.

[git]

Branch pushed to git repo; I updated commit sha1 and set ticket back to needs_review. This was a forced push. New commits:

​32aab07	update to GAP 4.10.2
​e1d5ab1	add libsemigroups package

[dimpase]

just a version bump and a rebase over the latest GAP

[dimpase]

this needs a modified tarball, as the ones available do not come with ./confugure, one needs autotools to get one.

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​300736d

correct checksum for the tarball

[dimpase]

I've linked the correct tarball, with ready ./configure. One needs to untar the original upstream tarball, which is bare, run ./autogen there (having autotools installed), and re-tar.

[dimpase]

Replying to isuruf: Could you also review #27295 (this is adding GAP interface to this library, whjch was the original purpose of this ticket all along)