Shorter printing of core linear code objects

[johanrosenkilde]

The central objects for linear codes in Sage prints unnecessarily long, even though there are established conventions for shorter description. This ticket proposes a change to that:

Linear code of length 7, dimension 3 over Finite Field of size 2^3
   should be
[7, 3] linear code over GF(8)

Printing GF(8) is the only controversial thing here, from a Sage POV. The
argument is that linear codes are always over finite fields, so it is
unnecessarily verbose to insist on the long printing. (formally, using LinearCode over a ring has not been deprecated yet, see #20387, but a code over a ring is not a "linear code" in the classical sense)

Other proposed changes are:

• Generalized Reed-Solomon Code -> Reed-Solomon Code (whenever the column multipliers are all-1)
• Extended code coming from <bla> -> Extension of <bla>
• Punctured code coming from <bla> -> Puncturing of <bla>
• Subfield subcode over <subfield> coming from <code> -> Subfield subcode of <bla> down to <subfield>

Related objects (e.g. encoders and decoders) have been shortened similarly
(though much more could be done here).

An unrelated change which I put here anyway is a polishing of AbstractLinearCode's class doc, and improved error messages for its encoder() and decoder() methods.

CC: @sagetrac-dlucas @jlavauzelle

Component: coding theory

Keywords: linear_code

Author: Johan Rosenkilde

Branch/Commit: 2316e35

Reviewer: Julien Lavauzelle

Issue created by migration from https://trac.sagemath.org/ticket/21691

[johanrosenkilde]

Branch: u/jsrn/improve_linear_code_messages

[johanrosenkilde]

New commits:

77772c3	Better error message for encode and decoder
d2f4c18	Improve some class doc in AbstractLinearCode
dafdadf	Shortened printing of many linear code objects.

[johanrosenkilde]

Commit: dafdadf

[jlavauzelle]

Changed branch from u/jsrn/improve_linear_code_messages to u/jlavauzelle/improve_linear_code_messages

[jlavauzelle]

comment:4

Hi Johan,

Nice changes, it makes it really more readable. I only have one suggestion for punctured codes. You wrote for instance: Puncturing of [11, 5] linear code over GF(7) punctured (...) but I think writing "Puncturing/punctured" only once would be even easier to read. However I'm not that fluent in english, so maybe I'm wrong...

I also made 3 optional doctests pass in guava.py and merge with the latest beta (7.5.beta1).

Julien

---

New commits:

4d65268	Merge branch 'u/jsrn/improve_linear_code_messages' of git://trac.sagemath.org/sage into 21691_shorter_printing
a3b2b66	Merged with latest beta (7.5.beta1).
ba19b0c	Fixed doctest on optional guava package.

[jlavauzelle]

Reviewer: Julien Lavauzelle

[jlavauzelle]

Changed commit from dafdadf to ba19b0c

[johanrosenkilde]

comment:5

Hi Julien,
Thanks for the review and good catch with the optional doctests. I agree with you on the puncturing. What about Puncturing of [11, 5] linear code over GF(7) at positions [1,2]?

I would make the change, but I'm running some heavy computations and I'm worried that I'll break something by recompiling and experimenting. Could you do it?

Best,
Johan

[sagetrac-git]

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

2316e35

Improved punctured codes representation and fixed doctests.

[sagetrac-git]

Changed commit from ba19b0c to 2316e35

[jlavauzelle]

comment:7

Hi Johan,

I did that change. Patchbot also found doctest failures in some crazy tutorials (one about linear codes and ciphers, the other about graphs...). I fixed them.

I'll wait your green light on my changes to give the positive review.

Best,

Julien

[johanrosenkilde]

comment:8

It looks great, thanks a lot!
Johan

[vbraun]

Changed branch from u/jlavauzelle/improve_linear_code_messages to 2316e35