Sébastien Labbé

Since two years I wrote thousands of line of private code for my own research. Each module having between 500 and 2000 lines of code. The code which is the more clean corresponds to code written in conjunction with research articles. People who know me know that I systematically put docstrings and doctests in my code to facilitate reuse of the code by myself, but also in the idea of sharing it and eventually making it public.

I did not made that code into Sage because it was not mature enough. Also, when I tried to make a complete module go into Sage (see #13069 and #13346), then the monstrous never evolving #12224 became a dependency of the first and the second was unofficially reviewed asking me to split it into smaller chunks to make the review process easier. I never did it because I spent already too much time on it (making a module 100% doctested takes time). Also, the module was corresponding to a published article and I wanted to leave it just like that.

Getting new modules into Sage is hard

In general, the introduction of a complete new module into Sage is hard especially for beginners. Here are two examples I feel responsible for: #10519 is 4 years old and counting, the author has a new work and responsabilities; in #12996, the author was decouraged by the amount of work given by the reviewers. There is a lot of things a beginner has to consider to obtain a positive review. And even for a more advanced developper, other difficulties arise. Indeed, a module introduces a lot of new functions and it may also introduce a lot of new bugs... and Sage developpers are sometimes reluctant to give it a positive review. And if it finally gets a positive review, it is not available easily to normal users of Sage until the next release of Sage.

Releasing my own Sage package

Still I felt the need around me to make my code public. But how? There are people (a few of course but I know there are) who are interested in reproducing computations and images done in my articles. This is why I came to the idea of releasing my own Sage package containing my public research code. This way both developpers and colleagues that are user of Sage but not developpers will be able to install and use my code. This will make people more aware if there is something useful in a module for them. And if one day, somebody tells me: "this should be in Sage", then I will be able to say : "I agree! Do you want to review it?".

Old style Sage package vs New sytle git Sage package

Then I had to chose between the old and the new style for Sage packages. I did not like the new style, because

• I wanted the history of my package to be independant of the history of Sage,
• I wanted it to be as easy to install as sage -i slabbe,
• I wanted it to work on any recent enough version of Sage,
• I wanted to be able to release a new version, give it to a colleague who could install it right away without changing its own Sage (i.e., updating the checksums).

Therefore, I choose the old style. I based my work on other optional Sage packages, namely the SageManifolds spkg and the ore_algebra spkg.

Content of the initial version

The initial version of the slabbe Sage package has modules concerning four topics: Digital geometry, Combinatorics on words, Combinatorics and Python class inheritance.

For installation or for release notes of the initial version of the spkg, consult the slabbe spkg section of the Sage page of this website.