With version 2.0 of Unidatab, Unidatab-CGI had lost its ability to modify a database. The problem was coming from the strengthened access restrictions server-side and not easy to tackle at once, involving changes at various levels.
Additionally, Unidatab-CGI (https://github.com/emilbarton/Unidatab-CGI) now offers a very basic style sheet in need for extension too; not real handy yet, but the step towards better HTML is taken.
These new versions (Unidatab 2.1.0, Unidatab-CGI 1.1.0) improve considerably the overall software situation, but probably several bugs remain to track and solve in cgi context (and elsewhere). This will be done gradually while in parallel I keep working on structural improvements, specifically output classes making the code faster and easier to read.
Another useful task I’m about to take over is the update of old posts on this blog so that it can serve as a tutorial for the present state of the code. The changed pages will be marked by the string [2018 edit] in order to let users know quickly that tips and tricks are up to date.
On the theoretical level I have started a (local) fork of Unidatab in order to test various improvements impying mathematical clarifications supposed to bring software enhancements in the medium term. In this perspective I could post a set-theoretic question on the Quora forum but am still waiting for an answer. Was the problem correctly stated? Probably not, but my ignorance is perhaps not the only thing to blame. There’s a general issue about software development: how do we convene scientific knowledge when confronted with technical challenges? Had I formulated my question in easier data science terms, I’d have probably received several answers already but missing the mathematical abstraction I wanted to reach. The goal wasn’t for me to speak of Unidatab on this occasion, but rather to discover how a similar subject matter would be handled in mathematical terms. This question could cover various mathematical domains: algebra, set theory but maybe newer fields too, like type theory or even homotopy type theory:
How would you describe the relations between the finite sets A, B, C? A is a set of 1-tuples. C is a set of 2-tuples where each element of a pair can be taken either from A or B but not C; B is a set of n-tuples of elements of C, where n varies.
I hope the present update in CGI will at least help the people who would be interested by the concept behind Unidatab but are repelled by code typing, to experience with this software more easily.
[2018 edit: done]