Le 22/06/2018 à 11:29, Frank Mittelbach a écrit :
> Am 22.06.18 um 10:59 schrieb Joseph Wright:
>> Ultimately, however, the
>> prefix is down to the package author: it's no good saying 'no we won't
>> log it', only for the package to go to CTAN anyway but ending up that
>> bit harder for other developers to see.
> As I said I would veto something like "stats", as for "statistics" it is
> long and it might be the package name so it is probably ok.
Some background info: the « rivmath » package started as a collection of
small mathematical settings of mine (like loading amsmath, unicode-math
and similar), but slowly grew over time to have loadable libraries. The
problem is that my typographical preferences are loaded by rivmath. I
had the choice between making rivmath a kind of all-math-related bundle
(a bit like the tkz suite from AlterMundus) or spin-off the parts that I
deem robust and useful enough to their own package.
This new « statistics » package is exactly that. I could call it
rivstats if you prefer. To choose the name I searched for statistics
related packages on CTAN and didn't find much, thus figured that using
the term wasn't so bad.
I'll submit it when I have written enough documentation, but the primer
is « automatic computation and typesetting of (currently 1-variable
descriptive) statistics tables and graphs, as can be taught in high
school, from the raw data, (value, count) pairs or (interval, count)
For instance, every table and graph in the following PDF has been
generated by the package which even supports keeping parts hidden for
teaching or demonstration purposes:
I don't think it qualifies as « everything you'll ever need about
statistics » as siunitx does for, well, SI Units typesetting.