The next devlog was initially supposed to be the one for the first alpha version, but as foreseeable our plans didn't quite work out.
Additionally, we have made a hard decision.
After two years of continuous annoyance and breakage, we have decided to get the fuck rid off pandas and as a result continued development on the data editor has become much less pain- and dreadful.
We're going to build our own analysis tooling – with blackjack and hookers!
And since our [de]serialization was already set up for numpy and shapely, we didn't even have to change anything there – things just got simpler because we don't have to take care of pandas' sometimes-but-not-always foregoing numpy datatypes for things like datetimes. We can have our consistency cake and eat it, too!
Implementing the functionality that pandas was already used for was almost trivial, with additional features (namely tolerances for join conditions and filtering for non-numeric data) on the way.
You can watch the progress on that in the
trashpandas branch, which we
will merge with
master as soon as the functionality currently being
worked on is done.
Once that's done, we are back on track for the alpha release – the rest should be a good bunch of hard work, but nothing requiring any new engineering.
Lastly, we've already started collecting issues, for the second alpha release, so if you're interested in seeing what the future has in store, feel free to have a look.
Well, that's it for now, back to the mines…