Blume

It started with just a class to draw tables alongside matplotlib plots. Years ago I wrote some matplotlib python to do the job of displaying a grid of text boxes. How to create a general interface that allows for things such as nested row and column labels?

So this got me thinking about tables.

Rows and columns of data.

Sometimes with column or row labels.

Tables: rows of data, each row a number of columns.

How to explore and share the data?

An aside that two types of tables are in mind: csv files of data and image files. An image is just a rectangular grid of values, so it fits the model.

Sometimes (usually?) I want to use both. An image in the background, overlaid with other data.

Then there is this idea of content addressable filesystems, and how this relates to git.

For starters, if blume is running in a git repository it would be good to let you navigate the git history of a file. I think I know enough now to do that reasonably efficiently without doing any changes to the files on the disk.

My inclination here is to let git do as much of the work that needs to be done.

Git is really good at lots of things.

I have also been thinking about bots in relation to blume and exploring data.

I am imagining some sort of bot that randomly plays with the attributes of things running in the system.

Perhaps manipulate meta data and save it in a git branch, in case it turns out to be useful later.