Make approximate dataset size more readily available

[colleenjg]

It occurs to me that users might benefit from seeing an estimate of a dataset's size when they launch a command to download it. Estimated size could, for example, be printed as a log.

An alternative would be to expand the brainsets list CLI command so that it can also list a few key attributes for each onboarded dataset (e.g., year, modality, nbr sessions, nbr subjects, approx size). This could be useful for users looking to start exploring torch_brain using a dataset that can be quickly downloaded and uses fewer compute resources. (An eventual sort option could be added.)

[milosobral]

I like the idea of adding it to brainsets list. Printing as a log could also be useful but it feels like the kind of information that is more useful to decide what to download rather than as a warning when you've already launched the command.

[colleenjg]

Yeah, I agree. Ideally users should be able to check the size before they start the download, as opposed to having to ctrl-c the download once it's started if it's going to be too big. I do think it can still be useful to have it as a log, though, to set expectations as to how long the full download will take.

[vinamarora8]

I like the idea of adding it to brainsets list, and perhaps also in the docs?

Another idea would be to add a sub-command brainsets info <brainset-id>.
This could include the download/prepared data size and even some more info, such as listing out the CLI args supported by its brainset-pipeline. We currently have no way of telling users what CLI args are supported, and they have to look at the code to actually find them. Though this might take slightly more architectural changes to achieve.

[colleenjg]

Great ideas. I'll put this on my to do list!

[colleenjg]

As discussed in our meeting, here are some functionalities we want to add:

• brainsets list: add modality and size and potentially enable sorting by modality or download size (default by name?)
• brainsets brainset_name info: list some information (download size, processed size, available commands, link to docs, and maybe a brief description?)

Eventually, it would be great to automatically generate a table with all the brainsets and key info in our docs.