feat: bead input sync

[korenmiklos]

We often update .bead-meta from outside bead, like under version control. We need a way to sync the actual inputs with the meta. It is hard to manually unload/load data. Even bead input unload would not work if an input has been deleted from meta.

bead input sync should

• check each input listed in meta, if content is the same (do we check checksum? there is no meta in input/ folders), do not touch
• if input folder is not listed in meta, delete
• if input is missing from the folder, do a bead input load

[korenmiklos]

The analogy is a package manager: the meta can be versioned and poetry install etc can bring the packages in sync with meta

[e3krisztian]

I am a bit confused with this issue.
Bead is intended to be a unit of work with small[ish] scripts in the range of a few hundred lines of code at most.
Bead is itself a version control for computations, it can save versioned backups, including work-in-progress versions, but it can not do diffs (you can have a different version in another directory and use directory diff), and it can not do merges (one bead = one developer, who to merge from?).

I have the feeling something is used in a way it was not intended.
Why put the directory under version control?
Even if it in version control, why pull/revert from a remote git repo?
I am curious: what is the use-case?

If it is the code to be managed by git, maybe the users should add a .gitignore at the root of the bead:

/.bead-meta
/input
/output
/temp

so as not to get bead's stuff (meta included) versioned?

[korenmiklos]

This follows from a deeper issue, let's discuss it there

https://github.com/e3krisztian/bead/issues/48