Mathieu Céraline: “On GitHub, commits usually look like e83c5163316f89bfbde7d9a…”

Mathieu Céraline @mathcrlnOct 19885b898

I notice the commits are regular uuids on /dev/null. To go even further with the git metaphor, what if we considered transitioning to content-addressed SHA-1 hashes like git does? Could also be a cool/clever way to approach editing commits etc if the feature ships later. What…

Kjell Kind @kachelleOct 19aa047b4

can you explain this content-addressed SHA-1 hashes?

Mathieu Céraline @mathcrlnOct 19

On GitHub, commits usually look like e83c5163316f89bfbde7d9ab23ca2e25604af290. Unlike uuids on here (https://dev-null.sh/@kachelle/commits/01a1563d-fe17-70ec-af96-4d137aa047b4) that are arbitrary, git essentially creates the id if your commit by hashing your data. Basically a very unique id specific to the changes you made. That’s why it’s unique and generally changes when you rewrite commits, rebase; etc. Definitely not an expert, just thought it was a pretty cool idea :)

/dev/nullKjell Kind (@kachelle) replying to @mathcrln on /dev/null: “can you explain this content-addressed SHA-1 hashes?”can you explain this content-addressed SHA-1 hashes?

22:23 · Oct 19, 2026commit 8b4c4b9

Stephan Meijer @smeijerOct 19

Downside would be that uuidv7 is time sorted and sha hashes are not. So that'd make the primary key inefficient for indexing.

Commit