More permissive license?

[grantmcdermott]

Hi Bob.

I saw a link to Ibex on my feed. Cool project!

I was wondering how (and why) you settled on licensing through AGPL-3.0?

For better or for worse, AGPL is heavily guarded against in tech and other commercial sectors. You might be able to use an AGPL licensed piece of software for an internal project (normally, only after escalating through legal, which can be a real pain in of itself). But it’s almost assuredly a no-go as an input to any production-level, consumer-facing product.

All of the alternative query engines that you are evaluating against (polars, DuckDB, data.table, etc.) run through permissive licenses (MIT, Apache, etc.) that don’t trigger these flags.

Your project, your choice obviously. But my feeling is that sticking to AGPL will foreclose a lot of your potential userbase.

[bobjansen]

Hi Grant,

Thank you!

And thanks for raising this. I’m actually not fully decided on the license yet, and the adoption cost of AGPL is one of the main reasons I’m still considering it.

Some considerations I have:

• I'm the only contributor at the moment and it would be easier to move to Apache from AGPL than vice versa.
• If Ibex becomes successful, I would prefer that someone cannot take it, build a closed fork or hosted derivative around it, and prevent users of that derivative from getting access to the corresponding source. For a pure analytical engine this is probably less of a concern than for a database, though.
• I fully agree that it creates friction for adoption that is not desired. I do hope to get lots of users.
• In the short term and in the happy scenario that somebody contacts me and wants to use Ibex I'd be happy to accommodate them somehow. I agree that would probably be impractical and therefore not likely to happen.

So I don’t regard AGPL as a final decision. It is the safer (for me) starting point while the project is young, with Apache-2.0 very much an option.