Replies (1)

• @david_chisnall@infosec.exchange 2026-09-15 11:20

  @SuperDicq@minidisc.tokyo GPLv3 would have had zero impact on the conversation. Big companies normally avoid using or contributing to GPLv3 projects entirely because it's even less well understood than GPLv2 and comes with exciting patent clauses that can translate to enormous liabilities. Small companies that hold patents also avoid it because they can accidentally lose their paper defence against big companies. This is always a risk-reward discussion. The risks of the license are larger for v3 than v2, but the license remains badly written (though less so) and easy for a competent legal team to find loopholes in if they really want to. The rewards are much lower. Linux is of interest to big cloud providers because it already had an active ecosystem and community when they started wanting to offer hosting. Without that, it wouldn't matter if the license were some perfect platonic ideal of a Free Software license, the reward would simply not be there and the project would be ignored. Switching to GPLv3 for GCC meant that Apple and Google shifted their contributions over to LLVM. Microsoft still does some work on GCC because it's used by non-Android Linux distributions, but they carefully firewall that work internally. Google and Apple simply decided it was too high risk relative to the rewards and moved their investment elsewhere. The Linux Foundation's dashboard thing reports: GCC has 188 active contributors, LLVM has 1750. Using the contribution accounting from the last year: The top three organisations that contribute to LLVM are AMD, Google, and Apple. Between them, their contributions add up to 32%. The next largest org on the list is 'LLVM', which is actually an umbrella for people with no inferable attribution, at 7%. NVIDIA has 6%, no other company breaks the 5% mark. On the GCC side, RedHat (an IBM subsidiary) alone is responsible for 33% of contributions. IBM, Qualcomm, and SUSE, between them, account for 51% of contributions. So, relative to a permissively licensed direct alternative, the outcome of the switch to GPLv3 for GCC has been: Fewer contributors overall.More concentration of contribution (more than half from three companies) If your goal is to give corporations more centralised control over a project, GPLv3 is great. And that's why it's important to measure these licenses against their actual effect and not their aspirations.

  Open ##4710881