4 ms·

> but complaints would have bubbled up to the Apache umbrella organization, which no doubt has plenty of Words written about "proper" versions, and the letter o

> but complaints would have bubbled up to the Apache umbrella organization, which no doubt has plenty of Words written about "proper" versions, and the letter of rules would have been used to add heavily to their burden, while the spirit of compatibility is ignored. Tailscale is wonderful software, but with all due respect I don't think David has much experience with Apache Foundation or Eclipse Foundation projects. I am a project lead of one (way smaller) Eclipse Foundation project and keep a close watch on another (bigger, but nowhere near the log4j scale) Apache Foundation project. We are free to make breaking changes (and make them we do). In both Foundations, there is a Project Management Committee that signs off on releases. They would normally raise an eyebrow if you try to push a breaking change into a patch release, but their reaction would normally be "why don't you bump the major version?". Edit: there is indeed a social contract between the project leadership and the users. We have a social contract of supporting the oldest possible JVM for our project. Now some of our dependencies (guess what, that ASF project!) dropped Java 8 support and published a few CVEs since and we are going to move to JDK 11 as a baseline soon. Yes, some users are grumpy but people understand that unless someone forks that ASF project and starts backporting CVE fixes, we got to make the move. Bottom line: it's a much more social process than a bureaucratic one.

optimalsolver 5y ago
>but complaints would have bubbled up to the Apache umbrella organization Like a thrown exception.
crawshaw 5y ago
Hello. You're right that I have little experience with them, I've submitted code but never run a project. I believe I misread some commentary to convince myself of this. Paragraph removed. Thanks! I suspect there is some kind of cultural pressure somewhere. Perhaps some of that comes from the versioning system, the maintainers are convinced they can't remove a feature without bumping a major version, and are simultaneously convinced they need to do a lot of work for others they don't have time to to bump the major version number? I believe neither statement should be true in reasonably run open source.

optimalsolver 5y ago

>but complaints would have bubbled up to the Apache umbrella organization Like a thrown exception.

crawshaw 5y ago

Hello. You're right that I have little experience with them, I've submitted code but never run a project. I believe I misread some commentary to convince myself of this. Paragraph removed. Thanks! I suspect there is some kind of cultural pressure somewhere. Perhaps some of that comes from the versioning system, the maintainers are convinced they can't remove a feature without bumping a major version, and are simultaneously convinced they need to do a lot of work for others they don't have time to to bump the major version number? I believe neither statement should be true in reasonably run open source.