Add binding breaking change to 1.10.439 Release Notes

[arichiardi]

Opening this PR in order to warn in the Release Notes that the behavior of bindings has changed in a breaking way.

One affected library is humane-test-output.

[mfikes]

This generally LGTM, but perhaps an example can be given that actually produces equivalent behavior to Clojure. One is given in the description of https://dev.clojure.org/jira/browse/CLJS-2541

[arichiardi]

While I see your point, I wanted to show the breakage more than the correct behavior, assuming folks would know (or google it). I also hoped the description and (doc binding) are also available to them.

However we could add both maybe?

[mfikes]

I was suggesting showing the breakage by simply showing code that previously produced a result that differed from Clojure, while showing that the new behavior matches Clojure (the correct behavior).

In other words, the breakage is the difference in behavior for such code.

To put it another way, with the example in CLJS-2541, the breaking change is that code that previously evaluated to 11 now evaluates to 2. This difference from 11 to 2 is sufficient to constitute breakage.

On the other hand, while the example you are showing does indeed also show that there is a breaking change, Clojure fails in a different way with that code because of the difference in handling unbound Vars in each dialect. So, while it indeed shows breakage, but it is doing so in a way that doesn't illustrate the correct behavior.

[arichiardi]

Uhm I am not sure I understand what you want me to change.

I wrote an example that shows what you should be careful of and what was the previous serial behavior. Isn't that what a dev, like me yesterday, wishes to see for a breaking change?

In any case I can copy your example from the Jira as well if you feel strongly about this.

[mfikes]

I think an example of a breaking change that also shows 1.10.439 matching Clojure would be an improvement, but it isn’t necessary in my opinion.

[swannodette]

Sorry this is not worth documenting IMO.

[arichiardi]

@swannodette respectfully, it would be great to explain a bit why you think that so that next time Mike and myself avoid spending time on something "not worth documenting": we would all learn something at least 😁

Thanks!

[swannodette]

@arichiardi it my view it was a bug fix, not a breaking change and just not interested in more discussion about that. Also this was opened up almost 2 years ago I don't see how it's relevant anymore given that many users do update with regular frequency.

[arichiardi]

Ok that's fair, thanks for explaining 👍