Larry Garfield

I don't like it either, but the language allows for it. A coding standard of "never use this language feature" seems... silly. What I could see is maybe saying that...

Sure, but we don't forbid compound namespaces entirely. That would be the equivalent here. I'm not comfortable just blanket disallowing a language feature, even if I never use it personally....

I tried tweaking the wording a little. Note that this still says MAY, only. So someone always and forever putting every attribute in its own block would be fully compliant.

Also rebased, so the section numbers all moved around.

Yeah, we should include a constructor, good call. For the blank lines, that's not in there now and I've heard differing opinions in the wild on it. I'm in favor...

But at that point wouldn't it be easier and just as effective to use separate blocks? I don't know how to say that really long ones are a bad idea,...

We can keep discussing in another thread, but I want to reiterate that "the language explicitly supports syntax X, but we're forbidding you from using it" is not a position...

So something like this?

We cannot reasonably pretend that they don't exist. I'd rather acknowledge their existence and put reasonable bounds on them (as here) and punt on the "when you should use them"...

Can we get this resolved? I don't really see much alternative to what is listed here now. Ignoring the existence of compound attributes is not viable, and forbidding the use...