Jon Brandvein

I guess instead of viewing this as privileging the main target over other macro targets, it's just a reflection of the fact that the UI the user sees is "I...

Definitely, the rhs should be fully evaluated before any assignments take place. We probably also have this problem elsewhere in the spec. For instance, we specify left-to-right evaluation for dictionary...

My first thought was that we shouldn't need to specify this at all, it should be assumed. But then again, is formatting specified to be implementation-defined in any way? If...

Sure, it's very easy to accidentally have this sensitivity to locale. The question is whether the spec bans this behavior by fully specifying formatting, or whether the spec permits an...

Created bazelbuild/bazel#25628. For the spec though, do we want to specify exact formatting conversions or leave it implementation defined? Or to put it another way, is a compliant implementation obligated...

> it's more important that the form of the output is predictable than that each number is accurate to 1 ulp If it's not 100% specified then program behavior may...

While I agree with everything Alan said two comments above, I can't see us prioritizing proper documentation for all the quirks of implicit outputs anytime soon. We know we'd like...

Hi, we're using Laurent's awesome-starlark list in lieu of maintaining a list of usages in this repository now. (I see you've already been added there.) Sorry to keep this PR...

Obsoleted/fixed by #289.

The design was rejected in its current form. Leaving this issue open to represent the unresolved question of whether/how this might be addressed in the future. The two possibilities are...