toddjonker
commented
4 months ago So after spending most of Friday trying to figure out a workable void-placeholder to go with separators, my weekend’s musing has left me much less enthusiastic about separators on the whole. I think I didn’t really consider the grammatical complexity, which effectively forces a brand-new third kind of sequence (beside list and sexp) with its own unique separation and quoting rules for both implementers and humans to keep track of. I mean it looks and reads nicely, but that’s hiding a bunch of cruft that worries me.
Ultimately, if we are comparing separators versus implicit-rest for all */+ I’m back to preferring the latter. Especially considering that it’s so close to what we’ve had all along, whereas separators are a big change both conceptually and syntactically.
Since @zslayton indicated that he's alright with sticking to the implicit-rest plan, I'm abandoning this in favor of #316
DO NOT MERGE! For illustration only.
I updated most (if not all) of the examples in the spec to illustrate how separators work as a replacement for argument groups. I've used semicolons for both E-expressions and TL invocations, since this process convinced me that the inconsistent use of comma was too confusing.
One thing I'm still considering is whether we should use separators after
!and?single-expression parameters. They aren't really needed there if we have a way to write "no value" for the latter case, and omitting them could avoid some verbiage. I'm not convinced, just considering. :-)By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.