Fork me on GitHub
#cljs-dev
<
2017-07-17
>
dnolen06:07:51

:npm-deps finishing touches I’d like to see https://dev.clojure.org/jira/browse/CLJS-2255

ambrosebs17:07:04

@dnolen does it make sense to propose expanding dotted symbols in the analyzer to :dot nodes?

ambrosebs17:07:04

It seems it would at least be useful to expand locals(?). I don't really understand how global/js dotted variables work.

tony.kay20:07:00

Have optimizations been well-tested with the new module support?

dnolen23:07:37

@ambrosebs I just need more information on what you’re trying to accomplish?

dnolen23:07:28

@juhoteperi I made bunch of minor edits to the post - if we can get those in would like publish tomorrow if possible

ambrosebs23:07:28

@dnolen as far as I understand, dotted symbols are shorthand for field accesses in CLJS. My hunch was that consumers of the AST format (like core.typed) would rather process an AST with :dot nodes containing a :var node, than just a :var node with an implicit field access. I'm just not familiar enough with CLJS's emission to JS to understand when/if this might be useful for tools.

dnolen23:07:48

@ambrosebs hrm will think about this and get back to you - if you want to open a enhancement ticket to track this, go for it

ambrosebs23:07:14

@dnolen thanks. Would you consider the tools.analyzer enhancements in smaller patches? I have a unit of work that I'm making into a nice patch to review, but it's not "done". I don't want it to linger for another year.

ambrosebs23:07:30

the first unit of work might just change :op's and :children to be more consistent, plus cosmetic renames of keys (:expr -> :body) plus the internal fixes needed to the compiler.

ambrosebs23:07:56

and larger questions like dotted symbols can be discussed later.

ambrosebs23:07:34

I created a ticket for the dotted var discussion https://dev.clojure.org/jira/browse/CLJS-2257