[issue41746] Add optional type information to asdl_seq objects

[Raymond Hettinger]

Raymond Hettinger <raymond.hettinger at gmail.com> added the comment:

> I know it's moot now, but still -- what benefit do we get 
> from using a "standard" like ASDL?

The "standard" part of it isn't important.  AFAICT, ASDL has a low adoption rate and is not maintained.

IMO, the part that matters is that ADSL was carefully balanced to be sufficiently expressive while keeping it easy to implement and easy to automatically translate into different languages.  Presumably, this will not only help other Python implementations and third-party tooling, it will also make life easier for us in the long run.

My understanding of the origin of ASDL is that it aspired to solve a common problem in language design where people commonly described their abstract syntax in way that was too tightly bound to underlying implementation language.  This caused long-run problems when reimplementing in other languages and when trying to automatically update downstream tools that interoperate with the AST.

In this regard, my personal experience with ASDL has been favorable.  I view it as the JSON spec of the AST world, intentionally minimal yet expressive.

That said, I think it failed to establish itself as a standard and almost no tooling was created for it.  The original authors expected that ASDL would sit side-by-side with BNF and regex notation.  That was a pipe dream.

----------

_______________________________________
Python tracker <report at bugs.python.org>
<https://bugs.python.org/issue41746>
_______________________________________