Codecov Report

Merging #17715 into master will increase coverage by 1.08%.
The diff coverage is n/a.

@@             Coverage Diff             @@
##           master   #17715       +/-   ##
===========================================
+ Coverage   82.11%   83.20%    +1.08%     
===========================================
  Files        1956     1571      -385     
  Lines      589319   414699   -174620     
  Branches    44450    44450               
===========================================
- Hits       483924   345046   -138878     
+ Misses      95746    60009    -35737     
+ Partials     9649     9644        -5     

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 9aeb0ef...9feb855. Read the comment docs.

Benchmarks (asked by @pablogsal) (not precise because lack of isolated cpus) (debug build / no pgo or lto):

 $ ./python -m pyperf timeit -s "import ast; f = open('Lib/ast.py'); c = f.read(); f.close(); t = ast.parse(c)" "compile(t, '<test>', 'exec')"
master: Mean +- std dev: 15.0 ms +- 0.4 ms
PR    : Mean +- std dev: 15.5 ms +- 0.5 ms

 $ ./python -m pyperf timeit -s "import ast; f = open('Lib/test/test_socket.py'); c = f.read(); f.close(); t = ast.parse(c)" "compile(t, '<test>', 'exec')"
master: Mean +- std dev: 75.4 ms +- 4.0 ms
PR    : Mean +- std dev: 77.2 ms +- 7.4 ms

Hey @asottile, could you give us a hand here? What is your oppinion on the error messages? Do you find odd/weird that they mention ASDL names? Do you think they may be a bit misleading for users that are not familiar with the ASDL definition? For instance:

field \"%s\" was expecting excepthandler type node, got ...

Do you have any suggestions on how to improve them?

agree with the expecting node of type %s as an improvement on the english standpoint

as for the actual messaging, this will only be exposed to those which are constructing ast manually right (for instance, pytest for assertion rewriting / werkzeug for "JIT")? as long as these aren't exposed to most users I think the quality of error message that's here is fine (though would be better if we could talk about ast types here but understandable that they're not available at this level).

at least for the expr ones I think it'll be fine (ast.expr is a thing after all)

Thanks for help Paul and Anthony

(though would be better if we could talk about ast types here but understandable that they're not available at this level).

What do you mean by ast types? Aren't all of these are real AST types (like expr, or expr_context etc.)

What do you mean by ast types?

I think he is referring to the python classes.

What do you mean by ast types?

I think he is referring to the python classes.

yeah I mean the classes in the ast module, sorry I should have been more specific :)

yeah I mean the classes in the ast module, sorry I should have been more specific :)

@asottile do you have any kind of error message in your mind (a specific one, let's say for this case)

>>> import ast
>>> t = ast.parse("node", mode="eval")
>>> t.body = [t.body]
>>> compile(t, "<test>", "eval")
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: required field "lineno" missing from expr

Maybe we could improve it

(update) the current version:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: field "body" was expecting expr type node, got list

yeah I mean the classes in the ast module, sorry I should have been more specific :)

@asottile do you have any kind of error message in your mind (a specific one, let's say for this case)

>>> import ast
>>> t = ast.parse("node", mode="eval")
>>> t.body = [t.body]
>>> compile(t, "<test>", "eval")
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: required field "lineno" missing from expr

Maybe we could improve it

(update) the current version:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: field "body" was expecting expr type node, got list

yeah the new code is definitely a huge improvement either way 👍

After applying suggestions, current state of the PR:

>>> import ast
>>> t = ast.parse("x", mode="eval")
>>> t.body = [t.body]
>>> compile(t, "<test>", "eval")
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: field 'body' was expecting node of type 'expr', got 'list'
>>> t = ast.parse("'str'", mode="eval")
>>> t.body.kind = 0xFF
>>> compile(t, "<test>", "eval")
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: field 'kind' was expecting a string or bytes object

When you're done making the requested changes, leave the comment: I have made the requested changes; please review again.

The following commit authors need to sign the Contributor License Agreement:

• [email protected]

Click the button to sign:

@isidentical, it seems you need to sign the CLA again. It is now tied to the email address.

This now has a nontrivial conflict.
@isidentical, do you want to resolve it, or should I look into it?