Using parser.models forces email-validator dependency.

[peterschutt]

I already use pydantic for domain modelling but before now haven't tried to use the parser models.

I figured seeing I already have pydantic around I would have been good to go, but the package namespace imports in utilities.parser.models from .ses cause pydantic to throw an import error due to from pydantic.networks import EmailStr in ses.py.

This is the lib it forces to install is https://github.com/JoshData/python-email-validator - not a big library by any stretch, so the bloat effect would be on the small side.

Just figured I'd canvas opinion before opening an issue.

Cheers!

[heitorlessa]

Hey Pete, we've got a PR for this and an issue on this already.

#1259

[peterschutt]

My apologies! Thanks for the reference.

[heitorlessa]

No worries at all ;) we should get this addressed in the next release.

Pydantic / Parser and Event Source Data Classes are two features I'd like to rethink from the ground up in the future -- overlap, package size (modularisation), duplication, and dangers of raising ValidationError for event sources that don't have a schema IF we were to make Pydantic v2 the default experience (not extra)

[peterschutt]

Thanks:) I really should have searched first 🤦.

I haven't tested with pydantic v2 yet - the latest Talk Python to Me pod features it - sounds like big performance gains.

Perhaps the role that the dataclasses play today could be replaced by TypedDict definitions so users looking for more performance could choose to build around them and still get the type hints. Overlap between the pydantic models and dataclasses is definitely a thing that many wouldn't appreciate the need for, especially if the pydantic performance is to improve as much as it seems.

Thanks mate.

[peterschutt]

The topic in #1404 has me thinking about whether the pydantic models could be a totally separate library that powertools depends on so that they could maybe have a wider version support than the core library.

[heitorlessa]

That was the intention but we wouldn’t have the bandwidth to maintain both - we might be able to next year. For now, I think modularisation will be a great stepping stone. We’ll look into that after 3.6 drop (v2).

…

On Sun, 7 Aug 2022 at 13:53, Peter Schutt ***@***.***> wrote: The topic in #1404 <#1404> has me thinking about whether the pydantic models could be a totally separate library that powertools depends on so that they could maybe have a wider version support than the core library. — Reply to this email directly, view it on GitHub <#1431 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAZPQBHP2OX6YKIBE2DNUKTVX6PUFANCNFSM55ZSEYUA> . You are receiving this because you commented.Message ID: <awslabs/aws-lambda-powertools-python/repo-discussions/1431/comments/3343059 @github.com>

[heitorlessa]

Update - this is no longer the case in V2. email-validator was dropped as a dependency. All dependencies being optional

https://github.com/awslabs/aws-lambda-powertools-python/releases/tag/v2.0.0