You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The consensus was that it probably won't land. There's a couple of reasons:
This breaks existing users for little benefit (e.g. new URL() currently works).
It's an overall change to the design of Express' APIs - if you look at any other method, you'll see they all accept anything and rely on coercion to correct types.
The conversation itself broke down into two classes of issues. The first is whether we should have validation at all, and the majority take was no (we should only validate if it'd otherwise result in an invalid state, not because someone is using the API with something they shouldn't be, the goal of the framework is to be minimal with low overhead), and the second was whether we should be coercing things to make them work (e.g. non-strings into strings) which was more contentious.
The main split of opinion seemed to be on maintainer burden, with the exhaustion on APIs that accept anything making it hard on maintainers, vs breaking existing users who might be using an API successfully with something that we don't expect.
Activity
kjugi commentedon Feb 25, 2026
I think agenda was untouched. We have discussed this PR instead: expressjs/express#6404
bjohansebas commentedon Feb 26, 2026
@kjugi what was discussed about that?
blakeembrey commentedon Feb 26, 2026
The consensus was that it probably won't land. There's a couple of reasons:
new URL()currently works).The conversation itself broke down into two classes of issues. The first is whether we should have validation at all, and the majority take was no (we should only validate if it'd otherwise result in an invalid state, not because someone is using the API with something they shouldn't be, the goal of the framework is to be minimal with low overhead), and the second was whether we should be coercing things to make them work (e.g. non-strings into strings) which was more contentious.
The main split of opinion seemed to be on maintainer burden, with the exhaustion on APIs that accept anything making it hard on maintainers, vs breaking existing users who might be using an API successfully with something that we don't expect.