Need less restrictive HTTP method validation

[wolfhechel]

First of all I'm sorry if this topic has been brought up before, I tried to search for a similar issue but couldn't find any.

As it is right now the only methods allowed for routing and requests are the ones defined in RFC7231 and RFC5789.
In reality there are far more methods used in the open like WebDAV methods (MOVE, COPY, UPDATE, LOCK, UNLOCK etc), GENA methods (SUBSCRIBE, UNSUBCRIBE, NOTIFY).

IANA lists the most common ones defined in RFC's at least, I was surprised to see GENA methods were missing though.

Point being, is there any reason as to why aiohttp is limited in this matter a part from the potential abuse in class-based views?

I know this is an issue already as I've done WebDAV and GENA work with aiohttp in projects, and as it is now I'm monkey-patching the method checks in order to allow route registration there.

[wolfhechel]

Also I'm sorry this issue doesn't follow the issue template, but as this is a discussion more than a bug report I hope you can find it in your heart to forgive me.

[asvetlov]

Please take a look on #960
Is the proposed solution satisfies your needs?

[wolfhechel]

Indeed that solves it for class based views at least, it still doesn't allow regular views as that's limited by the AbstractRoute implementation.

Would it be of any harm to remove lines 84:86 from AbstractRoute, since technically the only definition of a HTTP method is that it only consists of alphanumeric characters.

Actually, according to the HTTP 1.1 RFC a method can be any number of digits, letters or the symbols !#$%&'*+-.^_`|~.

[asvetlov]

My main reason for keeping checks was preventing user from silly types. For example I wanted to have a error on writing app.router.add_route('POSTT', '/', handler).

Now thanks to #989 we have shortcut methods for server api, and we already had these for client.

So my objection has gone away, we can relay checks significantly.

@wolfhechel are you volunteered for this task?

[wolfhechel]

@asvetlov Absolutely, I still think that methods should be validated according to RFC 1.1 specifications though so I'll implement that instead.

[asvetlov]

Fixed by #1037

[lock]

This thread has been automatically locked since there has not been
any recent activity after it was closed. Please open a new issue for
related bugs.

If you feel like there's important points made in this discussion,
please include those exceprts into that new issue.