Feature #5778: Allow WEBrick::HTTPResponse to send IO-duck-typed bodies

• Status: Assigned
• Priority: Normal
• Assignee: normalperson (Eric Wong)
• Target version: -

[regularfry (Alex Young)]

WEBrick::HTTPResponse currently type-checks for IO to determine how to transfer content to the outgoing socket. Because of this, it's not possible to use "IO-like" objects as data sources unless they specifically inherit from IO.

The interface HTTPResponse requires from its body objects is very simple: it only calls #read and #close, and optionally #bytesize. This means that the type-check is needlessly strict, and prevents custom objects from being used as data sources. The attached patch removes the type-check on IO, allowing duck-typed objects to be used.

The downside to this patch is that you can't pass in objects that duck-type to String's interface. It wouldn't be hard to remove that restriction if necessary, but I figured this was the less intrusive approach.

[mame (Yusuke Endoh)]

• Status changed from Open to Assigned
• Assignee set to mame (Yusuke Endoh)

It is trivial, and the patch looks good to me.
I'll commit it unless there is objection.

--
Yusuke Endoh [e-mail]

[nahi (Hiroshi Nakamura)]

Yusuke, you may want to assign it to me (maintainer)
I think it's good to improve, but not yet checked the patch...

[mame (Yusuke Endoh)]

• Assignee changed from mame (Yusuke Endoh) to nahi (Hiroshi Nakamura)

You have control.

--
Yusuke Endoh [e-mail]

[mame (Yusuke Endoh)]

• Target version set to 2.6

Blame NaHi!

--
Yusuke Endoh [e-mail]

[naruse (Yui NARUSE)]

• Target version deleted (2.6)

[mame (Yusuke Endoh)]

• Assignee changed from nahi (Hiroshi Nakamura) to normalperson (Eric Wong)

normalperson (Eric Wong), could you handle this ticket?

[normalperson (Eric Wong)]

[e-mail] wrote:

normalperson (Eric Wong), could you handle this ticket?
https://bugs.ruby-lang.org/issues/5778#change-72539

This seems identical to [Feature #8155] and already implemented.
https://bugs.ruby-lang.org/issues/8155