Bug 215563 – NetworkSocketChannel should delay sending back error message as done by NetworkSocketStream

• Component: WebCore Misc.
• Status: RESOLVED FIXED

[youenn fablet]

Delay sending back to WebProcess WebSocket error message as done in NetworkSocketStream

[youenn fablet]

Created attachment 406709 [details]
Patch

[Alex Christensen]

Comment on attachment 406709 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=406709&action=review

> Source/WebKit/NetworkProcess/NetworkSocketChannel.h:94
> +    String m_errorMessage;

It would be nice if we could make a version of Timer::startOneShot that takes a lambda so these could be in a lambda capture instead of members of this object.

> Source/WebKit/NetworkProcess/NetworkSocketChannel.h:95
> +    Optional<std::pair<unsigned short, String>> m_closeInfo;

Could we make this uint16_t?

[youenn fablet]

(In reply to Alex Christensen from comment #2)
> Comment on attachment 406709 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=406709&action=review
> 
> > Source/WebKit/NetworkProcess/NetworkSocketChannel.h:94
> > +    String m_errorMessage;
> 
> It would be nice if we could make a version of Timer::startOneShot that
> takes a lambda so these could be in a lambda capture instead of members of
> this object.

Agreed, but here we have two call sites, one for the error message and one for code and reason.

> > Source/WebKit/NetworkProcess/NetworkSocketChannel.h:95
> > +    Optional<std::pair<unsigned short, String>> m_closeInfo;
> 
> Could we make this uint16_t?

We probably could since it is an unsigned short for WebIDL as well.
But this would not be very consistent with other parts of the WebSocket implementation.

[EWS]

Committed r265814: <https://trac.webkit.org/changeset/265814>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 406709 [details].

[Radar WebKit Bug Importer]

<rdar://problem/67327819>