545854 - [HTML5][Patch] Add a <keygen> macro to the HTML5 parser

Status: RESOLVED FIXED

(Core :: DOM: HTML Parser, defect, P1)

Description

[Henri Sivonen (:hsivonen) (on vacation)]

To take bug 101019 off the critical path of getting the HTML5 parser enabled, implement <keygen> as a macro in the tree op executor.

Comment 1

[Henri Sivonen (:hsivonen) (on vacation)]

Attachment: Add a bugwards-compatible keygen parser macro

See CNavDTD for the old implementation.

This implementation is deliberately bugwards-compatible with the old impl.: The author can specify _moz-type that overrides the parser-generated attribute.

Demo for smoketesting at http://software.hixie.ch/utilities/js/live-dom-viewer/saved/376

Comment 2

[Ben Newman (:bnewman) (:benjamn)]

Comment on attachment 426987 [details] [diff] [review]
Add a bugwards-compatible keygen parser macro

>+      } else if (NS_UNLIKELY(isKeygen)) {
>+        // Adapted from CNavDTD
>+        nsCOMPtr<nsIFormProcessor> theFormProcessor =
>+          do_GetService(kFormProcessorCID, &rv);
>+        // ignore rv
>+        
>+        nsTArray<nsString> theContent;
>+        nsAutoString theAttribute;
>+         
>+        (void) theFormProcessor->ProvideContent(NS_LITERAL_STRING("select"),
>+                                                theContent,
>+                                                theAttribute);

Why ignore rv?  Could you at least NS_ENSURE_TRUE(theFormProcessor, rv), unless you're certain do_GetService is infallible?

Otherwise looking good.  The parts based on CNavDTD are a lot simpler than the CNavDTD implementation, too.

Comment 3

[Henri Sivonen (:hsivonen) (on vacation)]

(In reply to comment #2)
> (From update of attachment 426987 [details] [diff] [review])
> >+      } else if (NS_UNLIKELY(isKeygen)) {
> >+        // Adapted from CNavDTD
> >+        nsCOMPtr<nsIFormProcessor> theFormProcessor =
> >+          do_GetService(kFormProcessorCID, &rv);
> >+        // ignore rv
> >+        
> >+        nsTArray<nsString> theContent;
> >+        nsAutoString theAttribute;
> >+         
> >+        (void) theFormProcessor->ProvideContent(NS_LITERAL_STRING("select"),
> >+                                                theContent,
> >+                                                theAttribute);
> 
> Why ignore rv?

Because I thought malloc failures were supposed to be ignored per the meeting in December.

> Could you at least NS_ENSURE_TRUE(theFormProcessor, rv), unless
> you're certain do_GetService is infallible?

I added NS_ENSURE_SUCCESS.

Pushed:
http://hg.mozilla.org/mozilla-central/rev/b2d1b71e8321