Re: html 4 name=name

Hi all:

On Fri, Jan 17, 2020 at 6:05 PM John Rouillard <[email protected]> wrote:
> On Thu, Jan 16, 2020 at 11:40 AM Joseph Myers <[email protected]> wrote:
> > On Thu, 16 Jan 2020, Ralf Schlatterbeck wrote:
> > > Now my question: Is this expected to work with HTML 4 or is this a bug?
> > > Should we have an explicit list of attribute names that use the short
> > > form? Or a blacklist?
> >
> > The form attribute="attribute" for boolean attributes is fully valid HTML4
> > and HTML5. ...
> > So I suggest reverting the previous change.
>
> I agree with Joseph's assessment. I thought that an attribute name
> being the same as an attribute value was extremely unlikely. Guess not
> as unlikely as I expected.
> Reverting that part of the change works. CI is breaking though in
> FormTestCase.testFormValuePreserveOnError and
> HTMLClassTestCase.test_string_field.
> Not sure why. i don't see how your change affects these cases.

Now I do.

> I'll try to fix the tests tomorrow.

Tests are fixed and CI is working again. The change Ralf made resulted
in input tags being terminated with  /> (xhtml syle) and not > (html
4/5 style).
I fixed this by reverting Ralf's change and changing the html 4 entity
generation code to generate boolean style values only when the
attribute value
is None.  Setting the attribute value to the name of the atribute( eg.
{ "name": "name"}) just creates name="name" in the output.

Also cleaned up inline doc to remove the disabled functionality, fixed
the tests testing for the name="name" case and cleaned up
CHANGES.txt so it properly describes the fixed functionality.

-- rouilj


> Bummer.
>
> -- rouilj