[Edbrowse-dev] fwd: [Geoff McLane] Re: getting some traction on https://github.com/htacg/tidy-html5/issues/65

[Edbrowse-dev] fwd: [Geoff McLane] Re: getting some traction on https://github.com/htacg/tidy-html5/issues/65

[Chris Brannon]

[-- Attachment #0: Type: message/rfc822, Size: 4343 bytes --]

From: Geoff McLane <ubuntu@geoffair.info>
To: public-htacg@w3.org
Subject: Re: getting some traction on  https://github.com/htacg/tidy-html5/issues/65
Date: Thu, 17 Sep 2015 20:16:06 +0200
Message-ID: <55FB0366.1080401@geoffair.info>

Hi Chris,

As put on the Issue #65 posts, this fix is now in the issue-65 branch...

After some testing time, will be merged into master...

Would be great if you could pull and build this branch, and see if it 
lives up to your expectations ;=))

Regards,
Geoff

[Edbrowse-dev] [Geoff McLane] getting some traction on...

[Karl Dahlke]

You can completely cut out my preprocessing by commenting out line 131 in html-tidy.c.
That leaves just tidy, to test
	var a = "<script";

Karl Dahlke

Re: [Edbrowse-dev] [Geoff McLane] getting some traction on...

[Chris Brannon]

Karl Dahlke <eklhad@comcast.net> writes:

> You can completely cut out my preprocessing by commenting out line 131 in html-tidy.c.
> That leaves just tidy, to test

Ok, so here's what people need to do if they want to test the tidy code.
First, pull their repo, check out the issue-65 branch, and build from
that instead of master.
Next, pull edbrowse.  I've added some ifdef blocks that will let you
test the experimental code.  Rebuild edbrowse, but pass
-DHAVE_TIDYSKIPQUOTES in CFLAGS, like so:
CFLAGS='-DHAVE_TIDYSKIPQUOTES' make

So here's what I've discovered so far.
Their fix works beautifully on the basic testcases, both the one in the
tidy 5 tests/ subdirectory and the one that Karl posted the other day.
But I tried it on a cached page from the fanfiction site, and it is
still having some kind of problem.  Next step: figure out where it's
breaking now and see if we can come up with another test case.

-- Chris

[Edbrowse-dev] [Geoff McLane] getting some traction on...

[Karl Dahlke]

Remember there is a second related bug:

var a = "<\/script>";

tsp and I tracked this one down.
fanfiction.net shows both bugs.
So does fictionpress.com.

Karl Dahlke

Re: [Edbrowse-dev] [Geoff McLane] getting some traction on...

[Chris Brannon]

Karl Dahlke <eklhad@comcast.net> writes:

> Remember there is a second related bug:
>
> var a = "<\/script>";

And a third:
            document.write('<style>body{font-family:"'+XCOOKIE.gui_font+'",Verdana, Arial;}</style>');
Taken from Tyler's fanfiction site...

-- Chris

[Edbrowse-dev] [Geoff McLane] getting some traction on...

[Karl Dahlke]

> document.write('<style>body{font-family:"'+XCOOKIE.gui_font+'",Verdana, Arial;}</style>');

These things generate tidy errors but usually don't derail the script
or the correct parsing of the page.
In fact every </foo inside a javascript becomes an error
and is converted into <\/foo, but usually causes no trouble.

var a = "</style>";

Karl Dahlke