remove more optional tags

[paulirish]

these are probably the safest to add to the list:

</li>, </dt>, </dd>, </img>, </source>

[kangax]

Thanks for suggestions! From what I can see, img element has EMPTY content model, so can not have end tag. Ditto for source element (although as of now, I've really only targeted HTML 4.01). Removing </li>, </dt> and </dd> is theoretically safe, but I see that some browsers engage in some weird error-correction. For example, invalid markup like this — <ul><li>foo</li>x<li>bar</li>x</ul> — renders in FF3.6 as a list of 2 elements with "x" char following each item on a new line (making a list of 4 rows). If we were to remove </li> tags, the list would consist of 2 rows only, with "x" character appended right after "foo" and right after "bar". On the other hand, tolerating such invalid markup is probably not a good idea in the first place.

[mathiasbynens]

All these are optional:

</html>
</head>
</body>
</p>
</dt>
</dd>
</li>
</option>
</thead>
</th>
</tbody>
</tr>
</td>
</tfoot>
</colgroup>

Some of them are already being removed by the minifier.

I’m not aware of any problems with omitting the closing table elements like </td>. Anyone?

I agree tolerating invalid markup is not a good idea. The HTML minifier should expect valid, conforming markup.

[mathiasbynens]

From what I can see, img element has EMPTY content model, so can not have end tag.

Well, it can in (proper) XHTML.

[SunboX]

https://developers.google.com/speed/articles/optimizing-html?hl=de&csw=1

[kangax]

I've added remaining optional tags from @mathiasbynens list, except </p> and </li> to play it safe for now (no time to test edge/complex cases with unclosed <p> / <li>, but I'm all ears if anyone finds something)