Err, that second example should be $form.add( ".invalid", true ).

Any thoughts before @trac-o-bot auto-closes this?

It won't close since it's not "pending". It's worth discussion.

We've had some discussion about relative selectors, which always end up going through Sizzle's JavaScript path. Where would this feature simplify code? Are there some real-life examples?

Good comments.

Regarding relative selectors, I doubt this enhancement would much increase their use, because anyone wanting to use $set.add( strRelative, true ) is probably ''already'' using $set.add( selRelative, $set ) or $set.find( selRelative ).andSelf(). But as an aside, isn't @timmywil's big Sizzle overhaul going to seriously improve the Javascript path for scoped queries anyway?

On to examples. .and and its derivative .andSelf seem to be fairly uncommon, but this new syntax would clarify many uses of the former (by eliminating otherwise unnecessary variable declarations) and most of the latter (by avoiding clunky re-adding).

• http://api.jquery.com/andSelf/: $("div").find("p").andSelf().addClass("border") → $("div").add( "p", true ).addClass("border")
• https://ajax.googleapis.com/ajax/libs/jqueryui/1.8.18/jquery-ui.js: uiDialogTitlebar.find("*").add(uiDialogTitlebar).disableSelection() → uiDialogTitlebar.add( "*", true ).disableSelection()
• http://stackoverflow.com/questions/3037382/jquery-this-selector-help; http://stackoverflow.com/questions/2828019/looking-for-jquery-find-method-that-includes-the-current-node: $set.find(selector).andSelf() → $set.add( selector, true )
• http://forum.jquery.com/topic/advanced-use-of-load-function; http://jsfiddle.net/2g8Yv/8/: $set.filter(selector).add( $set.find(selector) ) → $set.filter( selector ).add( selector, true )

To me, the use of .andSelf() (perhaps renamed to .addPrior() per #9800) is much clearer than using a second Boolean argument to .add(). Admittedly it's not as mystical as .stop(true, true) but the added method name makes it much clearer what is going on.

I agree, and in my opinion we'd be better off if this were the default context for .add( selector ) rather than null (document), but that would be a breaking change.

Can you come up with a better special-case value (null and undefined are probably out, as are strings and user-defined objects—but perhaps a conveniently-mapped true like like jQuery.CONTEXT_SELF?) or (*shudder*) new method name, or is this ticket doomed to rejection by historical contingency?

I think it makes more sense to do this with .addPrior() or an .addFind(selector) plugin at this point, I'm not a fan at all of a magical second argument. If history were rewritten I agree it might make sense to have .add() use the set elements as context, since it's so easy to add rooted elements via $(). Perhaps we can add a time machine to the 1.9 roadmap.