Add {@any} helper.

[sethkinast]

• Must be contained in a {@select} block
• Renders its body if any truth test inside the {@select} block evaluates to true
• Can be included multiple times inside a {@select} block.

As part of this change, {@select} no longer clobbers the current context, so keys that begin with a dot work inside a {@select}. This closes #9, closes #42, closes #50, and closes #79.

[sethkinast]

Enables syntax such as:

{! foo is 1 or 2 or 3 !}
{@select key=foo}
  {@eq value="1"/}
  {@eq value="2"/}
  {@eq value="3"/}
  {@any}Foo is 1, 2, or 3{/any}
  {@default}Foo is not 1, 2, or 3{/default}
{/select}

{! Output something specific as well !}
{@select key=foo}
  {@any}Congratulations! You got: {/any}
  {@eq value="1"}First Place{/eq}
  {@eq value="2"}Second Place{/eq}
  {@eq value="3"}Third Place{/eq}
  {@default}Better luck next time!{/default}
{/select}

{! If any truth test passes, all {@any} blocks will be evaluated !}
{@select key=foo}
  {@any}<span class="place">{/any}
  {@eq value=1}First Place{/eq}
  {@eq value=2}Second Place{/eq}
  {@eq value=3}Third Place{/eq}
  {@any}</span>{/any}
{/select}

[sethkinast]

I purposely made the deferral generation generic in case we want to add the equivalent none. default mimics switch in that once it's hit, evaluation of further truth tests stops. I feel like it would have been more useful if you could have multiple default blocks that didn't stop truth testing, so that the default didn't have to be the last content in the block.

[prashn64]

lgtm, let's get one more set of eyes on it.

[sethkinast]

More than one more set I think, this is a fairly large grammar addition.

[rragan]

Has lots of value and slots in cleanly. Do you need a test with an any helper within another any helper within a select. That should just get diagnosed I think.

[sethkinast]

That's a really good call-- that actually crashes the template.

[sethkinast]

@rragan, are you OK with this syntax? I don't want you to feel like you're being run over-- your PR is the direct driver for this one. The team here was just worried about coming up with something that felt Dust-like.

[rragan]

The syntax works for me. It is a bit odd using a "select/case" mechanism for achieving a multi-output OR mechanism -- the name becomes a misnomer for someone looking at how to achieve OR behavior. Docs can help --, e.g. Need to do a an OR, see Select

[sethkinast]

@jimmyhchan whaddya think

[rragan]

When the "dust settles" on this round of changes and you are good to publish new dust and helpers, let me know and I can update the wiki.