</> terminating tag [Question/Bug] #1966

BuGlessRB · 2016-09-01T11:19:04Z

I created a tag to make the page-title auto-updating, and ended up with this:

<title>
 <yield /> - { title }
 </>
 <script>
  var self = this;
  self.mixin("bus");
  self.bus.on("settitle", function(title)
   { self.update({"title":title});
   });
 </script>
</title>

It turns out , that in order to get this working, I was forced to terminate the HTML part of the tag using </>, just before the <script>. The questions would be:

Is it possible to make that unnecessary by causing the <script> to autoterminate the HTML part instead?
Or if that is not possible, is </> a supported way to accomplish this?

Using <span>...</span> to encapsulate the HTML is not an option, since browsers do not tolerate any nested HTML directives in the <title> tag.

The text was updated successfully, but these errors were encountered:

aMarCruz · 2016-09-01T20:01:55Z

Interesting
@BuGlessRB are you using this tag in <head> ?

This is a compiler's bug.

BuGlessRB · 2016-09-01T22:04:43Z

Yes, just the normal <title> in <head>. Is there any reason why this might not work?
It looks like:

<head>
 <title>SiteName</title>
</head>

And the custom tag turns that into:

<head>
 <title>SiteName - Page Name Foo Bar</title>
</head>

aMarCruz · 2016-09-01T22:30:30Z

@BuGlessRB this doesn't works because is not ending with a valid closing tag, see the Guide.

With regexes, there's no way to detect expressions ending the html part, so I take your hack to give support for these cases in the compiler with a minor change: this will be <-/> so you can use this in tags inside html pages as well (browsers removes </> before the compiler can read it).

The addition of the BuGless-hack (may I?) is in the "dev" branch and will be release in v2.5.4 of the riot-compiler, and docummented in the section Untagged HTML content of the guide, with a brief example.

BuGlessRB · 2016-09-01T22:37:34Z

The nick is there for a reason, so BuGless-hack is fine :-).

The <-/> solution works for me.
However, I understand that using mere regexes to detect the end of the html part is difficult, but if you use an explicit <script></script>, shouldn't it be trivially possible to detect the start of the <script> part?

aMarCruz · 2016-09-01T22:47:08Z

nope, you can use multiple script tags + untagged js:

<my-tag>
  var title = 'foo'
  var x = { title }
  <script type="typescript">
    const y :string = "bar"
  </script>
</my-tag>

BuGlessRB · 2016-09-01T22:48:40Z

Hmmm, well, that does not strike me as a feature (putting it politely).

aMarCruz · 2016-09-01T22:49:57Z

agree, but is in the specs before me :)

BuGlessRB · 2016-09-01T22:52:37Z

Well, it shouldn't have been in those specs in the first place (if you ask me). So I vote to eliminate that part of the specs for 3.0.0.

I would even go as far as making <script></script> mandatory if you use any script, but would allow for multiple <script></script> blocks.

aMarCruz · 2016-09-01T23:00:45Z

It was propesed by some people yet, maybe we can do something about this for v3.
Making <script> mandatory solves gotchas and simplifies the code of the compiler, with a safer detection.

@riot/core-maintainers what do you think?

I'm not sure about such change because you can put <script> in riot tags inside html pages.

BuGlessRB · 2016-09-01T23:31:14Z

Ah, due to the fact that script tags cannot be nested.
Yes, true.

Nonetheless, I'd say, then make that an exception: i.e. only support missing <script> tags to start the javascript if defined inside a script tag already. If using a normal riottag definition, require a <script> tag to delimit the first use of any scripting language.

aMarCruz · 2016-09-01T23:51:36Z

mmm... I'll will think about it.

BuGlessRB · 2016-09-02T00:24:55Z

Besides solving some parsing issues, I expect for it to speed up the parsing process as well.

Incidentally, I just tried the dev branch from github, but it does not seem to contain the fix just yet. Am I missing something?

GianlucaGuarini · 2016-09-02T06:02:26Z

I would even go as far as making <script></script> mandatory if you use any script, but would allow for multiple <script></script> blocks.

@aMarCruz @BuGlessRB I would not force our users using scripts to handle their tags logic. Riot is build with the idea of keeping things simple, it's not opinionated so it will give you the freedom to write your code however you want. I wouldn't care that much about such edge cases and protecting users from themselves

BuGlessRB · 2016-09-02T08:28:14Z

Well, the only things I can say on that are that:

I love the philosophy of Riot to keep things simple.
I also wholeheartedly agree with not enforcing certain programming paradigms (i.e. leave the programmer to choose how he solves things).
I wholeheartedly disagree that this freedom extends into the used syntax and grammar of a language.
Tools to help a programmer do static analysis of code or syntax highlighting work a lot better if the syntax and grammar are well defined.
Things like optional delimiters (be it semicolons or <script> tags) I consider a design flaw. They unduly complicate the task of figuring out the intention of the programmer.
When enforcing a bit of structure even speeds up the runtime compilation in the browser, then that is considered a free bonus.

cognitom · 2016-09-03T02:59:13Z

</> doesn't make sense for me... It looks not like a part of HTML. And IMO we shouldn't use Riot tag in <head>. That could just introduce a bug into the code.

@aMarCruz honestly speaking, I like the idea of mandatory <script> and believe it's simpler than optional <script>, but it seems a bit inconsistent to the code outside the tag.

theCodeOutsideTagIsAllowedInRiot()
<my-tag>
  ...
</my-tag>

We might need a kind of strict mode but it needs more discussions from many aspects. At least, we shouldn't start the discussion from this issue, I think. Maybe, Riot v4 or later?

BuGlessRB · 2016-09-03T11:35:31Z

</> is not meant to make sense, it's just the only way that works to delimit the end of the HTML without using HTML.

Why would using riot tags in things as <title> or <meta name="description"> not be valid and simple use cases?

With regard to mandatory <script>, in inlined tags one could use a nested <riot-script>; and if this discussion should not be held here, then where?

aMarCruz · 2016-09-03T14:39:47Z

@cognitom mandatory script is good, I use this always, but JS outside the tag seems inconsistent with or without script, I think there's no syntax hilighter that can handle this :) ...anyway, the inability to place nested script tags on pages is the major drawback for implementing this requirement and solutions as riot-script go against the riot philosophy.

Maybe riot v4 can do something about this...
@BuGlessRB I will close this issue now, the bugless-hack is sufficient for cases like this. Thanks!

aMarCruz · 2016-09-03T14:48:22Z

@GianlucaGuarini agreed, thanks.

@BuGlessRB sorry, the update is in the dev branch of the riot-compiler.
Fork the repo, checkout 'dev' and compile this, then link to riot and compile:

git clone https://github.com/riot/compiler && cd compiler
git checkout dev && npm i && make
sudo npm link
cd ../<your riot repo>
npm un riot-compiler && npm link riot-compiler
make riot

aMarCruz self-assigned this Sep 1, 2016

aMarCruz added the bug label Sep 1, 2016

aMarCruz pushed a commit to riot/compiler that referenced this issue Sep 1, 2016

Hack for riot/riot#1966

8008b2b

koobitor pushed a commit to koobitor/riot that referenced this issue Sep 2, 2016

fix riot#1966 terminating tag

85838e3

aMarCruz added the wontfix label Sep 3, 2016

aMarCruz closed this as completed Sep 3, 2016

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

</> terminating tag [Question/Bug] #1966

</> terminating tag [Question/Bug] #1966

BuGlessRB commented Sep 1, 2016 •

edited

Loading

aMarCruz commented Sep 1, 2016 •

edited

Loading

BuGlessRB commented Sep 1, 2016 •

edited

Loading

aMarCruz commented Sep 1, 2016 •

edited

Loading

BuGlessRB commented Sep 1, 2016 •

edited

Loading

aMarCruz commented Sep 1, 2016

BuGlessRB commented Sep 1, 2016

aMarCruz commented Sep 1, 2016

BuGlessRB commented Sep 1, 2016

aMarCruz commented Sep 1, 2016 •

edited

Loading

BuGlessRB commented Sep 1, 2016

aMarCruz commented Sep 1, 2016

BuGlessRB commented Sep 2, 2016

GianlucaGuarini commented Sep 2, 2016

BuGlessRB commented Sep 2, 2016

cognitom commented Sep 3, 2016 •

edited

Loading

BuGlessRB commented Sep 3, 2016

aMarCruz commented Sep 3, 2016

aMarCruz commented Sep 3, 2016 •

edited

Loading

</> terminating tag [Question/Bug] #1966

</> terminating tag [Question/Bug] #1966

Comments

BuGlessRB commented Sep 1, 2016 • edited Loading

aMarCruz commented Sep 1, 2016 • edited Loading

BuGlessRB commented Sep 1, 2016 • edited Loading

aMarCruz commented Sep 1, 2016 • edited Loading

BuGlessRB commented Sep 1, 2016 • edited Loading

aMarCruz commented Sep 1, 2016

BuGlessRB commented Sep 1, 2016

aMarCruz commented Sep 1, 2016

BuGlessRB commented Sep 1, 2016

aMarCruz commented Sep 1, 2016 • edited Loading

BuGlessRB commented Sep 1, 2016

aMarCruz commented Sep 1, 2016

BuGlessRB commented Sep 2, 2016

GianlucaGuarini commented Sep 2, 2016

BuGlessRB commented Sep 2, 2016

cognitom commented Sep 3, 2016 • edited Loading

BuGlessRB commented Sep 3, 2016

aMarCruz commented Sep 3, 2016

aMarCruz commented Sep 3, 2016 • edited Loading

BuGlessRB commented Sep 1, 2016 •

edited

Loading

aMarCruz commented Sep 1, 2016 •

edited

Loading

BuGlessRB commented Sep 1, 2016 •

edited

Loading

aMarCruz commented Sep 1, 2016 •

edited

Loading

BuGlessRB commented Sep 1, 2016 •

edited

Loading

aMarCruz commented Sep 1, 2016 •

edited

Loading

cognitom commented Sep 3, 2016 •

edited

Loading

aMarCruz commented Sep 3, 2016 •

edited

Loading