Parser should report error

[raducoravu]

With this LESS fragment:

  .tip {
      font-size: 1em;
      line-height: 1.2em: 
     margin: 20px 0px 30px 0px;
     }

the "line-height" value ends with a ":" instead of ";" and the parser should report an error. It does not.

[infotexture]

Oddly enough, if a line comment is added to the invalid line, the parser recognizes the error:

.tip {
    font-size: 1em;
    line-height: 1.2em:         // Trailing colon should be a semicolon
    margin: 20px 0px 30px 0px;
    }

In this case, the parser responds with:

ParseError: Unrecognised input on line 3, column 20:

[seven-phases-max]

It's related to https://github.com/less/less.js/blob/v2.6.1/lib/less/parser/parser.js#L394. In fact if parser has to detect and throw a error there it will also has to throw a error for any "non-standard" syntax like: color: microsoft-forever: alpha(50);. It would probably make sense to discourage to use any unescaped MS-syntax in a long run (since it's mostly about IE6-9 which is weird in 2016), but still it will be quite breaking change.

[matthew-dean]

It would probably make sense to discourage to use any unescaped MS-syntax in a long run

Yes. Even MS browsers don't have this kind of backward-compatibility with some of their earlier syntax attempts. I think at this point (maybe for 3.0), we can just pull weird MS syntax checks since people should still be able to support them with string escapes if they need to.

That said, we probably want to strike a balance. There are still hacks being used in 2016 we wouldn't want to pull at the same time. See: http://browserhacks.com/. Anything that targets IE 8, 10, or 11 is still relevant and probably in use, especially because of this website.