- From: Brendan Eich <brendan@mozilla.org>
- Date: Fri, 07 Dec 2012 13:58:43 -0800
- To: David Sheets <kosmo.zb@gmail.com>
- CC: Anne van Kesteren <annevk@annevk.nl>, bugzilla@jessica.w3.org, public-script-coord@w3.org
David Sheets wrote: > On Fri, Dec 7, 2012 at 2:43 AM, Anne van Kesteren<annevk@annevk.nl> wrote: >> On Fri, Dec 7, 2012 at 5:41 AM, David Sheets<kosmo.zb@gmail.com> wrote: >>> How does this opinion resolve the bug as invalid? Either web-apps or >>> WebIDL is in error. >> The bug was filed against WebIDL. WebIDL does not consider its >> definition of enum to be a bug, so web-apps needs to be fixed. > > So WONTFIX, then? > >>> Is it really worth it to fail to parse WebIDL because of a (useful for >>> autogenerated lists) trailing comma? >> That seems like a slippery slope. > > What's at the bottom of the slope? Simple line-based elision? > >> And you could use join instead. > > This argument has convinced the C, ECMAScript, Python, Ruby, OCaml, > and other language communities to mandate fatal errors on trailing > delimiters. Developers in these languages and targeting these > languages enjoy these pointless fatal errors. There is no such error for JS (ECMAScript), for arrays literals: js> a = [1, 2, 3, ]; [1, 2, 3] I agree with you these are pointless. It isn't just the auto-generated list use-case that suffers. Manually maintained lists and version control of same using line-based diff also pay a small ongoing price. I was sad back in the day when (ANSI?) C standardized against optional trailing comma in enum. I fought hard for optional trailing comma in array initialisers in JS. /be
Received on Friday, 7 December 2012 21:59:15 UTC