Anne van Kesteren wrote:
On Mon, May 6, 2013 at 4:39 PM, Jason Orendorff
<[email protected]> wrote:
And this "weird" behavior is hardly without precedent: aren't *all*
other URLs a script deals with, including the existing APIs for
loading other scripts (adding a<script> tag or loading the code with
an XHR), resolved relative to the document base URL?
Typically the entry script's base URL, which can be different in
cross-document scenarios. And for XMLHttpRequest it's the document
associated with its constructor object (which annoyingly is somewhat
different from the rest).
Noted!
Good point. It can do what XHR does in workers: use the script's base
url. (If the site happens to have a single directory for serving
static js, and the worker script happens to be in it, that's
especially convenient.)
So now you have two conventions...
If the newer one is better, that's progress. Building on a broken
("annoyingly") convention to keep some unity-of-conventions ideal at the
expense of usability is not the obvious winner.
With the Web, you never can hope to have "one way to do it". The goal is
"better over time".
/be
_______________________________________________
es-discuss mailing list
[email protected]
https://mail.mozilla.org/listinfo/es-discuss
