On Tue, May 7, 2013 at 1:17 AM, Fitzgerald, Nick wrote:
> * IMO, it is impractical to just take every foo/bar.js and see if there is
> a foo/bar.map because the extra http request overhead (think when there are
> dozens/hundreds or scripts on a page), possibly slowing down the displaying
> of sources in debuggers depending on implementation, questions as to what
> to do when there is both a @sourceMappingURL and a foo/bar.map, etc
>
+1. Especially on just trying. It leads to unneccessary overhead. I'd
rather go for inlined (/dynamic) sourcemaps in that case. This is mostly a
developer step or stage anyways.
> * The possibility exists of adding a new @hasImplicitSourceMap style
> comment directive that would basically say that if the current script is
> foo/bar.js then there should be a source map located at foo/bar.map, which
> would make the whole renaming both the generated js and the source map
> scenario easier to deal with. I'm wary of adding these @ directives because
> once you do they tend to never go away and they have the potential to add a
> bunch of cruft that debugger authors will have to deal with. That said, if
> this is the best solution, then I'm all for it.
>
Bikeshedding; rather than introduce a new pragma for this, we could add a
rule that leaving out the name means implicit? (Or we could use a token for
the name to indicate the implicity, but that's really bikeshedding ;) I
dunno. Otherwise this is fine by me.
- peter
|
|