[Greasemonkey] Script scoping

[Aaron Boodman]

We can remove this; I'm not married to it.

The main reason I added it was to make GM_* functions local to user
scripts, but I can probably (?) think of another way to accomplish
that. Mark would have to revise his book.

It seems to be causing lots of problems and gotchas. 

-- 
Aaron

On 5/2/05, Jeremy Dunck <jdunck at gmail.com> wrote:
> On 5/2/05, Andrew Hayward <greasemonkey at mooncalf.me.uk> wrote:
> > I've just got round to installing GM0.3b, after finally getting fed up
> > of not having the new features at my disposal. However, I noticed the
> > following in the change log...
> >
> > > Improved error handling by isolating scripts from one another. Errors
> > > in one script will no longer stop other scripts from running. Errors
> > > loading a script will not stop other scripts from being loaded.
> >
> > That's all well and good, but what happens if I actually want to access
> > another script?
> 
> Hey Andrew, cross-script dependencies, eh.  There'd been some talk
> about this, but no support for it yet.
> 
> Others have already suggested the window.foo = yourfirstsscriptfunc() approach.
> 
> The only downside of this is that your first script has to be
> injected, err, first.  Which is the same as it was before, modulo the
> scoping issue.
> 
> That is to say, if you had cross script dependencies before, even
> without the anon function scoping added in 0.3b, the dependant script
> would have to inject after the lib script.
> 
> So..  that suggestion may already solve your problem.
> 
> Otherwise, I like the idea of @scriptinclude
> yournamespace/yourlibname, which, the way I'm thinking, would inject
> another (already installed) script, regardless of its
> @include/@exclude.
> 
> Would that work for you?  ...Injection would get more complicated,
> because the order of injection would have to be smarter to resolve
> dependencies.
> 
> Hmm, also, GM_getValue and GM_setValue on the library would be
> something else to discuss.  Should values stored as a library
> injection be under the top script name, or the lib script name?
> 
> ...Sure you can't preprocess just your user.js to just include all the
> stuff you want into the deployed user.js file?
> 
> We can work this out, but for now, I recommend you hack up your
> greasemonkey install.
> browser.xul line 124:
>               runBrowserScript(
>                 doc,
>                 [ "(function(){",
>                   "var GM_xmlhttpRequest = window.GM_xmlhttpRequest;",
>                   "var GM_registerMenuCommand = window.GM_registerMenuCommand;",
>                   "var GM_setValue = window.GM_setValue;",
>                   "var GM_getValue = window.GM_getValue;",
>                   "var GM_log = window.GM_log;\n",
>                   getContents(getScriptChrome(script.filename)),
>                   "})();"
>                   ].join("\n")
>               );
> to:
>               runBrowserScript(
>                 doc,
>                 [
>                   "var GM_xmlhttpRequest = window.GM_xmlhttpRequest;",
>                   "var GM_registerMenuCommand = window.GM_registerMenuCommand;",
>                   "var GM_setValue = window.GM_setValue;",
>                   "var GM_getValue = window.GM_getValue;",
>                   "var GM_log = window.GM_log;\n",
>                   getContents(getScriptChrome(script.filename))
>                   ].join("\n")
>               );
> _______________________________________________
> Greasemonkey mailing list
> Greasemonkey at mozdev.org
> http://mozdev.org/mailman/listinfo/greasemonkey
>