Consider Workbench class

[Elrond]

Out of random thinking, I was pondering about a Workbench class.

This ticket is mainly to not loose the discussion points, etc.

Ideas, etc will pop up on my dev/workbench\_class branch.

[Elrond]

Some parts of our discussion till now:

::

    <Elrond> I was considering to let the WorkbenchManager return "Workbench"-objects. The object could have the localize_file method on it for example.
    <paroneayea> yeah I thought about that too
    <paroneayea> actually localize_file doesn't need to be a method of anything
    <paroneayea> really
    <paroneayea> it doesn't even use the manager
    <Elrond> But it uses the "workbench" ;)
    <paroneayea> it just seemed like if the mgoblin_globals method was going to have the workbench that it would be convenient to have the method there
    <paroneayea> yeah
    <paroneayea> Elrond: do you think it'd be cleaner?
    <paroneayea> all it would need I guess is
    <paroneayea> well maybe it could be
    <paroneayea> Workbench.localize_file(), Workbench.directory, workbench.destroy_self()
    <Elrond> It might allow for "with ..." too. ;)
    <paroneayea> hm
    <paroneayea> yeah
    <Elrond> workbench.destroy_self was my actual question...
    <Elrond> You'd say, the desctruction should happen on the Workbench itself?
    <paroneayea> I'm not sure.
    <paroneayea> could still be that the manager takes a workbench object
    <paroneayea> might be "safer" since the manager does that check
    <paroneayea> I guess it isn't sanely really any safer.
    <Elrond> Yeah, that was my thinking somewhat.
    <paroneayea> Elrond: anyway, if you feel strongly about this, since I was thinking along similar lines and was unsure, make that and adjust the tests / processing code & I'll merge.
    <paroneayea> I was sitting there staring and debating internally and  eventually just decided "screw it, picking this one" 
    <Elrond> I did mostly. ;)
    <paroneayea> oh :)
    <paroneayea> haha

my tree is now online.

[Christopher Allan Webber]

Elrond, if you remove the WorkbenchManager.localized\_file()
wrapper and adjust the code to fit that, I'm happy to merge this.
Now that we have a real Workbench class I don't see any reason to
have the manager have a localized\_file() method.

[Elrond]

Not targetted at 0.0.3.
As it's done, it'll be in that version anyway.

[Will Kahn-Greene]

The original url for this bug was ​http://bugs.foocorp.net/issues/380 .