wanted: shallow recursiveness for :FORCE keyword

[Tobias C. Rittweiler]

When I use :FORCE, in most cases I do want to recompile solely the
specified system, not all its dependencies.

In my ideal world, :FORCE T would recompile the specified system only,
and :FORCE :ALL would recompile it along its dependencies.

Of course that would be backwards-incompatible, so perhaps better stick
with the current behaviour for :FORCE T, but add :FORCE :SYSTEM, or
:FORCE :SHALLOW, or similiar.

(Original posting: http://thread.gmane.org/gmane.lisp.asdf.devel/31)

[Nikodemus Siivola (nikodemus)]

Actually, I don't think breaking backwards compatibility would be too terrible here: my assumption is that :FORCE is in 99% of cases called manually, so as long as users are made aware of the change it should break very little.

That said, easiest way might be to make it

 :FORCE-RECURSIVE <boolean>
 :FORCE-SHALLOW <boolean>

and deprecate :FORCE, but keep it as an alias for :FORCE-RECURSIVE for a year or so.

[Robert P. Goldman (rpgoldman)]

How about keeping the argument as :FORCE and making it accept :recursive or :shallow as an argument? Then keep t as a synonym for :recursive?

Also, what about making it :deep versus :shallow or :recursive versus :immediate?

Not a big deal, but I prefer the alternatives to be parallel....

cheers!

[Nikodemus Siivola (nikodemus)]

2009/12/21 Robert P. Goldman <email address hidden>:
> How about keeping the argument as :FORCE and making it accept :recursive
> or :shallow as an argument?  Then keep t as a synonym for :recursive?
>
> Also, what about making it :deep versus :shallow or :recursive versus
> :immediate?

The reason why I prefer :FORCE-RECURSIVE and :FORCE-SHALLOW is that there is less room for mistake, and the meaning is always clear -- no wondering about what T means, and a clear upgrade path.

Also, if we support T, :RECURSIVE and :SHALLOW, we should signal an error for all others -- otherwise it's far too easy to accidentally pass in :SYSTEM or similar bogus argument and have it do the wrong thing.

No biggie, though. I will not cry if we go your route.

[Faré (fahree)]

Quite some time ago (2.014.4), I fixed the :FORCE (SYSTEM1 SYSTEM2 ...) feature, and less long ago (2.20.21), I added a :FORCE-NOT (SYSTEM3 SYSTEM4 ...) feature.

Can we say that this is now a solved issue?

[Robert P. Goldman (rpgoldman)]

One thing we might want to do is to document these. :force doesn't seem well documented in the manual.

I'm afraid I'm a travel monster this month, but if someone would like to assign me a ticket to write this up, and I can do it in November, LMK.

[Faré (fahree)]

OK, still not documented in the manual, but now fixed in 2.26.8 and documented in the operate docstring.

:force t only forces the current system.
:force :all forces everything, recursively.

[Faré (fahree)]

Note that these are as Dan Barlow specified them back in his e0d02781 commit from 2003-03-16, but which never actually worked completely as specified until 2.26.8 from 2012-12-06.

[Faré (fahree)]

Fix released in 2.27