On Fri, Apr 18, 2014 at 5:09 PM, Robert P. Goldman rpgoldman@sift.info wrote:
Faré wrote:
OK, so the main contender seem to be, without -system suffix:
one-package-per-file
Robert, are you alright with that name? Do you insist on the shorter variant package-per-file ?
I can do the renaming this weekend, and update the various relevant libraries.
I don't care so much about whether or not one- is there.
But it seems to me that it would be a LOT better if "SYSTEM" appeared somewhere in this name. Even to the point that PPF-SYSTEM would be better than PACKAGE-PER-FILE...
I guess
(DEFSYSTEM foo :CLASS ONE-PACKAGE-PER-FILE )
isn't terrible, but ONE-PACKAGE-PER-FILE doesn't scream "I am a subclass of SYSTEM" to me. Note for comparison that all of our file subtypes have -FILE (like STATIC-FILE, JAVA-FILE, etc.). That seems like a good convention.
On the one hand, I hear your arguments, on the other hand, they are mitigated by the fact that: * where there is a :class one-package-per-file, there are usually no :components being declared, and so little opportunity for confusion. * conversely, parse-component-form since ASDF3 has a guard against using a subclass of system as the class of a subcomponent: (when (and parent (subtypep class 'system)) (error 'non-toplevel-system :parent parent :name name))
That makes the -system suffix less necessary in my eyes, though I would usually have reacted just as you did, and in this case, I would drop it for the sake of concision, especially since one-package-per-file-system might confusingly sound like a filesystem that has only one package, which is the opposite of it.
I will still wait for comments, but unless a better proposal comes along and/or Robert vouches for a different name, I will rename package-system to: one-package-per-file
—♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org Love doesn't scale. — Eric S. Raymond