------- Comment #12 from burnus at gcc dot gnu dot org 2010-05-07 16:37 -------
Created an attachment (id=20599)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=20599&action=view)
Fourth version of the draft patch
And fourth version. I have just realized that the patch is too simplistic and
that I might need go back to the drawing board. The following is currently not
handled:
a) Constraint check (cf. comment 3 [F03: C489, F08: C496]): the
component-spec-list has to be different from the interface of any of the
generic procedures (i.e. an actual-arg-spec-list matching the
component-spec-list shall not match any of the generic functions). Note: This
included handling default initializers as they act as optional arguments.
b) The algorithm does not handle: Define function in one module,* define type
in a different module, and use associate them [including constraint checking,
cf. (a)]. Currently, the renaming is only handled if either the generic
interface or the derived type is available before the other type is declared.
(* That's possible because the functions do not need to return the derived type
- even though that's the most common case.)
I probably solve this by renaming derived types to <name>@ and automatically
generating an associated generic function with the specific name <name>@ and a
dummy arg list [including DT parents (plural)]. One needs to modify
gfc_typename to print the type without @ and then one has also no problem with
the trailing @ sign.
--
burnus at gcc dot gnu dot org changed:
What |Removed |Added
----------------------------------------------------------------------------
Attachment #20592|0 |1
is obsolete| |
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=39427
