------- Comment #6 from pault at gcc dot gnu dot org 2007-11-19 15:10 ------- (In reply to comment #5) > Off to generics and operators now!
Ahhh... I have run into a serious problem here. It transpires that renaming is not accomplished for generic interfaces by keeping the use-name symbol with multiple symtrees, each with a local-name. Instead, the symtree and the symbol have the local-name. This breaks the mechanism that I evolved above. The DEC Manual puts the rules rather nicely: If more than one USE statement for a given module appears in a scoping unit, the following rules apply: If one USE statement does not have the ONLY option, all public entities in the module are accessible, and any rename- lists and only-lists are interpreted as a single, concatenated rename-list. If all the USE statements have ONLY options, all the only-lists are interpreted as a single, concatenated only-list. Only those entities named in one or more of the only-lists are accessible. This, I think, shows the way forward. ie. In matching the USE statements, we only build up the rename-list and flag the presence of a USE statement without an ONLY. Then, at the last USE statement, we process all the referenced .mod files just once, using the concatenated rename-list as an only-list or a rename list, according to what we have seen. Paul -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33541
