[Patch, fortran] PR34672 - 4.3 Regression - .mod file misses renamed, USEd symbol

[Paul Richard Thomas]

:ADDPATCH fortran:

To fix this, I rewrote write_generic.  This function was so wrong that
I am amazed that it worked at all:)

In particular, it used the 'find_use_name' mechanism..... relying on
the use names left over from the previous module_read.  This is why
there was a sensitivity on the order of the USE statements.

I have not researched how this came into the tree because I rather
think that I am the guilty party.

Anyway, the fix is easy: scan the sym_root symtree, looking for
generic interfaces.  Give them the name of the symtree; ie. the local
name.

Regtested on Cywin_NT/PIV - OK for trunk? (I'll check that bootstrap
is OK tonight.)

Paul

2008-01-07  Paul Thomas  <pault@gcc.gnu.org>

	PR fortran/34672
	* module.c (write_generic): Rewrite completely.
	(write_module): Change call to write_generic.

2008-01-07  Paul Thomas  <pault@gcc.gnu.org>

	PR fortran/34672
	* gfortran.dg/use_only_2.f90: New test.
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: submit.diff
URL: <http://gcc.gnu.org/pipermail/gcc-patches/attachments/20080107/86e6da98/attachment.ksh>