Un-believable bug

[FX]

>> implicit real (a-h,o-z); parameter (i=0)
>> end

Tobias: your patch is reworking implicit parsing to some extent, maybe it fixes it? I can’t test, I have removed it from my tree after review.
Or otherwise, having the code fresh in memory, maybe you see an “obvious” fix?

Finally, it could be that we simply need this:

Index: decl.c
===================================================================
--- decl.c	(revision 215645)
+++ decl.c	(working copy)
@@ -3091,7 +3091,7 @@ gfc_match_implicit (void)
 	  /* We may have <TYPE> (<RANGE>).  */
 	  gfc_gobble_whitespace ();
 	  c = gfc_next_ascii_char ();
-	  if ((c == '\n') || (c == ','))
+	  if ((c == '\n') || (c == ',') || (c == ';'))
 	    {
 	      /* Check for CHARACTER with no length parameter.  */
 	      if (ts.type == BT_CHARACTER && !ts.u.cl)
@@ -3140,7 +3140,7 @@ gfc_match_implicit (void)
 
       gfc_gobble_whitespace ();
       c = gfc_next_ascii_char ();
-      if ((c != '\n') && (c != ','))
+      if ((c != '\n') && (c != ',') && (c != ';'))
 	goto syntax;
 
       if (!gfc_merge_new_implicit (&ts))




FX