Re: [patch,doc] Fix intrinsic arguments names

[Tobias Burnus <burnus at net-b dot de>]

Hi FX,

some remarks; it is a bit chaotic since I first didn't see your texi
changes. I will have another later.

+    *vl = "values", *p1 = "path1", *p2 = "path2", *com = "command";
[...]
+  add_sym_2 ("link", GFC_ISYM_LINK, NO_CLASS, ACTUAL_NO, BT_INTEGER, di,
+	     GFC_STD_GNU, gfc_check_link, NULL, gfc_resolve_link,
+	     "path1", BT_CHARACTER, dc, REQUIRED,
+	     "path2", BT_CHARACTER, dc, REQUIRED);

Is there any reason behind that you add "p1" above and still use "path1" here?


+  add_sym_2 ("fstat", GFC_ISYM_FSTAT, NO_CLASS, ACTUAL_NO, BT_INTEGER, di,
+	     GFC_STD_GNU, gfc_check_fstat, NULL, gfc_resolve_fstat,
+	     ut, BT_INTEGER, di, REQUIRED, vl, BT_INTEGER, di, REQUIRED);

I don't mind vl = "values", but shouldn't then also invoke.texi be updated?
   CALL FSTAT(UNIT, BUFF [, STATUS])
... OK, I now found it, you change it there as well.
(g77 had SArray=, ifort has statb=. values= is OK.)


Similarly for lstat (and STAT):
+  add_sym_2 ("lstat", GFC_ISYM_LSTAT, NO_CLASS, ACTUAL_NO, BT_INTEGER, di,
+	     GFC_STD_GNU, gfc_check_stat, NULL, gfc_resolve_lstat,
+	     nm, BT_CHARACTER, dc, REQUIRED, vl, BT_INTEGER, di, REQUIRED);

  CALL LSTAT(FILE, BUFF [, STATUS])
... I just saw that you did the change to name= also in texi. For completeness:

g77 had:
       CALL LStat(File, SArray, Status)
       integer(kind=1) = LStat(File, SArray)
while ifort has:
       result = STAT (name,statb)



+  add_sym_1 ("umask", GFC_ISYM_UMASK, NO_CLASS, ACTUAL_NO, BT_INTEGER, di,
+	     GFC_STD_GNU, gfc_check_umask, NULL, gfc_resolve_umask,
+	     msk, BT_INTEGER, di, REQUIRED);

I might miss something, but I think
  call UMASK(MASK=a, OLD=o)
does not work as OLD= is still not recognized.


-	     a, BT_CHARACTER, dc, REQUIRED);
+	     "path", BT_CHARACTER, dc, REQUIRED);

Shouldn't be there a constant above like it was done for the others?
I don't mind "path" but it is a bit inconsistent.


Ditto:
-	      c, BT_INTEGER, 4, REQUIRED);
+	      "seed", BT_INTEGER, 4, REQUIRED);

Ditto:
-	      c, BT_CHARACTER, dc, REQUIRED);
+	      "string", BT_CHARACTER, dc, REQUIRED);
(Doesn't there exist stg already?)o

-	      c, BT_CHARACTER, dc, REQUIRED, st, BT_INTEGER, di, OPTIONAL);
+	      "path", BT_CHARACTER, dc, REQUIRED, st, BT_INTEGER, di, OPTIONAL);


-@code{DTIME(TARRAY, RESULT)}
+@code{DTIME(VALUES, TIME)}

I don't mind VALUES=, but I want to point out that both g77 and ifort have TARRAY=


Ditto ETIME, IDATE [such as g77; except that ifort has idate(i,j,k) or idate(iarray)],
LTIME.


Tobias