Re: [Emacs-diffs] trunk r117464: Shrink Lisp_Sub_Char_Table by preferring C integers to Lisp_Objects.

[Stefan Monnier]

>> But I don't see why that necessarily implies casting
>> Lisp_Sub_Char_Table to a Lisp_Vector (I ask this question without
>> having spent much time looking for the answer in the source, maybe the
>> answer is obvious, but I think if you can tell me why it's a better use
>> of my time than trying to figure it out myself).

> Hm... look at mark_char_table in alloc.c - this function implies that
> both Lisp_Char_Table and Lisp_Sub_Char_Table pointers can be treated
> as a pointer to their base type Lisp_Vector. Having both mark_char_table
> and mark_sub_char_table looks a bit overengineered for me. This also
> applies to Lisp_Char_Table/Lisp_Sub_Char_Table printing code in
> print_object, etc.

Could you try to use something like the patch below to eliminate this
alignment assumption?


        Stefan


=== modified file 'src/alloc.c'
--- src/alloc.c 2014-07-02 03:26:19 +0000
+++ src/alloc.c 2014-07-03 16:47:25 +0000
@@ -5962,13 +5962,18 @@
 {
   int size = ptr->header.size & PSEUDOVECTOR_SIZE_MASK;
   /* Consult the Lisp_Sub_Char_Table layout before changing this.  */
-  int i, idx = (pvectype == PVEC_SUB_CHAR_TABLE ? SUB_CHAR_TABLE_OFFSET : 0);
+  int i;
+  Lisp_Object *contents
+    = pvectype == PVEC_SUB_CHAR_TABLE
+    ? (size -= SUB_CHAR_TABLE_OFFSET,
+       ((struct Lisp_Sub_Char_Table *)ptr)->contents)
+    : ((struct Lisp_Char_Table *)ptr)->contents;
 
   eassert (!VECTOR_MARKED_P (ptr));
   VECTOR_MARK (ptr);
-  for (i = idx; i < size; i++)
+  for (i = 0; i < size; i++)
     {
-      Lisp_Object val = ptr->contents[i];
+      Lisp_Object val = contents[i];
 
       if (INTEGERP (val) || (SYMBOLP (val) && XSYMBOL (val)->gcmarkbit))
        continue;