Re: Patch: implement new dynamic varobj spec

[Tom Tromey <tromey@redhat.com>]

>>>>> "Tom" == Tom Tromey <tromey@redhat.com> writes:

Tom> Oh, odd.  But not a bug, really, as the order is not specified.

Nick> It might not be a bug but it's desirable to list them in order as
Nick> that makes it easier to insert them into the existing list.

Tom> Ok, I will look into this.

Could you try the appended patch?
If it works for you I will roll it into the main patch.

Tom

diff --git a/gdb/varobj.c b/gdb/varobj.c
index e7fb589..7235635 100644
--- a/gdb/varobj.c
+++ b/gdb/varobj.c
@@ -1759,7 +1759,6 @@ VEC(varobj_update_result) *varobj_update (struct varobj **varp, int explicit)
 	{
 	  VEC (varobj_p) *changed = 0, *new = 0, *unchanged = 0;
 	  int i, children_changed;
-	  varobj_p tmp;
 
 	  if (v->frozen)
 	    continue;
@@ -1799,15 +1798,21 @@ VEC(varobj_update_result) *varobj_update (struct varobj **varp, int explicit)
 		  r.children_changed = 1;
 		  r.new = new;
 		}
-	      for (i = 0; VEC_iterate (varobj_p, changed, i, tmp); ++i)
+	      /* Push in reverse order so that the first child is
+		 popped from the work stack first, and so will be
+		 added to result first.  This does not affect
+		 correctness, just "nicer".  */
+	      for (i = VEC_length (varobj_p, changed) - 1; i >= 0; --i)
 		{
+		  varobj_p tmp = VEC_index (varobj_p, changed, i);
 		  varobj_update_result r = {tmp};
 		  r.changed = 1;
 		  r.value_installed = 1;
 		  VEC_safe_push (varobj_update_result, stack, &r);
 		}
-	      for (i = 0; VEC_iterate (varobj_p, unchanged, i, tmp); ++i)
+	      for (i = VEC_length (varobj_p, unchanged) - 1; i >= 0; --i)
 	      	{
+		  varobj_p tmp = VEC_index (varobj_p, unchanged, i);
 	      	  if (!tmp->frozen)
 	      	    {
 	      	      varobj_update_result r = {tmp};