[Sbcl-commits] master: Add comment

[Douglas K. <sn...@us...>]

The branch "master" has been updated in SBCL:
       via  65474ed6895f75201ba8ea99fb83423db21e41b3 (commit)
      from  66fc1e03cb3508f4b099e63eff76ccf1d5a3943e (commit)

- Log -----------------------------------------------------------------
commit 65474ed6895f75201ba8ea99fb83423db21e41b3
Author: Douglas Katzman <do...@go...>
Date:   Mon Dec 15 16:46:35 2014 -0500

    Add comment
---
 src/code/defstruct.lisp |    9 +++++++++
 1 files changed, 9 insertions(+), 0 deletions(-)

diff --git a/src/code/defstruct.lisp b/src/code/defstruct.lisp
index 9390b05..c01cf5c 100644
--- a/src/code/defstruct.lisp
+++ b/src/code/defstruct.lisp
@@ -1251,6 +1251,15 @@ unless :NAMED is also specified.")))
   ;; the object as indicated in the header, so the pad word needs to be
   ;; included in that length to guarantee proper alignment of raw double float
   ;; slots, necessary for (at least) the SPARC backend.
+  ;; On backends with interleaved raw slots, the convention of having the
+  ;; header possibly "lie" about an extra word is more of a bug than a feature.
+  ;; Because the structure base is aligned, double-word raw slots are properly
+  ;; aligned, and won't change alignment in descendant object types. It would
+  ;; be correct to store the true instance length even though GC preserves
+  ;; the extra data word (as it does for odd-length SIMPLE-VECTOR), treating
+  ;; the total physical length as rounded-to-even. But having two different
+  ;; conventions would be even more unnecessarily confusing, so we use
+  ;; the not-sensible convention even when it does not make sense.
   (logior (dd-layout-length dd) 1))
 
 (defun dd-bitmap (dd)

-----------------------------------------------------------------------


hooks/post-receive
-- 
SBCL