diff options
author | Steve Huston <shuston@riverace.com> | 1998-04-20 20:43:07 +0000 |
---|---|---|
committer | Steve Huston <shuston@riverace.com> | 1998-04-20 20:43:07 +0000 |
commit | 878496b1f7d89bf13d3ce3b25f234354efcbec56 (patch) | |
tree | fc53f1fade6d873f22b9e659f2b0ac5176293c24 | |
parent | 8ea84076a35728eaafe6f481630b7c76ac0654fd (diff) | |
download | ATCD-878496b1f7d89bf13d3ce3b25f234354efcbec56.tar.gz |
*** empty log message ***
-rw-r--r-- | ChangeLog-98a | 15 |
1 files changed, 15 insertions, 0 deletions
diff --git a/ChangeLog-98a b/ChangeLog-98a index 7de386317e9..4b6bd92c344 100644 --- a/ChangeLog-98a +++ b/ChangeLog-98a @@ -1,3 +1,18 @@ +Mon Apr 20 15:39:42 1998 Steve Huston <shuston@riverace.com> + + * ace/Handle_Set.cpp: Access fds_bits elements using an fd_mask + pointer rather than array indexing of the fds_bits array itself. + HP-UX 11 plays some games with the fd_mask type - fd_mask is defined + as an int32_t, but the fds_bits is an array of longs. This makes + plainly indexing through the array by hand tricky, since the FD_* + macros treat the array as int32_t. So the bits are in the right + place for int32_t, even though the array is long. This, they say, + is to preserve the same in-memory layout for 32-bit and 64-bit + processes. So, we play the same game as the FD_* macros to get + the bits right. See /usr/include/sys/_fd_macros.h. + On all other systems, this amounts to practically a NOP, since this + is what would have been done anyway, without all this type jazz. + Sat Apr 18 13:59:55 1998 David L. Levine <levine@cs.wustl.edu> * ace/Naming_Context.h: removed trailing ';' at end of |