On Tue, 4 Oct 2005, Joel E. Denny wrote:
On Mon, 26 Sep 2005, Joel E. Denny wrote:
On Tue, 20 Sep 2005, Joel E. Denny wrote:
In the event of memory exhaustion, if stack 0 happens to be marked for
deletion (guaranteeing that in a test case is not something I want to
think about), no user destructors are called on any symbols.
I created a test case.
<snip>
I created a patch using this approach.
I've only tested after applying my earlier `duplicated user destructor for
lookahead' patch, which I cut from the diff for this patch.
I just want to be sure the two patches and one test case mentioned above are
not forgotten. You can find them here:
http://lists.gnu.org/archive/html/bison-patches/2005-09/msg00078.html
http://lists.gnu.org/archive/html/bison-patches/2005-09/msg00109.html
In case it should help expedite these patches, I've retested them with the
most recent CVS sources for bison. Except for the line numbers in the
diffs, they should be fine. All test cases pass. Let me know if more
current diffs would help.
Joel
