[muscle] Re: 2.43?
- From: "Jeremy Friesner" <jaf@xxxxxxxxxxxx>
- To: muscle@xxxxxxxxxxxxx
- Date: Tue, 12 Aug 2003 23:24:59 PDT (-0700)
Hey David,
> When can we expect 2.43=3F
> What's new & what's retired...
Funny you should ask... just today I found a big fat efficiency-bug in the Queue
class (it made AddTail() and AddHead() into O(N) operations -- they should be
amortized O(1)), and so I was planning to release a v2.43 with the fix in a few
days (probably sometime this weekend). So if whomever is in charge of the
VC++ projects wants them included in the v2.43 release, the next day or two
would be a good time to send them to me. :^)
FWIW, here is my current change list for v2.43:
2.43 -
- Added an optional (count) argument to String::Append() and
String::Prepend().
- Added a DisconnectSession() method to the AbstractReflectSession class.
Calling this method force-disconnects the session's TCP connection.
- PR=5FCOMMAND=5FSETDATA now processes multiple sets to the same node in
a single Message.
- Added a (numPreallocSlots) argument to the Queue::EnsureSize() method,
so you can specify how many extra slots to preallocate in the event
of an array reallocation.
- Added SetSignalHandle(), SetSignalValue(), and SetReplyThreadID()
methods to the Win32MessageTransceiverThread class.
* Fixed a bug in Hashtable::Clear() that could cause crashes under
certain circumstances (iterators weren't being unregistered reliably).
* Fixed a bug in the Queue class that was disabling queue-slot-preallocation,
making AddHead() and AddTail() calls VERY inefficient -- O(N) instead
of amortized O(1). Oops!
* The Win32 implementation of GetHumanReadableTimeString() was returning
date strings 369 years too early. Fixed.
-Jeremy
Other related posts:
