[E-Lang] Announcing 0.8.9.1: The Bug Fix Branch Release
Mark S. Miller
markm@caplet.com
Fri, 27 Apr 2001 09:40:03 -0700
By popular demand, because 0.8.10 isn't out yet, I've applied some of the
bug fixes since 0.8.9 to create a side branch called 0.8.9.1 (actually,
0.8.9.1a). Get it at http://www.erights.org/download/0-8-9-1/index.html .
>From the "Highlights" section:
E 0.8.9.1 is a branch off of 0.8.9, in which many of the bugs fixed in the
internal development successors of 0.8.9 have been applied to 0.8.9 itself.
This allows production work in E to benefit from these bug fixes while we
all await the release of 0.8.10.
The main fixes are
* Bill's fixes to VatTP (aka net/data) to enable it to run without the cryptix
jars & properties files, and therefore to be able to run on Java 1.3. 0.8.9
itself could only run on JDK >= 1.1.7 & JDK < 1.3. This release should run
on all JDKs >= 1.2. This will be the last production release to support any
JDKs < 1.3; 0.8.10 will only support JDKs >= 1.3.
* Bill's further fixes to the random number generator so that it remains
unguessable even when the native microsecond timer code is absent (which,
currently, is always).
* The fixes to Runner to repair a deadlock bug diagnosed by Dean, and to fix a
bug that would have impaired multiple vats per JVM, noticed by Chris
Hibbert. Further fixes to remove the notion of a "global" default Runner,
with much help from Dean.
* The previous installer would crash on a null pointer if there wasn't a
java.exe on the PATH. This is now fixed. If this installer still crashes on
you, please let me know.
* Several improvements to the Makefiles and the C code for the installer that
should make them more portable.
* The 0.8.9.1 grammar accepts the "var" keyword, but treats it as a synonym
for define. Since "var" was one of the few non-downward compatible changes
to the E grammar that happened since 0.8.9, this change should allow most
normal 0.8.10 compatible code to run on 0.8.9.1.
Cheers,
--MarkM