sablevm[-classpath] maintenance (was Re: GregorianCalendar Problem)

Chris Pickett <[email protected]> Sun, 15 Oct 2006 19:12:10 -0400
Newsgroups gmane.comp.java.vm.sablevm.general
Message-ID <[email protected]>
Fernando Lozano wrote:
> Hi Chris,
> 
>>> AFAIK this is a known classpath bug that was solved some time ago,
>>> but SableVM is not updated to
>>> current classpath releases and as they joined Apache Harmony it's
>>> unlikely they will ever update.
>> That appears to be true, the last version of Classpath imported into
>> sablevm-classpath is 0.20, from Jan. 13, 2006.
> 
> Has this resulted in a new SableVM release? As far as I know SableVM
> 1.13 is from December 2005 and so included either Classpath 0.18 or
> 0.19. :-(

$ svn log svn://svn.sablevm.org/sablevm/tags | head -n 4

------------------------------------------------------------------------
r3819 | egagnon | 2005-03-24 15:50:37 -0500 (Thu, 24 Mar 2005) | 2 lines

* Tagged sablevm-1.11.3.

$ svn log svn://svn.sablevm.org/sablevm-classpath/tags | head -n 4

------------------------------------------------------------------------
r3820 | egagnon | 2005-03-24 15:51:28 -0500 (Thu, 24 Mar 2005) | 2 lines

* Tagged sablevm-classpath-1.11.3.

However, the trunks of both sablevm and sablevm-classpath were updated 
much more recently and I think those suffering from bugs might want to 
at least try updating to a more recent version.  Whether or not this 
fixes the specific GregorianCalendar issue I don't know, that would 
require some digging in the Classpath repository or bugzilla.

svn://svn.sablevm.org/sablevm[-classpath]/trunk

> 
>> I guess the question here is, do we care about keeping Classpath
>> compatability while waiting for Harmony?
> 
> I vaguely remember reading on this list changes on the classpath / vm
> interface (for memory management and the like) make a little hard to
> import a new classpath, and some sablevm developers are pissed off
> because of this and don't want to work with classpath any more.

I don't think it's a question of being pissed off.  People just don't 
have either time for or interest in maintaining SableVM right now.  It's 
used a lot for research purposes (I can count about 10 research projects 
off the top of my head---Etienne's 5 plus another 5 at McGill, including 
my own) and in many cases that means Classpath fixes are irrelevant to us.

It may be the case that this alienates non-academic users, but if 
someone outside of UQAM or McGill wants to prepare an updated 
SableVM+Classpath release I think the contribution would be quite 
welcome.  In order to prepare such a release, Subversion write access to 
branches of both sablevm[-classpath] is available, as Etienne has mentioned.

Cheers,
Chris