[jboss-dev] VFS 2.0.0.CR1

[Ales Justin]

> 1)The Windows file locking problem
> http://jira.jboss.com/jira/browse/JBAS-4310

This is solved in Marko's branch.
I had a talk with him about it yesterday, he said he still has some 
minor issues he wants to discuss (will post on the forum).

Scott, I remember you promising you'll have a look at it. ;-)
 >> Fantastic if it works! I'll pull it down into my jbossas workspace 
and try it out tonight.

He's also missing NoCopy feature, but said he'll look into it, how we 
currently do that, and how can be reused in his impl.

> 2) redeployment issues
> http://jira.jboss.com/jira/browse/JBAS-4220 - Hot deployment of jars 
> does not always work
> related to
> http://jira.jboss.com/jira/browse/JBVFS-7 - JarEntryHandler entryMap is 
> not being refreshed

I don't really see when people are modifying their jars @ runtime.
The workaround here is of course exploded dir. :-)

I wrote a test for this:
  - 
http://anonsvn.jboss.org/repos/jbossas/projects/vfs/trunk/src/test/java/org/jboss/test/virtual/test/JAREntryTestCase.java
Probably not the best test, but it exposes the problem.
In its current state (see TODO), it exposes the problem with caching 
(JBVSF-2).
To resolve JBVFS-7 I think we would need to implement a lot more:
  - http://www.jboss.com/index.html?module=bb&op=viewtopic&t=134200

> http://jira.jboss.com/jira/browse/JBVFS-2 - VFS caches old jar files

Adrian explained a few things here:
  - http://www.jboss.com/index.html?module=bb&op=viewtopic&t=134177&start=0

> There 2 more issues marked as critical:
> http://jira.jboss.com/jira/browse/JBVFS-11 - Validation of paths on case 
> insensative file systems

I can look at this, but dunno how to test it. :-)
Will Cygwin do the trick?

> http://jira.jboss.com/jira/browse/JBVFS-26 - optimization (nice to have)

I did this one today. ;-)
  - http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4152255#4152255
And Adrian added simple cache in classloading subproject.

But I still think we should do this more generic, see my yesterday's 
email to this list: 'improving AS boot time'

> There is also the question what we should do with Marko's alternative 
> implementation:
> http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4147669#4147669
> 
> In my view, unless it is a complete replacement of the existing 
> implementation that we can test ASAP, we can't rely on it at this point.

I agree.
If we don't have all what we currently have, it's still a step back, 
despite the obvious/cool locking fix.
But I'm sure Marko will surprise us again. ;-)