On Apr 5, 2006, at 8:14 PM, Charles O Nutter wrote:
- I'll be replacing Vector and Stack with ArrayList and
UnsynchronizedStack; we don't ever have more than one thread
accessing the given vars, and moving to unsynchronized collections
has given us a speed boost in many other areas.
OK, I used Stack to keep the code as simple possible since I figured
this wasn't core code likely to be executed a lot but I'll bear that
in mind in future.
Marc.
- The System.setOut stuff is definitely problematic, especially
considering we'd like this to be able to work across not only
multiple threads but across multiple JRuby VMs in the same JVM
process. The solution will probably involve providing a JRuby
runtime or JRuby thread-specific means of getting the "current
stdout", and making our change there. I don't have a fix at the
moment. The current working directory changes made recently are a
similar solution, since Java only has one CWD (" user.dir"
property) which we can't change willy-nilly. Long story short: your
change may stick for now with a big FIXME on it.
It's a good patch, and certainly cleaner and more complete than my
original hack. I'll see about cleaning it up and committing it.
Hopefully this will become a moot point someday if RCR 328 is
included in core Ruby. It would be far more preferable to implement
a Kernel#run_script for running scripts in a platform-specific way
than to have tricks like this in JRuby.
RCR 328: (appears to be down at the moment) http://www.rcrchive.net/
rcr/show/328
On 4/5/06, Charles O Nutter <[EMAIL PROTECTED]> wrote:
I saw it was a NetBeans patch, so I did what I'd been intending to
do for a while: make JRuby work as a project in NetBeans too. I
just committed project.xml, so it should be working correctly. I'm
pretty new to NetBeans, however, so if you (Marc) or someone else
more familiar could have a look, I'd appreciate it.
I should be able to apply and explore the patch now.
On 4/5/06, Charles O Nutter <[EMAIL PROTECTED] > wrote:
Ahh, one other detail: please use unified diff for your patches. It
came through ok this time, so I should be able to apply it anyway.
On 4/5/06, Marc Hadley <[EMAIL PROTECTED]> wrote: I tracked down
the shell wildcard problem with my original patch and
also tidied things up a bit and fixed a couple of additional bugs I
ran into during testing. With this patch Kernel#system and
Kernel#backquote now pass all the rubicon tests.
One change I made that could have future repercussions is to
temporarily redirect stdout to a buffer (see line 815 in the patched
file) while executing a second ruby script in-process (this is
required since you need to capture the output of the script executed
via Kernel#backtick). This works fine for a single threaded script
but might cause problems if multiple threads start executing
additional scripts concurrently. There doesn't appear to be an easy
fix to this but as I'm new to the code I might have missed something.
Marc.
---
Marc Hadley <marc.hadley at sun.com>
--
Charles Oliver Nutter @ headius.blogspot.com
JRuby Developer @ jruby.sourceforge.net
Application Architect @ www.ventera.com
--
Charles Oliver Nutter @ headius.blogspot.com
JRuby Developer @ jruby.sourceforge.net
Application Architect @ www.ventera.com
--
Charles Oliver Nutter @ headius.blogspot.com
JRuby Developer @ jruby.sourceforge.net
Application Architect @ www.ventera.com
---
Marc Hadley <marc.hadley at sun.com>
Business Alliances, CTO Office, Sun Microsystems.
-------------------------------------------------------
This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642
_______________________________________________
Jruby-devel mailing list
Jruby-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/jruby-devel
