Friday works for me...Perhaps if there is not a lot of opposition
we can set up a Friday evening time/location for a meet up?
I will update the web site and we can both blog the info.
Thirsty bear is close, but almost always reserved by other conference types.
Anyone know of a fairly close place?
Oops, looks like an autocomplete fumble, please ignore!
On 4/21/06, Charles O Nutter <[EMAIL PROTECTED]> wrote:
> Huzzah! Small, medium, and large varieties!
>
> --
> Charles Oliver Nutter @ headius.blogspot.com
> JRuby Developer @ jruby.sourceforge.net
> Application Architect @ www.ventera.com
>
It appears that the issue restoring a session is a problem with our marshalling implementation. The following code fails to work:class MyHash < Hash def foo; endendx = MyHash.newMarshal.load
(Marshal.dump(x)).fooThe MyHash is getting marshalled and unmarshalled as though it were a simple hash. The
There are plenty of 1.8 methods that have never been implemented, and many of them are not tested by Rubicon since it was originally written for 1.6. The lack of a spec or a complete set of unit tests continues to complicate our work.
I'll have a look at the patch and try to get it committed.On 4/5
On 4/5/06, Nick Sieger <[EMAIL PROTECTED]> wrote:
It seems like there are occasionally methods in the 1.8 API that still aren't implemented in JRuby. String#insert is one of those I ran into while doing my work on ActiveRecord.Attached is a patch to add it.
Eeek, I suppose it would help if I attac
I added a couple additional quick interpreter optimizations this morning. After running some numbers through, I saw that we're creating roughly 170 times as many InstructionBundle objects as there are unique AST nodes in the system. I knew this was wasteful when I wrote it, but I figured now is a g
I'm getting actual work done today, what with the SF.net CVS being down for over 12 hours now. At any rate, inbetween real work I ran another benchmark:nestedloop 15with fix: 55.422, 55.094; 55.258, 19% faster
without: 68.781, 68.89; 68.8355On 3/30/06, Charles O Nutter <[EMAIL PROTECTED]> wrote:
In
Another "What If" experiment, inlining a number of methods from EvaluationState, produced similar results; no more than .5% speed increase.On 3/28/06,
Charles O Nutter <[EMAIL PROTECTED]> wrote:
Another experiment to eliminate or narrow causes of performance trouble. In this case, IRubyObject was
There have been various changes since I last updated this patch, but it should be mostly applicable. As I mentioned, it does work, but it's certainly not as solid as I'd like.See also RCR 328
On 3/27/06, Marc Hadley <[EMAIL PROTECTED]> wrote:
On Mar 23, 2006, at 12:09 PM, Charles O Nutter wrote:> T
RCR 328: http://www.rcrchive.net/rcr/show/328On 3/27/06, Charles O Nutter <
[EMAIL PROTECTED]> wrote:There have been various changes since I last updated this patch, but it should be mostly applicable. As I mentioned, it does work, but it's certainly not as solid as I'd like.
See also RCR 328
On 3/
On Mar 23, 2006, at 12:09 PM, Charles O Nutter wrote:
That sounds good to me. Just let me/us know if you want that
Kernel#system patch (it's ugly but it works) or if you have any
questions about JRuby internals. It's still a bit of a maze, but
it's improving rapidly.
If you send me a copy
Well, actually this was more about getting RubyGems working better for
the release. (See Charles mail further down).
/O
- Original Message -
From: Thomas E Enebo <[EMAIL PROTECTED]>
Date: Monday, March 27, 2006 6:27 pm
Subject: Re: [Jruby-devel] Re: PATCH: Various fixes to make Ru
Charles O Nutter <[EMAIL PROTECTED]>
> > Date: Monday, March 27, 2006 5:27 pm
> > Subject: Re: [Jruby-devel] Re: PATCH: Various fixes to make RubyGems
> > work, almost
> > To: [email protected]
> >
> > > That seems fair. I can onl
>
> - Original Message -
> From: Charles O Nutter <[EMAIL PROTECTED]>
> Date: Monday, March 27, 2006 5:27 pm
> Subject: Re: [Jruby-devel] Re: PATCH: Various fixes to make RubyGems
> work, almost
> To: [email protected]
>
> > Th
I will check it out. When is the release "tonight" in GMT? ... I'm quite
a few timezones away, I guess.
/O
- Original Message -
From: Charles O Nutter <[EMAIL PROTECTED]>
Date: Monday, March 27, 2006 5:27 pm
Subject: Re: [Jruby-devel] Re: PATCH: Various fixes
That seems fair. I can only assume this check was put in there to prevent a more painful failure further on in the parser, but the error message is going to be pretty straightforward in any case. Perhaps we'll add the respond_to checks, and perhaps not.
I'm more concerned about the fileutils.rb twe
At 20:45 2006-03-26, you wrote:
Agreed on all points. The two hacks for IO=== and rescue returning the
exception are, I believe easy enough fixes. The latter is an interpreter
issue, so I'll tackle that, and I think the IO=== thing is just a matter
of findign an appropriate way to define it. On
The changes are committed. Happy day!The FileUtils hack is still necessary to get RubyGems to run, so we'll need a better solution for that, and of course there's still a few bugs with setup.rb and the corrupting of the sources gem...but we're SO CLOSE :)
On 3/26/06, Charles O Nutter <[EMAIL PROTEC
The hacks are resolved, and there's one additional fix:- GzipReadline.gets now properly uses the separator param. The impl is slow (character by character InputStream.read calls), but it works.I'll be committing shortly.
On 3/26/06, Charles O Nutter <[EMAIL PROTECTED]> wrote:
Agreed on all points.
Agreed on all points. The two hacks for IO=== and rescue returning the exception are, I believe easy enough fixes. The latter is an interpreter issue, so I'll tackle that, and I think the IO=== thing is just a matter of findign an appropriate way to define it. Once I have those modified and test ca
Great! I could tell yesterday when I was playing that gems was getting
close. Some of these could be landed right away and others not. My
guidelines:
1. Can it be tested in a unit test easily:
a. Both <=> with nil fit this category
b. String.oct
2. It is not a hack. I think the hacks are
Rubicon looks pretty good.On 3/26/06, Charles O Nutter <[EMAIL PROTECTED]> wrote:
The attached fixes make RubyGems setup.rb complete succesfully...but there's a catch.During the install of the sources-0.0.1.gem, two odd things seem to be happening:- sources-0.0.1.gem
is modified, and corrupted
- t
That sounds good to me. Just let me/us know if you want that Kernel#system patch (it's ugly but it works) or if you have any questions about JRuby internals. It's still a bit of a maze, but it's improving rapidly.
On 3/23/06, Marc Hadley <[EMAIL PROTECTED]> wrote:
>> Hello again Marc! I assume you
Hello again Marc! I assume you were finally able to get into
SourceForge's
flaky anon CVS server. Kudos!
Thanks, persistence paid off in the end !
- Many ruby scripts launch other scripts using Kernel#system. For
us, this
is very bad since it tries to launch another JVM. I have previously
Yes. That is true. Hopefully they agree so it can just be code in
our repo
-Tom
On Tue, 21 Mar 2006, Marc Hadley defenestrated me:
> On Mar 20, 2006, at 5:06 PM, Marc Hadley wrote:
> >
> >I found a GPL/LGPL version at: http://manticore.2y.net/Java/
> >examples/crypt.zip
> >
> It occurred
On Mar 20, 2006, at 5:06 PM, Marc Hadley wrote:
I found a GPL/LGPL version at: http://manticore.2y.net/Java/
examples/crypt.zip
It occurred to me that even if the authors won't agree to a CPL
license, the library could still be distributed and used with JRuby
under the terms of the LGPL.
Either way, thanks for finding this Marc! Perhaps we'll finally have crypt implemented now.On 3/20/06, Thomas E Enebo <
[EMAIL PROTECTED]> wrote:On Mon, 20 Mar 2006, Marc Hadley defenestrated me:
> On Mar 20, 2006, at 3:59 PM, jruby-devel-> [EMAIL PROTECTED] wrote:> >> >Send a patch to the mailing
On Mon, 20 Mar 2006, Marc Hadley defenestrated me:
> On Mar 20, 2006, at 3:59 PM, jruby-devel-
> [EMAIL PROTECTED] wrote:
> >
> >Send a patch to the mailing list and I will apply it. The biggest
> >problem with that bug was finding one that could be distributed
> >CPl/GPL/LGPL. If you found a p
On Mar 20, 2006, at 3:59 PM, jruby-devel-
[EMAIL PROTECTED] wrote:
Send a patch to the mailing list and I will apply it. The biggest
problem with that bug was finding one that could be distributed
CPl/GPL/LGPL. If you found a public domain one/license-friendly
one or
some hook in Java (that
Here's a simple script to compare Ruby and JRuby methods for various types:thing_to_compare = ARGV[0]method_type = ARGV[1]puts "Comparing JRuby #{thing_to_compare} #{method_type} methods to Ruby"
ruby_methods = eval `ruby -e "p #{thing_to_compare}.#{method_type}_methods.sort`jruby_methods = eval "#
I think it comes down to this: module methods should show up in the public_instance_methods of the module.C:\rails>jruby -e "p ::Kernel.public_instance_methods.sort"[]C:\rails>ruby -e "p ::
Kernel.public_instance_methods.sort"["==", "===", "=~", "__id__", "__send__", "class", "clone", "display", "d
The stack trace associated with the issue. This is the same thing I saw before, and I have debugged through in the past to find that this stack should not be empty.Complete Java stackTracejava.util.EmptyStackException
at org.jruby.util.collections.AbstractStack.pop(AbstractStack.java:59)
Actually, someone in a sh-compatible environment will need to clean up
the "jirb" script so it works with JRuby. It's obviously not quite
right, currently, since it still looks to run C Ruby directly. Oh
well, at least I'm set, and that's what really matters, right?
On 3/2/06, Charles O Nutter <[E
Here's an updated patch; I think some Rubygems stuff leaked into the other one.
On 2/19/06, Charles O Nutter <[EMAIL PROTECTED]> wrote:
> This is a somewhat less naive patch than my previous attempt, and this
> one appears to work well enough for Rails to continue. It basically
> just looks throug
Crisis averted! I added back in the single-digit pattern and the issue
is gone again. I hadn't noticed that got removed (or never added) to
the translator.
I also removed the "three digit" octal pattern. David, can you explain
that one? It didn't seem to be doing anything the "missing zero"
patter
Well this is definitely a bug, and I think it's related to the parser.
Clues:
- The parser determines dvar versus lvar at parse time, using
ParserSupport's localNamesStack. As blocks are entered and exited, the
topmost LocalNamesElement is modified appropriately. When a bare
identifier is encount
I just found this:
http://java-readline.sourceforge.net/
Apparently Jython uses it for terminal control stuff..
- Charlie
On 2/9/06, Charles O Nutter <[EMAIL PROTECTED]> wrote:
> It's become increasingly irritating that there's nothing equivalent to
> IRB that works with JRuby on the command-li
There's a standing question whether the parentModule should be
maintained in RubyModule. So far, eliminating it does not appear to
have broken anything, but I see where it could become confused.
parentModule was used incorrectly for finding constants, but it was
perhaps correctly used for Module#ge
Sorry, that's \[1-7][0-7]+
It *may* be possible to further narrow it to \[1-7][0-7][0-7]+ if Ruby
requires escaped octals to have at least three digits...which it may.
- Charlie
On 1/25/06, Charles O Nutter <[EMAIL PROTECTED]> wrote:
> This fix adds a zero to everything that matches \[1-7][1-7]+
Stephenson9Samuel
I can accept that argument. I'm also more keen on consistency than on
evangelizing one way or the other. Since we do it this way throughout
most of JRuby, so shall it continue to be done.
On 1/24/06, Thomas E Enebo <[EMAIL PROTECTED]> wrote:
> My early return style is based on the general patter
My early return style is based on the general pattern:
if (something_unusual1)
return error
if (something_else_unusual)
return error
return success
In thinking this way it is hard to argue that return success is
a side-effect. If it is not an error and success thing, then the
argument
I'll explain why I don't like either version and then move on..
With your two ways of doing things, it is much more complicated to do
something other than return in the future. Lets say that some day,
instead of doing a return we want to call another method (or some
other side effect). Using my ve
http://www.headius.com/jruby-200601231904.zip
On 1/23/06, robert kuzelj <[EMAIL PROTECTED]> wrote:
> Charles O Nutter wrote:
> > After talking with Tom we agreed that since external thread handling
> > is severely broken right now, this patch can only be a net gain. It
> > also does allow Robert's
Charles O Nutter wrote:
After talking with Tom we agreed that since external thread handling
is severely broken right now, this patch can only be a net gain. It
also does allow Robert's test to run successfully every time, so I
think that's very promising.
The test is also committed, but not par
> nit:
>
> if (a)
> return
> else if (b)
> return
> else
> return
>
> Should lop off the else:
>
> if (a)
> return
> else if (b)
> return
>
> return
Should lop off ALL the elses.
if (a)
return
if (b)
return
return
>
> Other than that it looks good...
>
> -Tom
No problem, will do that and commit.
- Charlie
>
> nit:
>
> if (a)
> return
> else if (b)
> return
> else
> return
>
> Should lop off the else:
>
> if (a)
> return
> else if (b)
> return
>
> return
>
---
This SF.net email is sponsore
On Sun, 22 Jan 2006, Charles O Nutter defenestrated me:
> I hate when I do that. Patch attached.
>
> On 1/22/06, Charles O Nutter <[EMAIL PROTECTED]> wrote:
> > This fix allows the module/class comparison operators to work like
> > they're supposed to: returning nil when two compared classes are
The trick to learn (which I'm still working on), is to not put the addressee
in when you start the message :)
On Sunday 22 January 2006 23:49, Charles O Nutter wrote:
> I hate when I do that. Patch attached.
>
> On 1/22/06, Charles O Nutter <[EMAIL PROTECTED]> wrote:
> > This fix allows the m
Grr. It introduced some other problem resulting in a NPE in
ConstNodeVisitor (strikingly similar to the JVM issue I ran into, but
does not go away with any combination of JVM or interpreted mode).
Back to the drawing board.
- Charliie
On 1/22/06, Charles O Nutter <[EMAIL PROTECTED]> wrote:
> I'm
I'm going with the binding2 patch on my branch. It doesn't add 100%
support for bindings, but it works for the most obvious ones. This
eliminates a couple failures in CRuby's test_env.
On 1/22/06, Charles O Nutter <[EMAIL PROTECTED]> wrote:
> A version cloning all of ThreadContext. Same results in
I hate when I do that. Patch attached.
On 1/22/06, Charles O Nutter <[EMAIL PROTECTED]> wrote:
> This fix allows the module/class comparison operators to work like
> they're supposed to: returning nil when two compared classes are on
> different branches of the hierarchy. I found this issue runnin
FYI, the correct output from that should be (I believe) what follows.
See http://ruby-talk.org/cgi-bin/scat.rb/ruby/ruby-dev/24229
[:B, :Ai, :B]
[:B, :Ai, :B]
[:B, :Ac, :B]
[:B, :Ac, :B]
[:B, :Ac, :B]
[:A, :Ac, :A]
On 1/22/06, Charles O Nutter <[EMAIL PROTECTED]> wrote:
> Given getSingletonClass
A version cloning all of ThreadContext. Same results in test_eval in C
Ruby source...one additional success, but still other failures. Seems
to have something to do with how we scope/frame lambdas.
- Charlie
On 1/22/06, Charles O Nutter <[EMAIL PROTECTED]> wrote:
> A few fixes. I added a couple t
A few fixes. I added a couple tests to Rubicon for this (TestBinding
was empty before) and found some impl issues.
- Charlie
On 1/22/06, Charles O Nutter <[EMAIL PROTECTED]> wrote:
> Here's a quick patch to add RubyBinding and make RubyKernel.eval
> (actually RubyObject.eval) work correctly with
Mccormick Zelma
It's worse than I thought. I confirmed that the code runs correctly
under 1.5.0_05, and it does not run correctly under 1.4.2_06 or
1.4.2_10 unless run in interpreted mode. It appears there's a problem
with HotSpot under 1.4.2 that was never fixed and which is causing
this issue.
I'm a bit burned
57 matches
Mail list logo
