I have found the following class useful for something similar, don't
know if it will be of help:
/** A {@link MutableCallSite} that can be reset to its initial target. */
public static class ResetableCallSite extends MutableCallSite {
private MethodHandle reset = null;
/**
* Make
Thanks Howard, this looks applicable as well.
I had not thought of this approach so it is helpful
mark___
mlvm-dev mailing list
mlvm-dev@openjdk.java.net
http://mail.openjdk.java.net/mailman/listinfo/mlvm-dev
Another couple of nit picky Javadoc comments:
1. Constructor MutableCallSite(MethodType) doesn't describe its argument.
2. Constructor MutableCallSite(MethodHandle) says "Make a blank call
site object, possibly equipped with an initial target method handle.",
but a non-null target ius required.
-
The changed makefiles neglect to add the java/lang/invoke classes to
jre/lib/ct.sym.
Anybody know offhand how to fix this? I had to do this once for java/dyn, but
I've forgotten how.
-- John
On Mar 5, 2011, at 2:45 PM, Stephen Bannasch wrote:
>>> Here's what my build script reports when I st
On Mar 7, 2011, at 12:17 AM, John Rose wrote:
> Anybody know offhand how to fix this? I had to do this once for java/dyn,
> but I've forgotten how.
Never mind: make/docs/CORE_PKGS.gmk
-- John
___
mlvm-dev mailing list
mlvm-dev@openjdk.java.net
http:
Le 07/03/2011 07:36, Attila Szegedi a écrit :
> If a SwitchPoint is invalidated, will all method handles created with its
> guardWithTest release the reference to "target" (since they no longer need
> it) and only keep the reference to "fallback", so in theory, "target" can be
> GCed?
>
> Atti
Changeset: 7539d9a8e017
Author:jrose
Date: 2011-03-07 01:13 -0800
URL: http://hg.openjdk.java.net/mlvm/mlvm/hotspot/rev/7539d9a8e017
meth-rename: fix config errors
! meth-rename-7012648.patch
___
mlvm-dev mailing list
mlvm-dev@openjdk.ja
Changeset: c7b1449d3379
Author:jrose
Date: 2011-03-07 01:13 -0800
URL: http://hg.openjdk.java.net/mlvm/mlvm/jdk/rev/c7b1449d3379
meth-rename: fix config errors
! meth-rename-7012648.patch
___
mlvm-dev mailing list
mlvm-dev@openjdk.java.n
Hi,
I got the following when trying to use CallSites on Stephen Bannasch's
latest build (which isn't the latest latest build and therefore this
issue might have been found already):
# To suppress the following error report, specify this argument
# after -XX: or in .hotspotrc: SuppressErrorAt=/re
I just pushed some tweaks to the meth-rename patches.
The jtreg tests now pass for me on the built jdk.
The next step is to have someone reproduce this good state of affairs.
-- John
___
mlvm-dev mailing list
mlvm-dev@openjdk.java.net
http://mail.openj
Hi all,
John, I really like the first part of the proposal even if I think that
the syntax can be tweaked a little bit :)
I don't see a need for the second part (bootstrap abstraction).
Testing a VM implementation without having a way to encode easily
invokedynamic
is really silly and I hope
At 1:15 AM -0800 3/7/11, John Rose wrote:
>I just pushed some tweaks to the meth-rename patches.
>
>The jtreg tests now pass for me on the built jdk.
>
>The next step is to have someone reproduce this good state of affairs.
The ClassValueTest and MethodTypeTest tests started passing but the others
At 10:41 AM + 3/2/11, lukas.stad...@jku.at wrote:
>Changeset: e16f8f75573f
>Author:Lukas Stadler
>Date: 2011-03-02 11:35 +0100
>URL: http://hg.openjdk.java.net/mlvm/mlvm/jdk/rev/e16f8f75573f
>
>coro: add jtreg tags to coro unit tests
>
>! coro.patch
With the latest changes the
That looks like an old java is being used. Try manually refreshing the
langtools repo to the latest revision.
Will rebase to bsd-port today.
-- John (on my iPhone)
On Mar 7, 2011, at 6:24 AM, Stephen Bannasch
wrote:
> At 1:15 AM -0800 3/7/11, John Rose wrote:
>> I just pushed some tweaks
At 9:38 AM -0800 3/7/11, John Rose wrote:
>That looks like an old java is being used. Try manually refreshing the
>langtools repo to the latestrevision.
langtools looks up-to-date to me -- after popping the patch queue a manual pull
adds nothing new.
also fyi: right now folks can't build the bs
On Mar 7, 2011, at 11:34 AM, Stephen Bannasch wrote:
> langtools looks up-to-date to me -- after popping the patch queue a manual
> pull adds nothing new.
Odd. When you get an error about ([Object)Object not matching some other
signature, it's usually because the compiled call sites of invokeE
At 11:46 AM -0800 3/7/11, John Rose wrote:
>On Mar 7, 2011, at 11:34 AM, Stephen Bannasch wrote:
>
>>langtools looks up-to-date to me -- after popping the patch queue a manual
>>pull adds nothing new.
>>
>
>Odd. When you get an error about ([Object)Object not matching some other
>signature, it's
17 matches
Mail list logo
