You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@harmony.apache.org by Mikhail Loenko <ml...@gmail.com> on 2008/02/13 14:09:30 UTC
[drlvm] support gc_cc and original verifier
there is a number of bugs in gc_cc and not that many volunteers to fix them,
do we need to continue support it or we better move to archive?
the same for original verifier: a number of bugs found in both
verifiers were fixed in the default one (verifier-3363) only, should
we move the original one into archive?
Thanks,
Mikhail
Re: [drlvm] support gc_cc and original verifier
Posted by Ivan Volosyuk <iv...@gmail.com>.
Аминь!
On Fri, Feb 15, 2008 at 4:11 PM, Alexey Varlamov
<al...@gmail.com> wrote:
> Done: old verifier and gc_cc are moved out of trunk, gc_cc mode
> support is dropped (bti, drlvm build).
>
> --
> Alexey
>
> 2008/2/15, Gregory Shimansky <gs...@apache.org>:
>
>
> > Tim Ellison said the following on 15.02.2008 12:04:
> > > Alexey Varlamov wrote:
> > >> OK, given nobody objected - should we do it before M5?
> > >> As old verifier is never used and gc_cc have known problems and
> > >> limitations (e.g. no support for uncompressed references), I think we
> > >> should.
> > >
> > > Since these are not used by default, I agree with removing them from M5.
> >
> > +1
> > I see no problem with this.
> >
> > --
> > Gregory
> >
> >
>
Re: [drlvm] support gc_cc and original verifier
Posted by Alexey Varlamov <al...@gmail.com>.
Done: old verifier and gc_cc are moved out of trunk, gc_cc mode
support is dropped (bti, drlvm build).
--
Alexey
2008/2/15, Gregory Shimansky <gs...@apache.org>:
> Tim Ellison said the following on 15.02.2008 12:04:
> > Alexey Varlamov wrote:
> >> OK, given nobody objected - should we do it before M5?
> >> As old verifier is never used and gc_cc have known problems and
> >> limitations (e.g. no support for uncompressed references), I think we
> >> should.
> >
> > Since these are not used by default, I agree with removing them from M5.
>
> +1
> I see no problem with this.
>
> --
> Gregory
>
>
Re: [drlvm] support gc_cc and original verifier
Posted by Gregory Shimansky <gs...@apache.org>.
Tim Ellison said the following on 15.02.2008 12:04:
> Alexey Varlamov wrote:
>> OK, given nobody objected - should we do it before M5?
>> As old verifier is never used and gc_cc have known problems and
>> limitations (e.g. no support for uncompressed references), I think we
>> should.
>
> Since these are not used by default, I agree with removing them from M5.
+1
I see no problem with this.
--
Gregory
Re: [drlvm] support gc_cc and original verifier
Posted by Tim Ellison <t....@gmail.com>.
Alexey Varlamov wrote:
> OK, given nobody objected - should we do it before M5?
> As old verifier is never used and gc_cc have known problems and
> limitations (e.g. no support for uncompressed references), I think we
> should.
Since these are not used by default, I agree with removing them from M5.
Regards,
Tim
Re: [drlvm] support gc_cc and original verifier
Posted by Alexey Varlamov <al...@gmail.com>.
OK, given nobody objected - should we do it before M5?
As old verifier is never used and gc_cc have known problems and
limitations (e.g. no support for uncompressed references), I think we
should.
--
Alexey
2008/2/15, Pavel Rebriy <pa...@gmail.com>:
> +1 - old verifier is moving to archive
>
> On 13/02/2008, Alexei Fedotov <al...@gmail.com> wrote:
> >
> > I'm ok for the verifier. Pavel R? I would suggest adding a link from
> > Harmony web site to the original verifier referencing it as
> > demonstrating an approach for java verification based on subroutine
> > inlining. At the moment I wrote the code this was interesting for some
> > researches and might attract them to our project.
> >
> >
> > On Feb 13, 2008 4:09 PM, Mikhail Loenko <ml...@gmail.com> wrote:
> > > there is a number of bugs in gc_cc and not that many volunteers to fix
> > them,
> > > do we need to continue support it or we better move to archive?
> > >
> > > the same for original verifier: a number of bugs found in both
> > > verifiers were fixed in the default one (verifier-3363) only, should
> > > we move the original one into archive?
> > >
> > > Thanks,
> > > Mikhail
> > >
> >
> >
> >
> >
> > --
> > With best regards,
> >
> > Alexei
> >
>
>
>
> --
> Best regards,
> Pavel Rebriy
>
Re: [drlvm] support gc_cc and original verifier
Posted by Pavel Rebriy <pa...@gmail.com>.
+1 - old verifier is moving to archive
On 13/02/2008, Alexei Fedotov <al...@gmail.com> wrote:
>
> I'm ok for the verifier. Pavel R? I would suggest adding a link from
> Harmony web site to the original verifier referencing it as
> demonstrating an approach for java verification based on subroutine
> inlining. At the moment I wrote the code this was interesting for some
> researches and might attract them to our project.
>
>
> On Feb 13, 2008 4:09 PM, Mikhail Loenko <ml...@gmail.com> wrote:
> > there is a number of bugs in gc_cc and not that many volunteers to fix
> them,
> > do we need to continue support it or we better move to archive?
> >
> > the same for original verifier: a number of bugs found in both
> > verifiers were fixed in the default one (verifier-3363) only, should
> > we move the original one into archive?
> >
> > Thanks,
> > Mikhail
> >
>
>
>
>
> --
> With best regards,
>
> Alexei
>
--
Best regards,
Pavel Rebriy
Re: [drlvm] support gc_cc and original verifier
Posted by Alexei Fedotov <al...@gmail.com>.
I'm ok for the verifier. Pavel R? I would suggest adding a link from
Harmony web site to the original verifier referencing it as
demonstrating an approach for java verification based on subroutine
inlining. At the moment I wrote the code this was interesting for some
researches and might attract them to our project.
On Feb 13, 2008 4:09 PM, Mikhail Loenko <ml...@gmail.com> wrote:
> there is a number of bugs in gc_cc and not that many volunteers to fix them,
> do we need to continue support it or we better move to archive?
>
> the same for original verifier: a number of bugs found in both
> verifiers were fixed in the default one (verifier-3363) only, should
> we move the original one into archive?
>
> Thanks,
> Mikhail
>
--
With best regards,
Alexei
Re: [drlvm] support gc_cc and original verifier
Posted by Aleksey Shipilev <al...@gmail.com>.
+1 also
gc_gen has now equipped with semi-space GC algorithms so the last
strong advantage of gc_cc vs. gc_gen disappeared.
Thanks,
Aleksey.
On Feb 13, 2008 4:30 PM, Vladimir Beliaev <vl...@gmail.com> wrote:
> +1 also for both
>
> 2008/2/13, Pavel Pervov <pm...@gmail.com>:
>
> >
> > +1 for both.
> >
> > We now have stable gc_gen which is more advanced in GC algorithms than
> > gc_cc.
> > We now have stable verifier which is smaller and faster than old one.
> >
> > Pavel.
> >
> > On 2/13/08, Mikhail Loenko <ml...@gmail.com> wrote:
> > > there is a number of bugs in gc_cc and not that many volunteers to fix
> > them,
> > > do we need to continue support it or we better move to archive?
> > >
> > > the same for original verifier: a number of bugs found in both
> > > verifiers were fixed in the default one (verifier-3363) only, should
> > > we move the original one into archive?
> > >
> > > Thanks,
> > > Mikhail
> > >
> >
>
>
>
> --
> Vladimir Beliaev
> Intel Middleware Products Division
>
Re: [drlvm] support gc_cc and original verifier
Posted by Vladimir Beliaev <vl...@gmail.com>.
+1 also for both
2008/2/13, Pavel Pervov <pm...@gmail.com>:
>
> +1 for both.
>
> We now have stable gc_gen which is more advanced in GC algorithms than
> gc_cc.
> We now have stable verifier which is smaller and faster than old one.
>
> Pavel.
>
> On 2/13/08, Mikhail Loenko <ml...@gmail.com> wrote:
> > there is a number of bugs in gc_cc and not that many volunteers to fix
> them,
> > do we need to continue support it or we better move to archive?
> >
> > the same for original verifier: a number of bugs found in both
> > verifiers were fixed in the default one (verifier-3363) only, should
> > we move the original one into archive?
> >
> > Thanks,
> > Mikhail
> >
>
--
Vladimir Beliaev
Intel Middleware Products Division
Re: [drlvm] support gc_cc and original verifier
Posted by Pavel Pervov <pm...@gmail.com>.
+1 for both.
We now have stable gc_gen which is more advanced in GC algorithms than gc_cc.
We now have stable verifier which is smaller and faster than old one.
Pavel.
On 2/13/08, Mikhail Loenko <ml...@gmail.com> wrote:
> there is a number of bugs in gc_cc and not that many volunteers to fix them,
> do we need to continue support it or we better move to archive?
>
> the same for original verifier: a number of bugs found in both
> verifiers were fixed in the default one (verifier-3363) only, should
> we move the original one into archive?
>
> Thanks,
> Mikhail
>
Re: [drlvm] support gc_cc and original verifier
Posted by Ilya Berezhniuk <il...@gmail.com>.
+1 for gc_cc
Although gc_cc is quite small, it's not easy to support.
When I repaired gc_cc in Q4, I found that gc_cc does not fit well with
current threading model (there are potential problems with suspending
and buffers flushing).
Also it has no support for uncompressed references mode :)
Thanks,
Ilya.
2008/2/13, Mikhail Loenko <ml...@gmail.com>:
> there is a number of bugs in gc_cc and not that many volunteers to fix them,
> do we need to continue support it or we better move to archive?
>
> the same for original verifier: a number of bugs found in both
> verifiers were fixed in the default one (verifier-3363) only, should
> we move the original one into archive?
>
> Thanks,
> Mikhail
>