[Vm-dev] Re: Marking the Squeak 5.1 release

Clément Bera bera.clement at gmail.com
Tue Sep 27 08:31:50 UTC 2016


On Mon, Sep 26, 2016 at 8:50 PM, Eliot Miranda <eliot.miranda at gmail.com>
wrote:

>
> Hi Fabio,
>
> On Sun, Sep 25, 2016 at 12:10 PM, Fabio Niephaus <lists at fniephaus.com>
> wrote:
>
>> Hi Eliot,
>>
>> I've created a GitHub release for 201608171728 (see [1]) which is
>> considered to be stable. Feel free to add a changelog if you have some
>> time. You can find all commits after r3732 at [2].
>>
>
> Thanks, will do.
>
>
>> Is there any particular milestone for the next release?
>>
>
> Looking through my emails I see the following for Squeak 6.0:
>
> "For 6.0 I'm hoping we will do
>
> - the Sista bytecode set (with backwards compatibility support for the
> existing set).  This gives us 32k literals,
> - FullBlockClosure, which means I depended method objects for blocks, and
> hence (slightly) faster block activation
> - read-only objects and hence read-only literals
>
> These three provide the necessary support for the image-level Scorch
> adaptive optimiser.  This will be demonstrated at ESUG in Prague and
> productized by next year"
>
> Clément and I have already put the byte code set and full block support
> into the latest 32-bit Spur VM source.  I should remember to put these into
> the 64-bit ones too.  I think Clément has added read-only-object support
> to the Pharo VM.  The Squeak VM will follow soon.
>
> So yes, these three, Sista Bytecode Set as secondary byte code set,
> FullBlockClosure support and read-only-object support are the major items
> for the next release.
>
>
>

Hi...

Some details.

I have integrated these 3 features, at the exception of the read-only
literals, in the Pharo image. I need to check to make sure all the features
are enabled in the Pharo VM.

For read-only objects I have some SUnit tests proving that it works, but it
needs further testing. The next step is read-only literals, which leads to
the question, will compiled method and compiled block be read-only ? I
think they should.

For FullBlockClosure and the new bytecode set, I have been using them for
weeks without any issues. I implemented the FullBlockClosure support in the
Opal compiler, one needs to implement something similar in the Squeak
Compiler. It may require some change in the debuggerMethodMap. If one is
willing to do so, he may want to contact me as there as a couple details
that may be useful to be time-efficient.

These features are important for the Sista architecture. I don't need to
load any more dependencies in the Pharo image to have the sista framework
running, though the latest SistaVM is broken *again* (Something related to
mustBeBoolean, it could be mcpc to bcpc mapping).

This semester I teach a lot and I write my phd, but I try to make progress
on the sista every week. Once all the benchmarks from my suite are running
without crashes nor slow-down with the latest Sista VM I will do another
call for contributors and make an open beta. This will happen before
Christmas.

Best,
>> Fabio
>>
>> [1] https://github.com/OpenSmalltalk/opensmalltalk-vm/releas
>> es/tag/201608171728
>> [2] https://github.com/OpenSmalltalk/opensmalltalk-vm/compar
>> e/r3732...201608171728
>>
>
> cheers,
> Eliot
>
>
>>
>> --
>>
>> On Fri, Sep 23, 2016 at 12:16 AM Fabio Niephaus <lists at fniephaus.com>
>> wrote:
>>
>>> Hi Eliot,
>>> I am currently traveling, but I'm happy to take care of this next week.
>>> The procedure is really simple: basically you need to create a Git tag
>>> [1] which you can then turn into a GitHub release, or you can just do all
>>> of this in one step using GitHub's UI (see [2]).
>>>
>>> Best,
>>> Fabio
>>>
>>> [1] https://git-scm.com/book/en/v2/Git-Basics-Tagging#Creating-Tags
>>> [2] https://help.github.com/articles/creating-releases/
>>>
>>> On Thu, Sep 22, 2016 at 11:38 PM Eliot Miranda <eliot.miranda at gmail.com>
>>> wrote:
>>>
>>>> Hi All, Hi Fabio,
>>>>
>>>>    if one looks at the Releases tab on https://github.com/OpenSmal
>>>> ltalk/opensmalltalk-vm it says there is only one release, the r3732
>>>> release from August when we moved to github.  Can we mark the 5.1 release
>>>> as a second release?  How is this done?
>>>>
>>>> _,,,^..^,,,_
>>>> best, Eliot
>>>>
>>>
>
>
> --
> _,,,^..^,,,_
> best, Eliot
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.squeakfoundation.org/pipermail/vm-dev/attachments/20160927/6a3113e9/attachment-0001.htm


More information about the Vm-dev mailing list