You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@lucy.apache.org by Marvin Humphrey <ma...@rectangular.com> on 2014/08/19 02:55:32 UTC

Re: [lucy-dev] 0.4 release?

Greets,

I think we're pretty much done prepping 0.4.0 and it's time to call a release
vote as soon as the prep_0.4.0 branches are merged to master.

Since Clownfish and Lucy will henceforth be separate products, I plan to run
separate votes for this first independent release of Clownfish.  In the future
we might run a combined vote for the sake of convenience.

Reviewing the checklist at <http://wiki.apache.org/lucy/ReleasePrep>:

*   The RAT buildbot is currently passing for Lucy.  We don't have a RAT
    buildbot for Clownfish AFAIK so I ran RAT locally and the output looked
    OK.
*   The issue tracker is clean for 0.4.0.
*   The Lucy CHANGES file has been brought up to date and the Clownfish
    CHANGES file has been inaugurated.
*   LICENSE and NOTICE look correct for both Lucy and Clownfish.
*   As of a week or two ago, Lucy and Clownfish built and passed tests on a
    variety of Windows configurations, plus Linux, FreeBSD (our buildbot) and
    OS X.  It might be nice to check NetBSD and Solaris.
*   The valgrind test targets have been modernized to use a more flexible
    suppressions format and are now more reliable and less fussy about
    compilation options and Perl versions.  I'll give it another pass before
    submitting the RC but they should be passing.

In addition, the update_version scripts have been been adapted for the changes
since 0.3.x and I've checked the product of the CPAN dist building process.
Every module listed in 'provides' has the correct version number (though to be
honest the list of classes is not exhaustive) so we should get a clean index
on CPAN.

Marvin Humphrey

Re: [lucy-dev] 0.4 release?

Posted by Marvin Humphrey <ma...@rectangular.com>.
On Mon, Aug 18, 2014 at 5:55 PM, Marvin Humphrey <ma...@rectangular.com> wrote:

> *   The valgrind test targets have been modernized to use a more flexible
>     suppressions format and are now more reliable and less fussy about
>     compilation options and Perl versions.  I'll give it another pass before
>     submitting the RC but they should be passing.

I found a memory leak in RC1, which I will now discard. After
committing a fix, I'll roll RC2.

Marvin Humphrey

Re: [lucy-dev] 0.4 release?

Posted by Nick Wellnhofer <we...@aevum.de>.
On Aug 19, 2014, at 19:52 , Marvin Humphrey <ma...@rectangular.com> wrote:

> Clownfish is likely to remain in significant flux at least until we've
> implemented bindings for more languages.  Here are some of the upcoming
> challenges which will likely require modifications to the Clownfish core:
> 
> *   Tracing GC (Ruby, Go)

I’ve also been thinking about this. It should be feasible if we keep track of all Clownfish objects that were passed to the host language.

> *   Non-UTF-8 host string type (Python)

A simple approach would be to reencode every string that is passed from the host language and back. But for a more performant solution, Clownfish needs support for multiple string encodings.

> *   Limited access to GC internals hampers host subclassing (Go)

I don’t know anything about how Go interfaces with C, but I can imagine that the lack of virtual methods in Go poses an interesting problem.

> This is super fun stuff to work on. :)

I hope you didn’t mean that ironically. It definitely sounds like fun stuff to me. :)

> Having Lucy as a real-world example is invaluable during this period and I
> think releases will be coordinated for a while yet.

Yes, but now that Clownfish is cleanly separated, it should be much easier to start supporting new host languages without having to care about Lucy at first.

Nick


Re: [lucy-dev] 0.4 release?

Posted by Marvin Humphrey <ma...@rectangular.com>.
On Tue, Aug 19, 2014 at 5:28 AM, Nick Wellnhofer <we...@aevum.de> wrote:

> I could imagine that the release schedules of Lucy and Clownfish will
> diverge at some point. Maybe we should also try to keep the Lucy master
> branch compatible with an official Clownfish release. Then Lucy would only
> be updated for Clownfish changes once a new version of Clownfish is
> released. This would essentially force us to release Lucy and Clownfish
> separately.

In the long term, divergence is inevitable.

> On the other hand, it might be useful to test whether changes to Clownfish
> work with Lucy as soon as possible.

Clownfish is likely to remain in significant flux at least until we've
implemented bindings for more languages.  Here are some of the upcoming
challenges which will likely require modifications to the Clownfish core:

*   Tracing GC (Ruby, Go)
*   Ordering requirements for optional parameters (Python)
*   Non-UTF-8 host string type (Python)
*   Limited access to GC internals hampers host subclassing (Go)
*   Mapping parcel/class names to host names and constructs (all)

This is super fun stuff to work on. :)

Having Lucy as a real-world example is invaluable during this period and I
think releases will be coordinated for a while yet.

Marvin Humphrey

Re: [lucy-dev] 0.4 release?

Posted by Nick Wellnhofer <we...@aevum.de>.
On 19/08/2014 02:55, Marvin Humphrey wrote:
> Since Clownfish and Lucy will henceforth be separate products, I plan to run
> separate votes for this first independent release of Clownfish.  In the future
> we might run a combined vote for the sake of convenience.

I could imagine that the release schedules of Lucy and Clownfish will diverge 
at some point. Maybe we should also try to keep the Lucy master branch 
compatible with an official Clownfish release. Then Lucy would only be updated 
for Clownfish changes once a new version of Clownfish is released. This would 
essentially force us to release Lucy and Clownfish separately.

On the other hand, it might be useful to test whether changes to Clownfish 
work with Lucy as soon as possible.

> Reviewing the checklist at <http://wiki.apache.org/lucy/ReleasePrep>:
>
> *   The RAT buildbot is currently passing for Lucy.  We don't have a RAT
>      buildbot for Clownfish AFAIK so I ran RAT locally and the output looked
>      OK.
> *   The issue tracker is clean for 0.4.0.
> *   The Lucy CHANGES file has been brought up to date and the Clownfish
>      CHANGES file has been inaugurated.
> *   LICENSE and NOTICE look correct for both Lucy and Clownfish.
> *   As of a week or two ago, Lucy and Clownfish built and passed tests on a
>      variety of Windows configurations, plus Linux, FreeBSD (our buildbot) and
>      OS X.  It might be nice to check NetBSD and Solaris.
> *   The valgrind test targets have been modernized to use a more flexible
>      suppressions format and are now more reliable and less fussy about
>      compilation options and Perl versions.  I'll give it another pass before
>      submitting the RC but they should be passing.
>
> In addition, the update_version scripts have been been adapted for the changes
> since 0.3.x and I've checked the product of the CPAN dist building process.
> Every module listed in 'provides' has the correct version number (though to be
> honest the list of classes is not exhaustive) so we should get a clean index
> on CPAN.

Looks good.

Nick


Re: [lucy-dev] 0.4 release?

Posted by Peter Karman <pe...@peknet.com>.
Marvin Humphrey wrote on 8/18/14 7:55 PM:
> Greets,
> 
> I think we're pretty much done prepping 0.4.0 and it's time to call a release
> vote as soon as the prep_0.4.0 branches are merged to master.

sweet.

A big thanks to you and Nick for pulling all this together. I look forward to
Life With Apache Clownfish and what comes next.


-- 
Peter Karman  .  http://peknet.com/  .  peter@peknet.com