platforms problems

Andrew Whitworth wknight8111 at gmail.com
Fri Aug 3 00:18:30 UTC 2012


On Thu, Aug 2, 2012 at 12:54 PM, Patrick R. Michaud <pmichaud at pobox.com> wrote:
> On Wed, Aug 01, 2012 at 08:19:11PM -0400, Andrew Whitworth wrote:
> Sadly, I don't have any suggestions, but I do want to make sure I
> understand the implications of this for Rakudo.
>
> In this email I'm not intending to place or indicate any value judgements
> on the options available; I just want to make sure I (and others)
> understand the situation fully.
>
> If I'm understanding you correctly, merging the threads and io_cleanup
> branches into Parrot will negate the ability to run Parrot on
> Darwin/PPC, some BSDs, and some Solaris systems.  So, if these branches
> are merged into Parrot's master branch and Parrot drops support for
> these platforms, then Rakudo must likewise make a choice:
>   1) continue following the mainline Parrot development, dropping
>      any support for Rakudo on Darwin/PPC, *BSD, and *Solaris, or
>   2) remain pegged to an older version of Parrot (e.g. 4.6.0) until support
>      for some or all of those platforms can be restored.

The problems are these:

1) Parrot is suffering from a pronounced shortage of developer
activity lately. Whether this is a long-term trend or something
temporary (and, possibly, related to real-world economic issues among
others) is not a question I have an answer to. I do know that as far
as hacking on Parrot core, I'm one of the few people doing it right
now with any regularity.

2) The two big projects I've got on my plate, threads and
whiteknight/io_cleanup1 are failing tests and exhibiting problems on
platforms that I do not have access to for fixing or development.

3) I cannot complete my work. Therefore I cannot merge it (unless we
change our support policy to exclude certain platforms, which is not
something I rule out in all cases).

4) I have had many free hours in the past two weeks or so when I
wanted to work on Parrot, but could not, because I do not have
reliable access to platforms where the last bits of work are needed.

So that's the problem we're in. I cannot work, because I am at a
roadblock that I cannot get past with my own resources. As a larger
issue, Parrot as a community simply does not have any core developers
on a variety of platforms able to deal with problems as they appear.
The fact that things work on those platforms in master is a temporary
and delicate situation. A single breakage due to changes in parrot or
changes elsewhere in the toolchain for those platforms would be, at
this point, mostly irrepairable.

> Speaking somewhat more selfishly:  What benefits will Rakudo gain from
> the branch merges?

Darwin/PPC is a special issue because it's basically a dead platform
and I don't feel like we need to bend over backwards to support it if
we don't have the resources to do it. Solaris and BSD are very live,
and failures there are very serious. We WILL NOT merge either of these
branches so long as we've got failures on those platforms and a
reasonable expectation that we should not have failures there.

The issue is not "we're going to merge these branches come hell or
high water". The issue is that we cannot merge these branches because
of failures that we cannot fix. Development effort on these and
several other projects are at a full stop. And I'm not going to start
another branch, to start yet another project, without finishing things
I've started and without having confidence that I'll ever be able to
merge what I've developed.

This email thread is more of a status update. I am not seriously
recommending any particular course of action, and I'm not threatening
to break anything. Things will continue to work, as they currently
work, on platforms where things are currently working. Whether that's
a good thing or not is subjective.

--Andrew Whitworth


More information about the parrot-dev mailing list