0.8.3 for mid-July?

0.8.3 for mid-July?

[Ludovic Courtès]

Hello!

I think we should release 0.8.3 for mid to end July.

The relevant things to be done include:

  • Merging wip-diet and wip-environment, and maybe wip-container,
    though that one could just as well be merged afterwards.

  • We must sit down and fix as many bugs as possible.
    <http://bugs.gnu.org/20765> is a blocker IMO, but help is needed
    from Pythonistas.  <http://bugs.gnu.org/20888> must be fixed too,
    but there are many others.

  • Several people reported getting a kernel panic, instead of a Guile
    prompt, when something goes wrong during early boot (in the initrd.)
    I haven’t been able to reproduce it in a VM; if someone could
    reproduce it, that would be great.

There are optimizations that I would like to look into it, though they
are not blocker IMO:

  • ‘guix build hydra -n’ literally takes ages, as Eric noted, so that
    needs to be investigated.

  • ‘guix system’ is quite slow, but it’s apparently computing the same
    derivations several times, which must be fixed.

Anything else?  Comments?

Thanks,
Ludo’.

Re: 0.8.3 for mid-July?

[Thompson, David]

On Wed, Jul 1, 2015 at 9:27 AM, Ludovic Courtès <ludo@gnu.org> wrote:
> Hello!
>
> I think we should release 0.8.3 for mid to end July.

That sounds good.  I guess now is a good time to mention that I will
be AFK from July 10th to the 17th.

> The relevant things to be done include:
>
>   • Merging wip-diet and wip-environment, and maybe wip-container,
>     though that one could just as well be merged afterwards.

I looked at wip-environment last night and I liked what I saw.  I did
a couple of smoke tests and everything worked as expected.  Thanks for
figuring out some of the difficult issues there, like only including
the relevant outputs in the environment.

I'd really like wip-container to make it in so that we can start
showing it off, but I'm currently blocked on a working
eval-in-container implementation.  Perhaps I could save that for next
release and polish what works ('guix environment --container', 'guix
system container')?

>   • We must sit down and fix as many bugs as possible.
>     <http://bugs.gnu.org/20765> is a blocker IMO, but help is needed
>     from Pythonistas.  <http://bugs.gnu.org/20888> must be fixed too,
>     but there are many others.
>
>   • Several people reported getting a kernel panic, instead of a Guile
>     prompt, when something goes wrong during early boot (in the initrd.)
>     I haven’t been able to reproduce it in a VM; if someone could
>     reproduce it, that would be great.
>
> There are optimizations that I would like to look into it, though they
> are not blocker IMO:
>
>   • ‘guix build hydra -n’ literally takes ages, as Eric noted, so that
>     needs to be investigated.
>
>   • ‘guix system’ is quite slow, but it’s apparently computing the same
>     derivations several times, which must be fixed.
>
> Anything else?  Comments?

Let's do this! :)

- Dave

Re: 0.8.3 for mid-July?

[Ludovic Courtès]

"Thompson, David" <dthompson2@worcester.edu> skribis:

> On Wed, Jul 1, 2015 at 9:27 AM, Ludovic Courtès <ludo@gnu.org> wrote:
>> Hello!
>>
>> I think we should release 0.8.3 for mid to end July.
>
> That sounds good.  I guess now is a good time to mention that I will
> be AFK from July 10th to the 17th.

Noted.  I’ll be mostly AFK after July 20th.

>> The relevant things to be done include:
>>
>>   • Merging wip-diet and wip-environment, and maybe wip-container,
>>     though that one could just as well be merged afterwards.
>
> I looked at wip-environment last night and I liked what I saw.  I did
> a couple of smoke tests and everything worked as expected.  Thanks for
> figuring out some of the difficult issues there, like only including
> the relevant outputs in the environment.

Thanks for looking at it.  I’ll merge it later today, then.

> I'd really like wip-container to make it in so that we can start
> showing it off, but I'm currently blocked on a working
> eval-in-container implementation.  Perhaps I could save that for next
> release and polish what works ('guix environment --container', 'guix
> system container')?

Sure, let’s start polishing what’s already there to merge it, and then
we’ll see if we can debug ‘eval-in-container’.

Thanks!

Ludo’.

Re: 0.8.3 for mid-July?

[Ludovic Courtès]

ludo@gnu.org (Ludovic Courtès) skribis:

> I think we should release 0.8.3 for mid to end July.
>
> The relevant things to be done include:

I forgot something: the ELPA importer and emacs-build-system.
Federico, WDYT?  Does that sound doable?

Thanks,
Ludo’.

Re: 0.8.3 for mid-July?

[Federico Beffa]

On Sat, Jul 4, 2015 at 4:45 PM, Ludovic Courtès <ludo@gnu.org> wrote:
> ludo@gnu.org (Ludovic Courtès) skribis:
>
>> I think we should release 0.8.3 for mid to end July.
>>
>> The relevant things to be done include:
>
> I forgot something: the ELPA importer and emacs-build-system.
> Federico, WDYT?  Does that sound doable?

Yes, the second half of July should be OK.

Regards,
Fede

Re: 0.8.3 for mid-July?

[Ludovic Courtès]

ludo@gnu.org (Ludovic Courtès) skribis:

>   • ‘guix build hydra -n’ literally takes ages, as Eric noted, so that
>     needs to be investigated.

Commit 161094c addresses this issue, which was in the
‘transitive-inputs’ procedure, which computes the set of propagated
inputs that are inherited (see commit log for details.)

This change is in ‘core-updates’ because it triggers a full rebuild, due
to the fact that it computes different lists of inputs.

I think that this and the other changes and upgrades that are currently
in ‘core-updates’ may well justify a merge before the release.

Thoughts?

Ludo’.

0.8.3 update

[Ludovic Courtès]

Hi!

ludo@gnu.org (Ludovic Courtès) skribis:

> I think we should release 0.8.3 for mid to end July.

I would like to target Wednesday, 22nd, which means the release must be
tagged on Tuesday.  That sounds doable.

Mark: will you be around to produce the armhf tarball?

By then we’ll have to merge core-updates.  It’s mostly built on x86
already; mips64el may be largely built by tomorrow, hopefully.

>   • Merging wip-diet and wip-environment, and maybe wip-container,
>     though that one could just as well be merged afterwards.

There are missing bits from wip-container.  We’ll see whether David can
help with that, but otherwise that’ll be for the next time.

>   • We must sit down and fix as many bugs as possible.
>     <http://bugs.gnu.org/20765> is a blocker IMO, but help is needed
>     from Pythonistas.  <http://bugs.gnu.org/20888> must be fixed too,
>     but there are many others.

Python people, help!

>   • ‘guix build hydra -n’ literally takes ages, as Eric noted, so that
>     needs to be investigated.

Fixed in core-updates.

>   • ‘guix system’ is quite slow, but it’s apparently computing the same
>     derivations several times, which must be fixed.

Haven’t checked that yet.


Now is the time to run the test suite and report any issues:

  https://www.gnu.org/software/guix/manual/html_node/Running-the-Test-Suite.html

I’ll push an installation image after the core-updates merge.  In the
meantime, you can also build it yourself from ‘master’ to give it a try:

  https://www.gnu.org/software/guix/manual/html_node/System-Installation.html#Building-the-Installation-Image

Ludo’.