Regarding acl2 Port

classic Classic list List threaded Threaded
5 messages Options
Reply | Threaded
Open this post in threaded view
|

Regarding acl2 Port

Jackson Isaac-2
Hi,

I was working on updating the acl2 port version from 3.5 to 8.0

There has been lot of changes ever since to upstream. They moved to github
and the stable versions are released under 'acl2-devel' username
instead of acl2 (upstream).

I was thinking to obsolete acl2 and move the new port to acl2-devel.

Any thoughts or advice regarding this would be really helpful.
Probably from experienced Port developers and/or PortMgr.

I would also like to know how do I test all the variants and different
phases that are mentioned in initial Portfile. There were lot of
things that didn't work correctly with updated version but I was able
to pin point to only errors that I encountered (by running just port
-d build/destroot/install acl2) but there might be other changes too
that may be required. Do we run each and every variant combination and
test the portfile or is there any easy way to do so (maybe run a build
bot job ?).

--
Jackson Isaac
Reply | Threaded
Open this post in threaded view
|

Re: Regarding acl2 Port

Ryan Schmidt-24

On Apr 16, 2018, at 00:10, Jackson Isaac wrote:

> I would also like to know how do I test all the variants and different
> phases that are mentioned in initial Portfile. There were lot of
> things that didn't work correctly with updated version but I was able
> to pin point to only errors that I encountered (by running just port
> -d build/destroot/install acl2) but there might be other changes too
> that may be required. Do we run each and every variant combination and
> test the portfile or is there any easy way to do so (maybe run a build
> bot job ?).

Our Buildbot cannot currently build nondefault variants.

https://trac.macports.org/ticket/52742

Once we add support for building with variants, we will still not by default build all variants.


If you add a variant to a port (or make any other change to a port), you're expected to verify before committing that it at least works on your system.

Reply | Threaded
Open this post in threaded view
|

Re: Regarding acl2 Port

Clemens Lang-2
In reply to this post by Jackson Isaac-2
Hi,

----- On 16 Apr, 2018, at 07:10, Jackson Isaac [hidden email] wrote:

> I was working on updating the acl2 port version from 3.5 to 8.0
>
> There has been lot of changes ever since to upstream. They moved to github
> and the stable versions are released under 'acl2-devel' username
> instead of acl2 (upstream).

I think acl2-devel is the one to go with here. Consider their descriptions
on GitHub:

 acl2: ACL2 System and Books as Maintained by the Community
 acl2-devel: ACL2 System as Maintained by the ACL2 System Authors

The former seems de-facto unmaintaineed.


> I was thinking to obsolete acl2 and move the new port to acl2-devel.

I'd just package acl2-devel in the acl2 port.

> I would also like to know how do I test all the variants and different
> phases that are mentioned in initial Portfile. There were lot of
> things that didn't work correctly with updated version but I was able
> to pin point to only errors that I encountered (by running just port
> -d build/destroot/install acl2) but there might be other changes too
> that may be required. Do we run each and every variant combination and
> test the portfile or is there any easy way to do so (maybe run a build
> bot job ?).

There's no easy way to test variant combinations. With acl2's 5 variants,
testing each combination would already mean starting 32 builds. This is
computationally not feasible.

I usually test each variant separately, and sometimes all variants
together.

--
Clemens Lang
Reply | Threaded
Open this post in threaded view
|

Re: Regarding acl2 Port

Perry E. Metzger
In reply to this post by Jackson Isaac-2
On Mon, 16 Apr 2018 10:40:12 +0530 Jackson Isaac
<[hidden email]> wrote:
> Hi,
>
> I was working on updating the acl2 port version from 3.5 to 8.0
>
> There has been lot of changes ever since to upstream. They moved to
> github and the stable versions are released under 'acl2-devel'
> username instead of acl2 (upstream).
>
> I was thinking to obsolete acl2 and move the new port to acl2-devel.

Which ACL2 are you referring to? Because there's also this:

https://github.com/acl2/acl2

which is I believe where the thing is now being maintained.


> Any thoughts or advice regarding this would be really helpful.
> Probably from experienced Port developers and/or PortMgr.
>
> I would also like to know how do I test all the variants and
> different phases that are mentioned in initial Portfile. There were
> lot of things that didn't work correctly with updated version but I
> was able to pin point to only errors that I encountered (by running
> just port -d build/destroot/install acl2) but there might be other
> changes too that may be required. Do we run each and every variant
> combination and test the portfile or is there any easy way to do so
> (maybe run a build bot job ?).
>



--
Perry E. Metzger [hidden email]
Reply | Threaded
Open this post in threaded view
|

Re: Regarding acl2 Port

Perry E. Metzger
In reply to this post by Clemens Lang-2
On Mon, 16 Apr 2018 10:13:18 +0200 (CEST) Clemens Lang
<[hidden email]> wrote:

> Hi,
>
> ----- On 16 Apr, 2018, at 07:10, Jackson Isaac
> [hidden email] wrote:
>
> > I was working on updating the acl2 port version from 3.5 to 8.0
> >
> > There has been lot of changes ever since to upstream. They moved
> > to github and the stable versions are released under 'acl2-devel'
> > username instead of acl2 (upstream).  
>
> I think acl2-devel is the one to go with here. Consider their
> descriptions on GitHub:
>
>  acl2: ACL2 System and Books as Maintained by the Community
>  acl2-devel: ACL2 System as Maintained by the ACL2 System Authors
>
> The former seems de-facto unmaintaineed.

Last commit on the former was nine hours ago.
Last commit on the latter was in mid-December.

That said, I have no idea which is superior.

I suspect that the names aren't reasonable, in that -devel isn't the
development branch of acl2, but is an unrelated fork at this point.

Perry
--
Perry E. Metzger [hidden email]