Deprecating support for ports

[Adam Mercer]

Hi

I work as part of the LIGO project and for a while now we've used
MacPorts to provide support for our collaboration members running our
software under macOS. For various reasons we are in the process of
switching over to Conda as this allows us to support Linux, macOS, and
even Windows (in some circumstances) from the same package source
thereby reducing the amount of work needed to support all the
different platforms used.

As such we would like to indicate that the ports we have in MacPorts
are no longer supported by us, looking through the various PortGroups
it seems that the deprecated group allows us to specify that a Port is
no longer under support in MacPorts. Therefore would the correct
course of action here be to add the following to the Ports we wish to
mark as no longer supported:

PortGroup deprecated 1.0
deprecated.macports_support no

In addition to setting the maintainer to nomaintainer?

At the moment all the ports work but that may not the be case as time
goes on, so eventually these will probably need to be removed. However
as they are still currently working I'd like to keep them around in
case anyone else would like to maintain them.

Any comments or suggestions regarding this?

Cheers

Adam

Re: macOS10.14 testers ?

[Adam Mercer]

On Wed, Jul 4, 2018 at 8:41 AM Christopher Jones
 wrote:

> I am just curious, has anyone else signed up for the 10.14 beta and played 
> with it and MacPorts ?
>
> I have been doing this myself, in a VM, and whilst we cannot discuss anything 
> here on list, as per the NDA, I would be interested to in private share what 
> I have found, as there are perhaps some interesting things to discuss a bit….

I have been testing the developer seeds as well, ran into a few
problems but haven't had time to investigate further. I'd be
interested on seeing if we'd run into the same problems and if you've
found solutions.

Cheers

Adam

Re: [macports-ports] branch master updated: science/lscsoft-deps: add cmake

[Adam Mercer]

On Tue, Jun 12, 2018 at 7:17 AM Ryan Schmidt  wrote:

> Assuming there's no reason why cmake-devel wouldn't work, this should instead 
> be written as "path:bin/cmake:cmake".

Thanks, change pushed.

Cheers

Adam

Re: [macports-ports] branch master updated: devel/automake: update to 1.16.1

[Adam Mercer]

On Thu, Mar 15, 2018 at 3:32 PM, Ryan Schmidt  wrote:

>> Good point. I'll get a patch ready to switch back to the .gz as
>> circular dependencies are always a nightmare.
>
> Well, I see autoconf already uses xz too, so I guess we can leave it for now.

Pushed a change just before this came through.

Cheers

Adam

Re: [macports-ports] branch master updated: devel/automake: update to 1.16.1

[Adam Mercer]

On Thu, Mar 15, 2018 at 3:23 PM, Ryan Schmidt  wrote:

>>  master_sitesgnu
>> -checksums   rmd160  3dfd5d5aa037db5aae281829539eef328694cd75 \
>> -sha256  
>> 988e32527abe052307d21c8ca000aa238b914df363a617e38f4fb89f5abf6260
>> +use_xz  yes
>> +checksums   rmd160  8c2a91e8fc0595dbf4854be3cef4d14bb8c5e756 \
>> +sha256  
>> 5d05bb38a23fd3312b10aea93840feec685bdf4a41146e78882848165d3ae921
>
> This works right now, but it's very easy for this to cause problems in the 
> future, because it means xz and xz's dependencies can't use automake; if they 
> did, it would be a circular dependency and new installs would fail with a 
> confusing lack of error message.

Good point. I'll get a patch ready to switch back to the .gz as
circular dependencies are always a nightmare.

Cheers

Adam

Re: science/lalinference: update to 1.8.2

[Adam Mercer]

On Sat, Jan 7, 2017 at 6:04 AM, Joshua Root  wrote:

>> 
>>
>>
>> This also reverted
>>
>> ,
>> which really is necessary. In general because you don't control the
>> value of $frameworks_dir, and it could contain characters which have
>> special meaning in a regex, and because string replacement on a Tcl list
>> is hard enough to get right. And specifically because in MacPorts 2.4 we
>> will once again be setting prefix to '${prefix}' while building the
>> PortIndex (the code that does that was accidentally non-functional in
>> 2.3).

Thanks for pointing that out, I'll get that change back in place. This
change was dropped due to the way we currently managed these ports, we
develop them out of a different repository and then when they're ready
we just copy them to the MacPorts git reposiry and commit. We really
need to find a more sane way as things like this are bound to happen
again.

It also sounds like a similar change should be made to the other lal* ports.

> Although I see I also made a typo; sorry about that. The dollar sign should
> be in front of the first brace on line 121, not in front of the bracket.

I assume you mean the following:

diff --git a/science/lalinference/Portfile b/science/lalinference/Portfile
index 018a9d96ed..f92ad99bbb 100644
--- a/science/lalinference/Portfile
+++ b/science/lalinference/Portfile
@@ -118,7 +118,7 @@ foreach v {27 34 35} {
 depends_lib-replace   port:lal port:py${v}-lal
 depends_lib-appendport:${name} port:py${v}-numpy

-configure.args-strsed s|$[quotemeta
{default_pythonarg}]|PYTHON=[quotemeta
${pythonhome}/bin/python${pythonversion}]|
+configure.args-strsed s|[quotemeta
${default_pythonarg}]|PYTHON=[quotemeta
${pythonhome}/bin/python${pythonversion}]|
 configure.args-replace --disable-python --enable-python
 configure.args-append --enable-swig-python

Cheers

Adam

Re: Triggering a rebuild on the buildbot

[Adam Mercer]

On Tue, Nov 15, 2016 at 9:50 PM, Ryan Schmidt  wrote:

> I usually start from the "Builders" page:
>
> https://build.macports.org/builders
>
> Scroll down to the "portwatcher_force" section.

Thanks.

Cheers

Adam

Re: Triggering a rebuild on the buildbot

[Adam Mercer]

Thanks Ryan

I don't have a buildbot login, what do you need from me to set one up?

Cheers

Adam

-- 
Sent from Android

On Nov 15, 2016 15:04, "Ryan Schmidt"  wrote:

>
>
> > On Nov 15, 2016, at 13:53, Adam Mercer  wrote:
> >
> > Hi
> >
> > A recent commit failed to build and I tracked the problem down and it
> > was due to a missing dependency on a dependency of the port that
> > failed. I fixed this is the dependency port and I was wondering if
> > there's a way to trigger a rebuild of the port on the buildbot without
> > revbumping?
>
> Yes, if you have a buildbot login you can force a new build of any port(s).
>
> If you don't have a login, I can make one for you.
>

Triggering a rebuild on the buildbot

[Adam Mercer]

Hi

A recent commit failed to build and I tracked the problem down and it
was due to a missing dependency on a dependency of the port that
failed. I fixed this is the dependency port and I was wondering if
there's a way to trigger a rebuild of the port on the buildbot without
revbumping?

Cheers

Adam