Bug 216926 - [maintainer] math/amath: update to 1.6.4
Summary: [maintainer] math/amath: update to 1.6.4
Status: Closed FIXED
Alias: None
Product: Ports & Packages
Classification: Unclassified
Component: Individual Port(s) (show other bugs)
Version: Latest
Hardware: Any Any
: --- Affects Only Me
Assignee: Joseph Mingrone
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2017-02-08 21:17 UTC by Carsten Larsen
Modified: 2017-02-13 22:28 UTC (History)
1 user (show)

See Also:


Attachments
patch 1.6.4 (1.39 KB, patch)
2017-02-08 21:17 UTC, Carsten Larsen
cs: maintainer-approval+
Details | Diff

Note You need to log in before you can comment on or make changes to this bug.
Description Carsten Larsen 2017-02-08 21:17:53 UTC
Created attachment 179775 [details]
patch 1.6.4

portlint:
  looks fine.

poudriere:
  9.3  {amd64, i386} Fails
  10.3 {amd64, i386} OK
  11.0 {amd64, i386, arm} OK
  11.0 {arm64, powerpc} Not tested

make test:
  {amd64, i386, arm} no errors

arm64 & powerpc is not tested since I couldn't get poudriere cross compiling to work (C compiler cannot create executables).
Comment 1 Joseph Mingrone freebsd_committer freebsd_triage 2017-02-12 13:45:07 UTC
How about https://reviews.freebsd.org/D9541 ?

I am not clear why you redefine ARCH.  What's wrong with the default value, 'uname -p'?

I feel the ONLY_FOR_ARCHS definition is unnecessary.  You are accounting for most supported architectures except for uncommon ones like sparc64.  Also, I tested the build on armv6 and it passed.
Comment 2 Carsten Larsen 2017-02-13 19:56:05 UTC
(In reply to Joseph Mingrone from comment #1)

My intention for adding ONLY_FOR_ARCHS was partly to explicit state approved architectures, and partly to allow pre-build packages as shown on PortsMon:
http://portsmon.freebsd.org/portoverview.py?category=math&portname=amath

The redefinition of ARCH was due to the variety of arm architectures and inspired by this mail: 
https://lists.freebsd.org/pipermail/freebsd-ports/2015-November/100968.html

If there is a better way to mark specific architectures as approved (and tested) I would be happy to learn about it.
Comment 3 Joseph Mingrone freebsd_committer freebsd_triage 2017-02-13 21:34:06 UTC
I think documentation might the appropriate place to convey to users that the application/port has been tested on specific architectures.  In this case, your code seems robust across many architectures, so it makes sense to list those that you do not plan to support in NOT_FOR_ARCHS or those that are temporarily broken in BROKEN_ARCH.

As far as the packages builders go, I am not aware of builds being triggered based on an architecture being added to ONLY_FOR_ARCHS.
Comment 4 Joseph Mingrone freebsd_committer freebsd_triage 2017-02-13 22:15:39 UTC
I confirmed that removing ONLY_FOR_ARCH means a build will be attempted on all architectures (a package for aarch64 won't be attempted because it's currently marked broken) and pkg.freebsd.org currently supports aarch64 amd64 armv6 and i386 right now.  The patch at https://reviews.freebsd.org/D9541 was confirmed correct by a few developers, so I will commit it.
Comment 5 commit-hook freebsd_committer freebsd_triage 2017-02-13 22:27:26 UTC
A commit references this bug:

Author: jrm
Date: Mon Feb 13 22:26:26 UTC 2017
New revision: 434058
URL: https://svnweb.freebsd.org/changeset/ports/434058

Log:
  math/amath: Update to version 1.6.4.

  Remove ONLY_FOR_ARCHS.

  PR:		216926
  Submitted by:	cs@innolan.dk (maintainer)
  Reviewed by:	linimon
  Approved by:	swills (mentor)
  Differential Revision:	https://reviews.freebsd.org/D9541

Changes:
  head/math/amath/Makefile
  head/math/amath/distinfo
Comment 6 Joseph Mingrone freebsd_committer freebsd_triage 2017-02-13 22:28:48 UTC
Committed. Thanks.