Bug#1019413: debcrossgen: host_arch.cpu_family() reports powerpc64le on ppc64el but should report ppc64

2022-11-09 Thread Helmut Grohne
Hi Jussi,

On Wed, Sep 14, 2022 at 12:06:20AM +0300, Jussi Pakkanen wrote:
> We have moved this functionality from debcrossgen to Meson itself.
> This means that the packaging tools should switch from calling
> debcrossgen to `meson env2mfile` as described in this bug:
> 
> https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1017436
> 
> Dunno what is the correct way to get that fixed, though...

I'm sorry. This fell through the cracks somehow, but you're still
looking for an answer, right?

You need to locate all users of debcrossgen and fix them individually. A
very easy way of doing so is using codesearch:

https://codesearch.debian.net/search?q=debcrossgen&literal=1

 - debhelper is the place you really want to fix
 + glib2.0 is a false positive and no longer runs it directly
 - mesa, monado and phosh call debcrossgen for gitlab-ci

So that's probably sending 4 distinct patches. The most important one is
debhelper. I think all Debian package builds run debcrossgen via
debhelper.

I think this should be reasonably simple, see:
https://sources.debian.org/src/debhelper/13.10.1/lib/Debian/Debhelper/Buildsystem/meson.pm/?hl=70#L70

Helmut



Bug#1019413: debcrossgen: host_arch.cpu_family() reports powerpc64le on ppc64el but should report ppc64

2022-09-13 Thread Jussi Pakkanen
On Thu, 8 Sept 2022 at 23:33, Helmut Grohne  wrote:

> host_arch.cpu_family() should be reporting ppc64 on ppc64el according to
> https://mesonbuild.com/Reference-tables.html, but it actually reports
> powerpc64le when debcrossgen is in use. The relevant mapping should be
> fixed in debcrossgen. I'm attaching a patch for your convenience.

We have moved this functionality from debcrossgen to Meson itself.
This means that the packaging tools should switch from calling
debcrossgen to `meson env2mfile` as described in this bug:

https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1017436

Dunno what is the correct way to get that fixed, though...



Bug#1019413: debcrossgen: host_arch.cpu_family() reports powerpc64le on ppc64el but should report ppc64

2022-09-08 Thread Helmut Grohne
Package: meson
Version: 0.63.2-1
Tags: patch
User: debian-cr...@lists.debian.org
Usertags: ftcbfs
Control: affects -1 + src:dpdk

host_arch.cpu_family() should be reporting ppc64 on ppc64el according to
https://mesonbuild.com/Reference-tables.html, but it actually reports
powerpc64le when debcrossgen is in use. The relevant mapping should be
fixed in debcrossgen. I'm attaching a patch for your convenience.

Helmut
diff --minimal -Nru meson-0.63.2/debian/changelog meson-0.63.2/debian/changelog
--- meson-0.63.2/debian/changelog   2022-09-03 18:33:44.0 +0200
+++ meson-0.63.2/debian/changelog   2022-09-08 22:26:00.0 +0200
@@ -1,3 +1,10 @@
+meson (0.63.2-1.1) UNRELEASED; urgency=medium
+
+  * Non-maintainer upload.
+  * debcrossgen: fix host_arch.cpu_family() on ppc64el. (Closes: #-1)
+
+ -- Helmut Grohne   Thu, 08 Sep 2022 22:26:00 +0200
+
 meson (0.63.2-1) unstable; urgency=medium
 
   * New upstream release. Closes: #1017087.
diff --minimal -Nru meson-0.63.2/debian/debcrossgen 
meson-0.63.2/debian/debcrossgen
--- meson-0.63.2/debian/debcrossgen 2021-02-14 13:58:40.0 +0100
+++ meson-0.63.2/debian/debcrossgen 2022-09-08 22:25:57.0 +0200
@@ -73,7 +73,8 @@
 cpu_family_map = dict(mips64el="mips64",
   i686='x86')
 cpu_map = dict(armhf="arm7hlf",
-   mips64el="mips64",)
+   mips64el="mips64",
+   powerpc64le="ppc64",)
 
 def run(options):
 if options.arch is None: