pulseaudio install fails

[Ryan Schmidt]

On Mar 10, 2021, at 12:11, joerg van den hoff wrote:

> On 10.03.21 19:00, Ryan Schmidt wrote:
>> On Mar 10, 2021, at 05:58, joerg van den hoff wrote:
>>> I did a (somewhat overdue, thus lengthy ...) `port upgraded outdated' yesterday and ran into some hickups most of which went away after forced uninstalling of affected packages and dependencies and reinstalling them afterwards.
>>> 
>>> however I cannot manage to get `pulseaudio' to install again (a dependency of `surf'). from the log file:
>>> 
>>> 2347 :info:configure checking whether C compiler accepts -std=gnu11... no
>>>   2348 :info:configure configure: error: *** Compiler does not support -std=gnu11
>>>   2349 :info:configure Command failed:  cd "/opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_macports_release_tarballs_ports_a   2349 udio_pulseaudio/pulseaudio/work/pulseaudio-14.2" && ./autogen.sh --prefix=/opt/local --with-mac-version-min=10.15 --with-caps --with-soxr --with-speex    2349 --disable-silent-rules --disable-tests --disable-gtk3 --enable-x11 --disable-oss-output --disable-alsa --disable-esound --disable-solaris --disable-wav   2349 eout --disable-avahi --disable-gconf --disable-gsettings --disable-jack --disable-asyncns --disable-tcpwrap --disable-lirc --disable-bluez5 --disable-u 2349 dev --disable-hal-compat --disable-openssl --disable-systemd-daemon --disable-systemd-login --disable-systemd-journal --enable-neon-opt=no --enable-orc   2349  --enable-adrian-aec --with-mac-sysroot=/Library/Developer/CommandLineTools/SDKs/MacOSX10.15.sdk
>>>   2350 :info:configure Exit code: 1
>>> 
>>> so the `-std=gnu11' compiler flag is not recognized it seems..
>> It should be recognized. There may be a different cause. What's in the config.log?
> 
> a lot (3146 lines) ... :). if I grep for `gnu11' I find these lines:
> 
> 2299 configure:10349: checking whether C compiler accepts -std=gnu11
>   2300 configure:10369: /usr/bin/clang -c -pipe -Os -isysroot/Library/Developer/CommandLineTools/SDKs/MacOSX10.15.sdk -arch x86_64 -pedantic -Werror -std=gnu11 -I/opt/local/include -isysroot/Library/Developer/CommandLineTools/SDKs/MacOSX10.15.sdk conftest.c >&5
>   2301 conftest.c:74:9: error: '_DARWIN_C_SOURCE' macro redefined [-Werror,-Wmacro-redefined]
>   2302 #define _DARWIN_C_SOURCE 200112L
>   2303         ^
>   2304 conftest.c:50:9: note: previous definition is here
>   2305 #define _DARWIN_C_SOURCE 1
>   2306         ^
>   2307 1 error generated.
> 
> or what information would be relevant in this context (did not dare to copy+paste those approx 50 pages of config.log here...)?

Yup, that looks relevant. The configure test fails to do what it was trying to do because it redefines a macro, which is usually accepted with a warning, but it requests that warnings be turned into errors by using -Werror. 

autoconf 2.71 caused this problem to appear. autoconf 2.69 didn't first define _DARWIN_C_SOURCE to 1 but autoconf 2.71 does.

The autoconf 2.70 release notes specifically state that "Running configure tests with warnings promoted to errors is not supported."

Please file a bug report.