On Fri, Sep 02, 2016 at 12:59:53AM +0200, doncatnip wrote:
> Hey Guix !
> Let's try this again.
> Subject: [PATCH 1/3] gnu: lua: Remove conflicting flag, pass MYCFLAGS
> * gnu/packages/lua.scm (lua)[arguments]: Use MYCFLAGS instead of
> CFLAGS and remove conflicting -DLUA_USE_POSIX since -DLUA_USE_LINUX
> is passed implicitly for build target "linux".
I made the commit title into a complete sentence and pushed.
> Subject: [PATCH 2/3] gnu: lua: Add lua-lgi.
> * gnu/packages/lua.scm (lua-lgi): New variable.
I took ng0's advice about the comment styles:
I also made the phases 'set-env', 'set-lua-version', 'skip-test-gtk',
and 'start-xserver-instance' return #t. We want successful phases to
return #t, but things like setenv and substitute* do not specify a
return value, so we explicity return #t.
I moved #:make-flags to the top of the arguments block, which kept
it shorter than 80 columns. And I corrected the indentation of
#:phases by one column.
> Subject: [PATCH 3/3] gnu: awesome: Update to 3.5.9.
> * gnu/packages/wm.scm (awesome): Update to 3.5.9.
> (awesome)[inputs]: Add gobject-introspection, lua-lgi, cairo. Use
> latest available lua.
> (awesome)[arguments]: Set lua search paths. Add cairo to
> LD_LIBRARY_PATH. Wrap binary in respect to those paths plus
The commit message only needs to refer to the (awesome) variable once in
this case. Referring to the fields [inputs] and [arguments] later does
not require the variable to be mentioned again. Fixed before pushing.
I sorted the new inputs alphabetically, since the pre-existing inputs
were already sorted. I made the comment about loading Cairo dynamically
into a complete sentence. I made the phase 'set-lua-paths' return #t.
Finally, since the new 'wrap' phase had a too-long line that was ugly to
break up, I shifted the indentation of #:phases down one line and to the
left, giving some more room to breathe on the right.
Pushed as 22037a327a0340341df7ae71a9c1d3551c28c705 !