build cleanups.

build cleanups.

* Condense config.make.in

* Reduce make style expansion in config.make; this makes interop with
  other build systems easier.

* Trim dependencies on version.hh

* Add ALL_CXXFLAGS to link line

[hanwenn, 2020-02-29 20:50:54]

ld

[lemzwerg, 2020-02-29 21:36:33]

LGTM

[dak, 2020-02-29 22:03:04]

https://codereview.appspot.com/547690053/diff/567300043/config.make.in
File config.make.in (right):

https://codereview.appspot.com/547690053/diff/567300043/config.make.in#newcode48
config.make.in:48: GROFF = @GROFF@
These sort of drive-by changes without any mention in commit message or issue
make it a bit harder to review.  I find

stepmake/stepmake/documentation-rules.make:     troff -man -Tascii $< | grotty
-b -u -o > $@

which is awkward by not using anything actually tested for.  Nevertheless, use
of grotty probably requires the presence of the groff package.  The test should
likely be better targeted than what we currently have, though.

[hanwenn, 2020-02-29 22:12:06]

commit msg

[hanwenn, 2020-02-29 22:13:42]

added 

    * Remove unused GROFF and LD autoconf vars

to the commit msg.

https://codereview.appspot.com/547690053/diff/567300043/config.make.in
File config.make.in (right):

https://codereview.appspot.com/547690053/diff/567300043/config.make.in#newcode48
config.make.in:48: GROFF = @GROFF@
On 2020/02/29 22:03:03, dak wrote:
> These sort of drive-by changes without any mention in commit message or issue
> make it a bit harder to review.  I find
> 
> stepmake/stepmake/documentation-rules.make:     troff -man -Tascii $< | grotty
> -b -u -o > $@
> 
> which is awkward by not using anything actually tested for.  Nevertheless, use
> of grotty probably requires the presence of the groff package.  The test
should
> likely be better targeted than what we currently have, though.

do we use the .txt anywhere? The docker images (which don't include groff) don't
seem to fail for this.

[dak, 2020-02-29 22:28:14]

On 2020/02/29 22:13:42, hanwenn wrote:
> added 
> 
>     * Remove unused GROFF and LD autoconf vars
> 
> to the commit msg.
> 
> https://codereview.appspot.com/547690053/diff/567300043/config.make.in
> File config.make.in (right):
> 
>
https://codereview.appspot.com/547690053/diff/567300043/config.make.in#newcode48
> config.make.in:48: GROFF = @GROFF@
> On 2020/02/29 22:03:03, dak wrote:
> > These sort of drive-by changes without any mention in commit message or
issue
> > make it a bit harder to review.  I find
> > 
> > stepmake/stepmake/documentation-rules.make:     troff -man -Tascii $< |
grotty
> > -b -u -o > $@
> > 
> > which is awkward by not using anything actually tested for.  Nevertheless,
use
> > of grotty probably requires the presence of the groff package.  The test
> should
> > likely be better targeted than what we currently have, though.
> 
> do we use the .txt anywhere? The docker images (which don't include groff)
don't
> seem to fail for this.

What do you mean with "fail for this"?  Since the results of the test for groff
are not actually used anywhere, removing the test will not change anything. 
Removing the groff package from the docker image, in contrast, could be
problematic.  I really have no clue about the workings of our build system.

[hanwenn, 2020-02-29 22:35:49]

On 2020/02/29 22:28:14, dak wrote:
> On 2020/02/29 22:13:42, hanwenn wrote:
> > added 
> > 
> >     * Remove unused GROFF and LD autoconf vars
> > 
> > to the commit msg.
> > 
> > https://codereview.appspot.com/547690053/diff/567300043/config.make.in
> > File config.make.in (right):
> > 
> >
>
https://codereview.appspot.com/547690053/diff/567300043/config.make.in#newcode48
> > config.make.in:48: GROFF = @GROFF@
> > On 2020/02/29 22:03:03, dak wrote:
> > > These sort of drive-by changes without any mention in commit message or
> issue
> > > make it a bit harder to review.  I find
> > > 
> > > stepmake/stepmake/documentation-rules.make:     troff -man -Tascii $< |
> grotty
> > > -b -u -o > $@
> > > 
> > > which is awkward by not using anything actually tested for.  Nevertheless,
> use
> > > of grotty probably requires the presence of the groff package.  The test
> > should
> > > likely be better targeted than what we currently have, though.
> > 
> > do we use the .txt anywhere? The docker images (which don't include groff)
> don't
> > seem to fail for this.
> 
> What do you mean with "fail for this"?  Since the results of the test for
groff

You'd expect failures to occur if groff wasn't installed; but come to think of
it, I'm only testing the regtests through docker (there is an obscure error with
make doc that I haven't fixed yet.), so I may not have full coverage. However, I
think we don't generate .txt files, except maybe in the "make dist". Let me test
that.

> are not actually used anywhere, removing the test will not change anything. 
> Removing the groff package from the docker image, in contrast, could be
> problematic.  

> I really have no clue about the workings of our build system.

[hanwenn, 2020-02-29 22:38:18]

https://codereview.appspot.com/557560044/

[hanwenn, 2020-03-01 16:58:03]

rebase

[hanwenn, 2020-03-06 20:11:14]

commit 35501a692374ddff7c367967b7c9d0dbc5c6e626
Author: Han-Wen Nienhuys <hanwen@lilypond.org>
Date:   Thu Mar 5 11:20:15 2020 +0100

    build cleanups.