Re: [GIT PULL] Follow-up patches of cleanup-makefile

[Akira Yokosawa <akiyks@gmail.com>]

On 2016/06/10 1:17, Paul E. McKenney wrote:
> On Wed, Jun 08, 2016 at 11:36:14PM +0900, Akira Yokosawa wrote:
>> Hi Paul,
>>
>> These are follow-up patches of the previous patch series.
>> They will suppress warning messages and streamline messages displayed
>> during build.
>>
>> I hope you like them.
> 
> I pulled them in, but I am getting the following when I do "make clean":
> 
> Makefile:124: target `SMPdesign/DiningPhilosopher5.pdf' given more than once in the same rule.
> Makefile:124: target `SMPdesign/DiningPhilosopher5TB.pdf' given more than once in the same rule.
> Makefile:124: target `SMPdesign/DiningPhilosopher4part-b.pdf' given more than once in the same rule.
> Makefile:124: target `SMPdesign/DiningPhilosopher5PEM.pdf' given more than once in the same rule.
> Makefile:124: target `advsync/store15tred.pdf' given more than once in the same rule.

I can't reproduce these warnings.
I tested with GNU Make 4.1 and GNU Make 3.81 on Ubuntu 16.04, Ubuntu 14.04 and Ubuntu 12.04.

Which version of make are you using?

> 
> But I was getting them before these changes, so I pushed them out.
> 
> Which raises a question:  What should we be doing to test Makefile changes?
> 
> I am guessing something like this:
> 
> 	git clone git://git.kernel.org/pub/scm/linux/kernel/git/paulmck/perfbook.git
> 	cd perfbook
> 	make -j16
> 	# check output
> 	make distclean
> 	# check output
> 	make -j16 1c
> 	# check output
> 	make distclean
> 	make -j16 hb
> 	# check output
> 
> Is there some way of automating the checking of output?

Well, the output should depend on the environment.
Automating the checking would be not so simple...

> BTW, the reason I was so quick to accept your changes last week was that
> I -really- -really- like the fact that make -j now works.  This makes
> building from an empty archive much nicer.  But the above errors do
> underscore the need to test Makefile changes.  ;-)

Yes. In the past cleanup of build scripts, I made several regressions.

Or should we introduce 'configure' approach to check the existence and
versions of necessary tools?

                                                      Thanks, Akira

> 
> 							Thanx, Paul
> 
>> --- 
>> The following changes since commit ca71bc6cf0bf303257b589d7edf251d627e90ce3:
>>
>>   Fix typo in epigram attribution (2016-06-07 13:30:31 -0700)
>>
>> are available in the git repository at:
>>
>>   https://github.com/akiyks/perfbook.git cleanup-makefile-v2
>>
>> for you to fetch changes up to 35e4d20da570eab7e6ca6a59b92ab438d2efd7b2:
>>
>>   Makefile: Use wildcard for .dot files (2016-06-08 23:05:39 +0900)
>>
>> ----------------------------------------------------------------
>> Akira Yokosawa (3):
>>       build scripts: Suppress warning messages
>>       Hide sources of unused figures
>>       Makefile: Use wildcard for .dot files
>>
>>  Makefile                                           |  58 +++---
>>  SMPdesign/lockdeq.fig                              | 219 ---------------------
>>  .../MoreThanOneValue-15CPU-noresp.eps              |   0
>>  .../MoreThanOneValue-15CPU-noresp.fig              |   0
>>  .../rcuimpl/{ => .unused}/RCUpreemptCounters.fig   |   0
>>  cartoons/{ => .unused}/patterns.fig                |   0
>>  cartoons/{ => .unused}/patterns.svg                |   0
>>  utilities/runfirstlatex.sh                         |   2 +-
>>  utilities/runlatex.sh                              |  16 +-
>>  9 files changed, 32 insertions(+), 263 deletions(-)
>>  delete mode 100644 SMPdesign/lockdeq.fig
>>  rename advsync/{ => .unused}/MoreThanOneValue-15CPU-noresp.eps (100%)
>>  rename advsync/{ => .unused}/MoreThanOneValue-15CPU-noresp.fig (100%)
>>  rename appendix/rcuimpl/{ => .unused}/RCUpreemptCounters.fig (100%)
>>  rename cartoons/{ => .unused}/patterns.fig (100%)
>>  rename cartoons/{ => .unused}/patterns.svg (100%)
>>
> 
>