Re: [PROBLEM] perf requires python-devel to compile

[Michael Witten <mfwitten@gmail.com>]

On Mon, 15 Apr 2013 21:26:40 -0700, David Ahern wrote:

> On 4/15/13 7:14 PM, Namhyung Kim wrote:
>>> Makefile:755: The path '/usr/bin/python-config' is not executable.
>>> Makefile:755: *** Please set 'PYTHON_CONFIG' appropriately.  Stop.
>>>
>>> The problem is that I didn't have python-devel package installed and
>>> get-executable-or-default decides to error out instead of letting the
>>> Makefile disable Python support.
>>
>> Right.  I think the get-executable-or-default should not error out in
>> this case but just emit a warning and keep building.
>>
>> Does following patch fix your problem?
>>
>> -----------8<-------------8<-------------
>> diff --git a/tools/perf/config/utilities.mak b/tools/perf/config/utilities.mak
>> index 8ef3bd30a549..3b8036f8aca4 100644
>> --- a/tools/perf/config/utilities.mak
>> +++ b/tools/perf/config/utilities.mak
>> @@ -177,7 +177,7 @@ $(if $($(1)),$(call _ge_attempt,$($(1)),$(1)),$(call _ge_attempt,$(2),$(1)))
>>   endef
>>   _ge_attempt = $(if $(get-executable),$(get-executable),$(_gea_warn)$(call _gea_err,$(2)))
>>   _gea_warn = $(warning The path '$(1)' is not executable.)
>> -_gea_err  = $(if $(1),$(error Please set '$(1)' appropriately))
>> +_gea_err  = $(if $(1),$(warning Please set '$(1)' appropriately))
>
> In this case you don't want a warning, you just want python support 
> disabled and move on. I've been getting around this in a minimal install 
> of F18 in a VM using PYTHON_CONFIG=/bin/false; haven't had time to 
> search for the proper solution.

You should probably disable python support more directly:

  make NO_LIBPYTHON=1

That being said, this issue was introduced with the following commit:

  31160d7feab786c991780d7f0ce2755a469e0e5e

namely due to:

  ... Also fix an issue where _get_attempt was called with only
  one argument. This prevented the error message from printing
  the name of the variable that can be used to fix the problem.

specifically:

  -$(if $($(1)),$(call _ge_attempt,$($(1)),$(1)),$(call _ge_attempt,$(2)))
  +$(if $($(1)),$(call _ge_attempt,$($(1)),$(1)),$(call _ge_attempt,$(2),$(1)))

The "missing" argument was in fact missing on purpose; it's a signal
that the error message should be skipped, because the failure was due
to the default value, not the user-supplied value.

That being said, I think there's room for improvement; for instance,
the error handling should perhaps not belong there. I will look into
it presently.

Sincerely,
Michael Witten