Re: bug#14560: C Compilation variables present in output Makefiles unconditionally

[Stefano Lattarini]

submitter 14560 Ralf Corsepius <address@hidden>
severity 14560 minor
tags 14560 + moreinfo
thanks

On 06/05/2013 12:03 PM, Stefano Lattarini wrote:
> [+cc bug-automake, so this will be registered in the bug tracker]
> 
> On 06/05/2013 07:16 AM, Ralf Corsepius wrote:
>> On 06/03/2013 09:14 PM, Stefano Lattarini wrote:
>>> We are pleased to announce the GNU Automake 1.13.3 maintenance release.
>>
>> When comparing automake-1.13.2 generated Makefile.ins against
>> automake-1.13.3 generated Makefile.in, in projects which are
>> _not_ using "c" I am observing changes like this one below:
>>
>> --- a/Makefile.in
>> +++ b/Makefile.in
>> ...
>> @@ -109,6 +109,18 @@ AM_V_at = $(address@hidden@)
>>  am__v_at_ = $(address@hidden@)
>>  am__v_at_0 = @
>>  am__v_at_1 =
>> +COMPILE = $(CC) $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) \
>> +       $(CPPFLAGS) $(AM_CFLAGS) $(CFLAGS)
>> +AM_V_CC = $(address@hidden@)
>> +am__v_CC_ = $(address@hidden@)
>> +am__v_CC_0 = @echo "  CC      " $@;
>> +am__v_CC_1 =
>> +CCLD = $(CC)
>> +LINK = $(CCLD) $(AM_CFLAGS) $(CFLAGS) $(AM_LDFLAGS) $(LDFLAGS) -o $@
>> +AM_V_CCLD = $(address@hidden@)
>> +am__v_CCLD_ = $(address@hidden@)
>> +am__v_CCLD_0 = @echo "  CCLD    " $@;
>> +am__v_CCLD_1 =
>>  SOURCES =
>>  DIST_SOURCES =
>>  AM_V_DVIPS = $(address@hidden@)
>> ...
>>
> Yeah, this shouldn't happen.  Not a serious regression thankfully,
> but still unpleasant.
> 
>> So far, I havn't had sufficient time to implement a simple reproducer,
>>
Any progress with this?  I haven't been able to reproduce the bug so
far.  Even if you cannot find a minimal reproducer, a tarball of the
package where you are encountering the issue might help.

>> but I am inclined to believe, automake-1.13.3 inserts c-compiler
>> related vars into Makefile.ins, in cases no C-compiler is being used.
>
> Thanks for the report, I'll try to look into this ASAP (might be
> few days though).  I already have in mind which set of changes
> can be responsible for the regression (that is, those related to
> the fixing of automake bug#14441).
> 

Thanks,
  Stefano