[petsc-dev] getting rid of TESTEXAMPLES_XXX_XXX stuff

[Barry Smith]

On Dec 2, 2013, at 10:10 PM, Barry Smith <bsmith at mcs.anl.gov> wrote:

> 
> On Dec 2, 2013, at 8:49 PM, Jed Brown <jedbrown at mcs.anl.gov> wrote:
> 
>> Barry Smith <bsmith at mcs.anl.gov> writes:
>> 
>>>  There reason for TESTEXAMPLES_C TESTEXAMPLES_etc is that make
>>>  doesn’t have if tests. Since we have shifted to gnumake can we
>>>  eliminate all these variables and handle the cases with appropriate
>>>  if tests in the makefiles? Then we could also handle more
>>>  complicated situations like if parmetis, 64 bit indices,
>>>  datafilespath.
>> 
>> If we continue to specify tests as targets in makefiles, we don't
>> necessarily need ugly if tests.  One option would be to use
>> target-specific variables.
>> 
>> runex123: REQUIRE = parmetis mumps 64-bit-indices

   Ok, I don’t understand this syntax at all, there is no mention of REQUIRE used in this way in the gnumake manual.  How are you proposing to use this variable?

>> runex123:
>> 	${MPIEXEC} -n 2 ./ex123 …
>> 
>   I used to hate gnumake but if it supports the syntax above that is much better than our current approach.
>> 
>> I don't know how to avoid needing to list all the targets once, other
>> than to have a script that parses the makefiles and makes that list
>> (easy, and can update automatically like the $PETSC_ARCH/conf/files that
>> is relatively transparent).
> 
> tests: $(shell grep ":" makefile | grep runex | grep -v makefile | sed "s/://g” )
> 
> Unfortunately it is not that simple because for each runexX we need to compile the example and we only want to compile it once for all its uses and we need to rm it as soon as it is no longer needed. Perhaps we can list them in the correct order 
> 
> ex1:  REQUIRE = ccx
> 
> runex1: REQUIRE = ccx
> 	stuff
> 
> runex1: REQUIRE = ccx ddy
> 	stuff
> 
> rmex1: 
> 	${RM} -f ex1
> 
> then generate at make time with shell the entire list of rules to run starting at the top
> 
>