Dmitry Eremin-Solenikov(lumag) replied on github web page:
example/traffic_mgmt/Makefile.am
line 6
@@ -1,9 +1,5 @@
include $(top_srcdir)/example/Makefile.inc
-bin_PROGRAMS = odp_traffic_mgmt$(EXEEXT)
-odp_traffic_mgmt_LDFLAGS = $(AM_LDFLAGS) -static
-odp_traffic_mgmt_CFLAGS = $(AM_CFLAGS) -I${top_srcdir}/example
+bin_PROGRAMS = odp_traffic_mgmt
Comment:
@muvarov The mentioned commit is plain wrong, starting from the
`AC_SUBST([EXEEXT])`, which is not necessary. All `_PROGRAMS` variables will be
automatically extended with `$(EXEEXT)` during `Makefile.am` -> `Makefile.in`
conversion by Automake, it is stated in the documentation. In our case this
suffix is only necessary if `dirA` uses (in `TESTS` variable) program built in
`dirB`. For example, like `test/linux-generic/Makefile.am` does for
`validation/api/shmem/shmem_linux` or `ring/ring_main`. Oh, wait. Where is the
`$(EXEEXT)` suffix in the former case?
> muvarov wrote
> point is to name programs with extension like test.exe and run this under
> make check. Commit 0274087a added that.
>> Dmitry Eremin-Solenikov(lumag) wrote:
>> What for? `_PROGRAMS` will automatically receive `$(EXEEXT)` if necessary,
>> so do `$(TESTS)`. What purpose do these `$(EXEEXT)` surve at this point?
>>> muvarov wrote
>>> Looks like some misunderstanding here. I wrote that we should keep ${EXEEXT}
>>>> muvarov wrote
>>>> @lumag I still do not understand do you remove extensions? That link
>>>> http://www.gnu.org/software/automake/manual/html_node/EXEEXT.html says
>>>> that there is no way to do it anyhow different. I think removing will
>>>> break existence functionality.
>>>>> Dmitry Eremin-Solenikov(lumag) wrote:
>>>>> It is used as is by linux-dpdk. Other platforms may also benefit from
>>>>> atomic-related checks, in which case this test should just be
>>>>> generatlized.
>>>>>> Dmitry Eremin-Solenikov(lumag) wrote:
>>>>>> I will drop remaining EXEEXT from examples, where necessary. Looks like
>>>>>> I skipped some. Stripping EXEEXT from test (and platform/l-g/test) will
>>>>>> come in a separate patch.
>>>>>>> Dmitry Eremin-Solenikov(lumag) wrote:
>>>>>>> It will fail with this PR. I should fix that.
>>>>>>>> muvarov wrote
>>>>>>>> make check will build it. Optional change.
>>>>>>>>> muvarov wrote
>>>>>>>>> this patch needs to be revising. EXEEXT somewhere is remove somewhere
>>>>>>>>> exist. EXEEXT was added by Kalrey to run make check on their platform
>>>>>>>>> where they can not run binaries without EXEEXT. I think we should
>>>>>>>>> leave extensions in the code.
>>>>>>>>>> muvarov wrote
>>>>>>>>>> to compile tests you need odp library compiled. To execute
>>>>>>>>>> performance tests examples are needed to be compile. "." defines
>>>>>>>>>> order for parallel make. Moving test_common to the top should bake
>>>>>>>>>> parallel make.
>>>>>>>>>>> muvarov wrote
>>>>>>>>>>> this atomic things are specific to linux-generic. Are you panning
>>>>>>>>>>> to use they anywhere else?
>>>>>>>>>>>> Dmitry Eremin-Solenikov(lumag) wrote:
>>>>>>>>>>>> @Bill-Fischofer-Linaro updated
>>>>>>>>>>>>> Dmitry Eremin-Solenikov(lumag) wrote:
>>>>>>>>>>>>> just for the sake of 80 chars limit? Fine, I will update this.
>>>>>>>>>>>>>> Bill Fischofer(Bill-Fischofer-Linaro) wrote:
>>>>>>>>>>>>>> What's wrong with factoring this like:
>>>>>>>>>>>>>> ```
>>>>>>>>>>>>>> VALIDATION_TESTDIR=platform/$ODP_PLATFORM/test/validation
>>>>>>>>>>>>>> PLATFORM_VALIDATION=${TEST_SRC_DIR}/../../$VALIDATION_TESTDIR
>>>>>>>>>>>>>> ```
>>>>>>>>>>>>>> That seems to solve the line length problem cleanly.
>>>>>>>>>>>>>>> Dmitry Eremin-Solenikov(lumag) wrote:
>>>>>>>>>>>>>>> Not in a clean way
>>>>>>>>>>>>>>>> Bill Fischofer(Bill-Fischofer-Linaro) wrote:
>>>>>>>>>>>>>>>> Checkpatch flags this line as being > 80 chars. Can it not be
>>>>>>>>>>>>>>>> split into two lines or otherwise shortened?
https://github.com/Linaro/odp/pull/213#discussion_r144600790
updated_at 2017-10-13 16:32:36
