On Thu, Aug 31, 2023 at 11:18:06AM +0100, Bruce Richardson wrote:
> When called from automated tools, like meson test, it is often useful to
> skip tests in a test suite, without having to alter the test build. To
> do so, we add support for DPDK_TEST_SKIP environment variable, where one
> can provide a comma-separated list of tests. When the test binary is
> called to run one of the tests on the list via either cmdline parameter
> or environment variable (as done with meson test), the test will not
> actually be run, but will be reported skipped.
> 
> Example run:
>   $ DPDK_TEST_SKIP=dump_devargs,dump_ring meson test --suite=debug-tests
>   ...
>   1/9 DPDK:debug-tests / dump_devargs             SKIP            1.11s
>   2/9 DPDK:debug-tests / dump_log_types           OK              1.06s
>   3/9 DPDK:debug-tests / dump_malloc_heaps        OK              1.11s
>   4/9 DPDK:debug-tests / dump_malloc_stats        OK              1.07s
>   5/9 DPDK:debug-tests / dump_mempool             OK              1.11s
>   6/9 DPDK:debug-tests / dump_memzone             OK              1.06s
>   7/9 DPDK:debug-tests / dump_physmem             OK              1.13s
>   8/9 DPDK:debug-tests / dump_ring                SKIP            1.04s
>   9/9 DPDK:debug-tests / dump_struct_sizes        OK              1.10s
> 
>   Ok:                 7
>   Expected Fail:      0
>   Fail:               0
>   Unexpected Pass:    0
>   Skipped:            2
>   Timeout:            0
> 
> Signed-off-by: Bruce Richardson <bruce.richard...@intel.com>
> Acked-by: Thomas Monjalon <tho...@monjalon.net>
> 
+Tyler

I see this set is failing CI checks due to breaking Windows builds. The
issue seems to be the use of the "strdup" function. I notice in the log
library, that we have a "#define strdup _strdup" macro. Since strdup is
fairly common, widespread function, I think we should consider a more
general approach to it.

Tyler, looking for your input here: should we just globally define strdup
as _strdup for windows in DPDK? Alternatively, some googling indicates that
there is the "_CRT_NONSTDC_NO_DEPRECATE" define which could be used to
enable a whole range of POSIX functions. Should we, or could we, just set
this to ease porting of code over? I'd hate each of our C files to have a
bunch of duplicated #defines at the start to prefix standard unix functions
with "_"s.

Thoughts?
/Bruce

Reply via email to