[PATCH 01/31] shell: Use #include "..." for local header files

Chris Johns chrisj at rtems.org
Tue Oct 9 12:50:46 UTC 2018


On 8/10/18 10:35 pm, Sebastian Huber wrote:
> On 09/10/2018 03:41, Chris Johns wrote:
>> On 8/10/18 3:43 am, Sebastian Huber wrote:
>>> Update #3375.
>> Why?
> 
> To keep the count of  command line arguments small. These three header files are
> used only in the shell directory. They are not installed.

Thanks, I see the issue now. This is a limitation of building this way. Waf
allows options to added based on grouping of source.

> The <err.h> and <fts.h> are candidates for Newlib. It is a bit of work to make
> them available via Newlib since they are already provided by Cygwin. This patch
> set is a preparation for this:
> 
> https://sourceware.org/ml/newlib/2018/msg01004.html

Nice.

>>
>> Some of these files are imported ....
>>
>>> ---
>>>   cpukit/libmisc/Makefile.am              | 1 -
>>>   cpukit/libmisc/shell/cmp-ls.c           | 2 +-
>> ... for this example this file.
> 
> I don't think the changed #include statements will cause major problems if
> someone decides to merge upstream patches.

Yes I agree. The change is fine.

Chris



More information about the devel mailing list