exporting functions

[Sébastien Bourdeauducq]

Hi,

<on-list>

On 03/03/2012 12:20 PM, Ralf Corsepius wrote:
>> By the way, what do you think about my "export functions" RTEMS patches?
>> https://www.rtems.org/bugzilla/show_bug.cgi?id=2030
>> https://www.rtems.org/bugzilla/show_bug.cgi?id=2031
> I actually haven't looked into all details of these PRs and only had a
> brief glimpse into them.
>
> However, when we met in Nuremberg last Wednesday, Joel and I had a brief
> (face-to-face) conversation about such kind of "export requests" in
> general and agreed upon to be having to be careful about them.
>
> What I mean is, we need to carefully check whether these functions are
> actually useful in longer terms and should be exported, or if such
> "export requests" are attempts to "legalize" what were API violations -
> Or if these request are symptoms of bugs/deficiencies elsewhere.

Well, in our case it's easy:
1) Implementing cp/mv/... is tedious on POSIX systems due to the lack of 
libc support. In fact, many Unix applications use system() in this case. 
If you really do not want to export those shell functions, an 
alternative is to add system() to RTEMS. I'm OK with both options, I 
just want it to work without messy duplicating of cp/mv/... code.
2) The second one (bootp functions) simply fixes an internal RTEMS problem.

Thanks,
Sébastien