vxworks header files vs. unix header files
Jeffrey W Percival
modules use sockets, time structures, and many, many things defined in
the standard Unix header files (struct tm, struct hostent, and on and on).
Several reliable and experienced vxworks users have told me not to use
unix header files, and should use only vxworks header files. The
problem is, the vxworks header files do not define all of the
"standard" things found in unix files.
How is this problem handled? Do I hunt through all the unix header
files, seeking out each and every vx-absent definition and copying it
to my own header file?
Is there some "foo.h" out there that defines all the vx-absent structs
and types?
--
Jeffrey W Percival (j...@larry.sal.wisc.edu) (608)262-8686
hwajin bae
>Several reliable and experienced vxworks users have told me not to use
>unix header files, and should use only vxworks header files. The
>problem is, the vxworks header files do not define all of the
>"standard" things found in unix files.
probably because those "standard" things are not supported in vxworks.
vxworks is not unix. i wouldn't expect it to support all the "standard"
things in unix. alot of similarities exist between unix and vxworks,
especially in networking area, but they're two very different operating
systems.
copying definitions of things from unix to vxworks header and using them
as is may or may not work. if it works at all, it'll only be by accident,
not by design.
hwajin
Jeffrey W Percival
>In article <1993Sep24.2...@sal.wisc.edu> j...@bernie.sal.wisc.edu (Jeffrey W Percival) writes:
>>Several reliable and experienced vxworks users have told me not to use
>>unix header files, and should use only vxworks header files. The
>>problem is, the vxworks header files do not define all of the
>>"standard" things found in unix files.
>
>vxworks is not unix.
>...
>copying definitions of things from unix to vxworks header and using them
>as is may or may not work. if it works at all, it'll only be by accident,
>not by design.
I see. I think my confusion is understandable, when I see comments like this
in vxworks header files:
/* systime.h - UNIX-style timeval structure defenition */
/* the following conditional is to avoid conflict with SunOS header...
Clearly, *some* unixy definitions are duplicated, and more than a nod
is given to collision with unix header files, so I feel that Mr Bae's
disclaimer above may be a little too stringent.
Anyway, the point is well taken and I thank Mr Bae for his helpful comments.
Medi Montaseri
Its called host/target issue. You are developing on a host for a target.
If your target is a Unix box, then sure, be my guest, use all the Unix libs and system
calls (hope you are getting them from the same unix vendor). But if your target
is some box running vxworks or joe_blow operting system, then how do you
expect these libs or system calls to work for you. vxworks is not unix, but their
marketing people are good. Their file system sucks. No virtual memory. But it allows you
to download stuff using sockets.
Medi
--
| Medi Montaseri "The fact that it works, is immaterial." |
| mont...@ntmtv.com |
| me...@earth.milkyway_galaxy.local_group.local_supercluster.universe |
+--------------------------------------------------------------------+
Hwajin
>vxworks is not unix, but their
>marketing people are good.
>Their file system sucks. No virtual memory. But it allows you
>to download stuff using sockets.
couldn't have said it better myself.
hwajin
-- not speaking for wrs, etc.