1. The more things can be standardized, the less they have to be reinvented
with each new implementation. I approve of /usr/lib/xxx where xxx is
something mnemonic. I've already done that for each major project I have
on my development machines.
2. I have my own "standard" header files. For user programs it's called
libc.h and for modules it's device-libc.h. I've used these names since
1983 and I don't have to think about what I want to include when I start a
new project. It's one of the best "inventions" I've ever made.
3. It's too easy to get a dynamic major number and just as easy to look in
/proc/devices (even with a script file) that it doesn't justify the
potential confusion (or worse) when collisions occur.
Norm
At 11:25 AM 7/26/2000 -0700, David Schleef wrote:
>
>I'd like to get some comments on some of the following directions:
>
>1. Creation of /usr/lib/realtime/include and population with header
> files.
>
> Currently, RTAI/RTLinux neutral projects (like RTnet and Comedi)
> don't have an adequate place to install header files for compilation
> with kernel modules. /usr/include is not appropriate, because it
> conflicts with existing files. (For example, RTnet can't use
> sockets.h, etc. because it aleady exists.) Comedi (with comedi.h)
> can use #ifdef __KERNEL__, but I don't feel that is a full
> solution.
>
> Also, I'd like to see RTAI and RTLinux header files installed
> into /usr/lib/realtime/include/rtai and .../rtlinux, assuming
> that the respective maintainers are interested in the required
> amount of source-compatibility. (Don't need to be perfect here--
> we're all still learning.)
>
> Anything that has a user-space interface, such as fifos, will
> still want to install a /usr/include header file.
>
> (People familiar with cross compiling will realise that
> /usr/lib/realtime is probably the appropriate directory for
> this.)
>
>2. Development of some standard header files, such as stdlib.h, and
> a few of the libc functions that people have asked about on the
> mailing list.
>
> (I think this just needs someone to create the project.)
>
>3. Allocation of a real-time misc device major number.
>
> It appears that an increasing number of projects need access
> to a ioctl()-like interface, like Tomasz's shared memory and
> RTnet. Currently, both of these use unallocated/experimental
> Misc-device numbers, which eventually will lead to conflicts.
> I'd like to get these numbers permanently allocated, and I
> think a new major specifically for this purpose is a good idea,
> since it allows greater flexibility than using major 10,
> including autoloading of appropriate modules.
>
>Anyway, i don't have too strong of an opinion on any of this, just
>that it seems like a good way to solve some of the problems I've
>been having, and generally making things neater and cleaner. let
>me know if you have better ways to solve similar problems.
>
>
>
>dave...
>
>-- [rtl] ---
>To unsubscribe:
>echo "unsubscribe rtl" | mail [EMAIL PROTECTED] OR
>echo "unsubscribe rtl <Your_email>" | mail [EMAIL PROTECTED]
>---
>For more information on Real-Time Linux see:
>http://www.rtlinux.org/rtlinux/
>
>
-- [rtl] ---
To unsubscribe:
echo "unsubscribe rtl" | mail [EMAIL PROTECTED] OR
echo "unsubscribe rtl <Your_email>" | mail [EMAIL PROTECTED]
---
For more information on Real-Time Linux see:
http://www.rtlinux.org/rtlinux/
