| 1 |
On Wednesday 25 September 2002 16:29, Moritz Schulte wrote: |
| 2 |
> Paul de Vrieze <pauldv@××××××.nl> writes: |
| 3 |
> > I know that, although one would probably still want a hurd that |
| 4 |
> > complies to the linux interface, (provides the linux chasis) [...] |
| 5 |
> |
| 6 |
> What do you mean with "linux interface"? I guess you mean the system |
| 7 |
> call interface of Linux. The Hurd itself cannot conform to that, |
| 8 |
> because of it's design. |
| 9 |
> |
| 10 |
> In Linux - and in Unix in general - you can use the system services |
| 11 |
> via system calls, a uniform way. This includes accessing files, |
| 12 |
> creating processes, networking, etc. In the Hurd, there are seperate |
| 13 |
> "servers" for such things and the services are used via RPCs. |
| 14 |
> |
| 15 |
> But since GNU aims POSIX compliance, of course, we have a POSIX |
| 16 |
> interface , which is encapsulated in glibc. So, POSIX compliance |
| 17 |
> programs should compile on GNU/Hurd just as they do on GNU/Linux - |
| 18 |
> with few exceptions. |
| 19 |
> |
| 20 |
> In case you were referring to a binary compatibility - which means: |
| 21 |
> having the same ABI - that does not exist at the moment although there |
| 22 |
> have been some discussions on that topic on the Hurd lists. |
| 23 |
> |
| 24 |
|
| 25 |
What I mean is binary compatibility meaning that you could one day decide to |
| 26 |
tell your bootloader (e.g. grub) not to load your favourite linux kernel, but |
| 27 |
to load another kernel (in the broad sense) such as HURD, and everything |
| 28 |
would still work. That doesn't mean it needs to be the most optimal way, but |
| 29 |
it must run. |
| 30 |
|
| 31 |
Paul |
| 32 |
|
| 33 |
-- |
| 34 |
Paul de Vrieze |
| 35 |
Junior Researcher |
| 36 |
Mail: pauldv@××××××.nl |
| 37 |
Homepage: http://www.devrieze.net |
Archives