Questionable statement in article

[Joel Dahl]

Mon 2004-08-09 klockan 22.15 skrev Ceri Davies:
> On Mon, Aug 09, 2004 at 01:39:40PM -0400, John Baldwin wrote:
> > On Monday 09 August 2004 01:29 pm, Joel Dahl wrote:
> > > Mon 2004-08-09 klockan 14.07 skrev Ceri Davies:
> > > > On Mon, Aug 09, 2004 at 01:40:28PM +0200, Devon H. O'Dell wrote:
> > > > > Okay, this is getting really ridiculous, and the statement is false. It
> > > > > would be rather simple to figure out which syscalls FreeBSD was unable
> > > > > to translate and thereby make a certain piece of software fail to run
> > > > > on FreeBSD. For instance, there are certain socket options in Linux
> > > > > that are not avaialble on FreeBSD and cannot be emulated. Software that
> > > > > makes use of these options will _not_ run on FreeBSD.
> > > >
> > > > Firstly, I'll note that the article is talking about BSD, not FreeBSD.
> > > >
> > > > > A more accurate statement would be:
> > > > >
> > > > > FreeBSD_Compilable_Code + FreeBSD_Binaries + FreeBSD_Emulatable(Linux)
> > > > > > Binaries(Linux)
> > > > >
> > > > > You can't blindly make this statement, however, without first proving
> > > > > the following:
> > > > >
> > > > > Binaries(Linux) - FreeBSD_Emulatable(Linux) < FreeBSD_Compilable_code +
> > > > > FreeBSD_Binaries.
> > > > >
> > > > > Now, once you factor in the SVR4 compatibility and others, this
> > > > > statement gets exceedingly difficult to make. When somebody wants to
> > > > > audit the amount of binaries that will run on FreeBSD and get a number,
> > > > > let me know.
> > > >
> > > > Since SVR4 gets bundled on the right hand side of the equation above,
> > > > along with BSDI, IBCS2, Interactive Unix, SCO Unix, SCO Xenix, and
> > > > Solaris (this selection just from the i386 NetBSD port and excluding
> > > > other free BSDs), the statement becomes slightly easier to make, I
> > > > think.
> > > >
> > > > > Also, it's interesting to note that OpenBSD will do the same -- it has
> > > > > Linux syscall translation as well -- it will also run FreeBSD binaries.
> > > > > Does this mean that OpenBSD has a conceviably larger amount of binaries
> > > > > that will run on it than FreeBSD?
> > > >
> > > > Well, yes.
> > > >
> > > > Ceri
> > >
> > > Whoops, my intention was not to cause any hard feelings with my original
> > > question about the statement. I'm just trying to make our docs correct.
> > >
> > > :)
> > >
> > > As I see it, the statement can't be confirmed as true OR false, and
> > > should therefore be removed, if someone with commit privileges agree. To
> > > remove the "As a result, more software is available for BSD than for
> > > Linux." -part would be perfectly sufficient. :)
> > 
> > FWIW, it seems to me that the statement has more downside potential ("FREEBSD 
> > LIES ON ITS WEBSITE, FILM AT 11" (if we are ever caught out on it b/c, in 
> > fact, there are Linux binaries that FreeBSD doesn't run or at least run well) 
> > than upside.
> 
> I've discussed this with Devon offlist - how do people like this patch?
> 
> Index: article.sgml
> ===================================================================
> RCS file: /home/ncvs/doc/en_US.ISO8859-1/articles/explaining-bsd/article.sgml,v
> retrieving revision 1.12
> diff -u -r1.12 article.sgml
> --- article.sgml	8 Aug 2004 13:43:54 -0000	1.12
> +++ article.sgml	9 Aug 2004 20:13:07 -0000
> @@ -529,9 +529,11 @@
>  	</listitem>
>  
>  	<listitem>
> -	  <para>BSD can execute Linux code, while Linux can not execute BSD
> -	    code.  As a result, more software is available for BSD than for
> -	    Linux.</para>
> +	  <para>BSD can execute most Linux binaries, while Linux can not execute BSD
> +	    binaries.  Many BSD implementations can also execute binaries
> +	    from other UNIX-like systems.  As a result, BSD may present an
> +	    easier migration route from other systems than
> +	    Linux would.</para>
>  	</listitem>
>        </itemizedlist>
>      </sect2>
> 
> Ceri

The patch looks great.

-- 
regards,
Joel Dahl