[lsb-discuss] appchk - command line processing

[Robert Schweikert]

Picking this back up.

Wichmann, Mats D wrote:
>> <snip>
>>
>> So I take your point. The current text covers programs sanely but not
>> libraries 
>>     
>
> I'm not even sure the current text conveys the right
> impression,  as we've seen; "If a feature is not
> present, then a conforming application shall not use it."
> could of course be interpreted as including "and in
> fact, we're not even going to try to run because
> that feature is missing" but that's not the way I'd
> take it at a quick look.  Obviously we'll entertain
> discussion on alternative wordings.
>   
I think it should be clearly stated that an application may choose not 
to install or provide an appropriate error message if an attempt is made 
to execute on an x86 CPU which does not support features the application 
needs.
> Anyway, what could we do on the library side?
>   
Maybe we could add some special conditions for libraries.  The library 
itself can be LSB compliant. The certification could state the minimum 
instruction set. For example if parasolid requires SSE2 instruction John 
could certify to be LSB compliant and then state that the minimum x86 
cpu is one that supports SSE2. The app vendor incorporating the library 
is then responsible for the CPU feature check while at the same time 
being assured that the library itself will not give any trouble when the 
app vendor tries to certify his/her own application.
> Distros look like they mainly handle this at
> the package level, that is, at install time.
>
>
> _______________________________________________
> lsb-discuss mailing list
> lsb-discuss at lists.freestandards.org
> http://lists.freestandards.org/mailman/listinfo/lsb-discuss
>   

-- 
Robert Schweikert                       MAY THE SOURCE BE WITH YOU
(robert.schweikert at mathworks.com)                 LINUX
The MathWorks Inc.
Phone : 508-647-2042