[alicebot-archcomm] Posix Standards

[Noel Bush]

Another drawback to Posix is that it seems to be unavailable online.  I
have a problem with the idea of hooking up with a standard which is
purposely exclusionary -- if we have a committee member without the
means to obtain a dead-tree copy of the standard then that person can't
participate in the discussion.

I also don't think that the specific details of disk I/O or process
management are part of our scope; these may be wholly irrelevant or even
senseless in a given application of AIML.  It would be unfortunate to
have to call a well-engineered AIML interpreter "non-compliant" because
it didn't implement a host of irrelevant operating system hooks.

I tend to think that what we really need is a much simpler extension
mechanism that lets someone put whatever they want inside a tag that is
qualified with a namespace.  That makes it easy.

<extension namespace="http://www.php.net">
...some php code here...
</extension>

Then it's just necessary to have a good way of agreeing on namespaces.

On Thu, 2002-03-14 at 07:00, Dr. Richard S. Wallace wrote:
> It has become apparent to several of us that the extensibility of AIML with
> <system> and <javascript> present several problems, not least of which is
> portability between various implementations and platforms.
> 
> Linux succeeded in part because the developers coded an existing standard,
> the Posix operating system standard.
> 
> I would think that we could "inherit" a lot of nice work in operating
> systems interfaces, is we also chose to follow the Posix standard.
> 
> Some examples of stuff in the standard:
> 
> F = open() : open a file
> close(F) : close it
> read(F) : low-level read
> write(F, data) : low level write
> stat(F) : return 0 is the file F exists, 1 otherwise.
> pid : returns process id.
> tmpnam : Generate unique temporary file name
> kill(p) : Send signal to process p.
> 
> also more stuff dealing with multiple processes, network sockets, and
> waiting for I/O interrupts.
> 
> Our <system> tag is already quite close to the Posix system(3)
> specification.  I intentionally trying to resist the urge to make up any new
> XML syntax for Posix concepts.  Maybe someone has done this already?  If
> not, we could consider making up tags like <tmpnam/>, <kill pid="10"/> or
> even the Unix <select>.  In any case, I'd like to keep the converation for
> now focused on whether or not adopting Posix standards for AIML is a good
> idea, not what the tag syntax will be.
> 
> Our <gossip> tag might be defined as:
> <category>
> <pattern>GOSSIP *</pattern>
> <template>
> <set var="filename">gossip.txt</set>
> <set var="F"><open mode="a"><get var="filename"/></open></set>
> <write var="F"><star/></write>
> <close>F</close>
> </template>
> </category>
> 
> 
> One minor drawback to the Posix standard: Microsoft does not comply.  But we
> would probably not need the entire standard, and the parts we do need are
> likely to have equivalent functions in the world of Microsoft.
> 
> Dr. Rich
> 
> 
> _______________________________________________
> alicebot-archcomm mailing list
> alicebot-archcomm@list.alicebot.org
> http://list.alicebot.org/mailman/listinfo/alicebot-archcomm
>