--- "Richard G. DAVIS" <[EMAIL PROTECTED]> wrote:

> Greg.....
> 
> Is the VistA KERNEL an example of what you have in mind?
> 

Indeed, the VistA Kernel could serve as a foundation for an
implementation of the standard library.

> What, if any, are the distinctive characteristics of the library you
> propose
> that are not present in the VA KERNEL implementation of platform
> independent
> services?

There are a few things. For one, given that the library would be
supported by a special syntax, it would not be necessary to use
extrinsic functions or ask users to load third party libraries (like VA
Kernel). It would also not be possible for users (programmers) to make
illicit use of internal entry points. It would free the implementers to
avoid writing implementations in MUMPS when it is awkward, inefficient,
or simply impossible to do so. The programmer would not have to know or
care whether his/her library call was really "MUMPS under the hood" or
not.

> 
> Moreover, why push these solutions down into the 'language' level of
> abstraction?  Would it not be wiser to create a HIGHER level of
> abstraction ...

I don't see a conflict here. Providing access to basic functionality
like UDP, device handling, asynchronous interrupts, etc. has nothing to
do with implementing high level abstractions. They're just different
pieces of the language puzzle.


> in which these services reside, and where the generality of these
> services
> can have a wider scope than just within the MUMPS (sic) context?
> 

I'm not sure that I follow you here. By "generality" do you mean I
higher level of abstraction, or are you speaking of availability that
goes beyond MUMPS code? In fact, I have some thoughts about
incorporating high level abstractions into the MUMPS language, but did
not mention it here because I see it as an entirely different issue. 
If you are referring to issues like linking MUMPS into applications
written in other languages, allowing classes to be implemented in MUMPS
but used in other environments (like Java, Python or C++) then, again I
agree that this would be a good thing, but it's also quite a separate
issue -- that is except insofar as a standard library would facilitate
linking (in the technical sense) code modules written in MUMPS with
other applications.

> I prefer thinking that attempts to cast the problem at a level that
> makes
> the next solution, the last solution needed for the foreseeable
> future, and
> perhaps beyond.
> 

I guess it just depends on the problem you're trying to solve. If I
want to write a TCP/IP application in MUMPS, then I have to use
features of implementations that go beyond the standard. I don't
consider that an acceptable option. If cannot receive notification that
data has arrived on a socket and am, instead, forced to poll for data,
my implementation will be awkward, slow, and difficult to maintain.
These are the types of issues I believe that could be appropriately
addressed through a standard library. But I suspect you are talking
about something rather different, such as beinf forced to write
relatively low level procedural code when an application could more
naturally be built using a functional or object oriented language. And
as I said before, those are also issues I'd like to see addressed, but
it's best to take them one at a time.

> Regards,
> 
> Richard.
> 


===
Gregory Woodhouse  <[EMAIL PROTECTED]>
"Interaction is the mind-body problem of computing."
--Philip L. Wadler


-------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc. Do you grep through log files
for problems?  Stop!  Download the new AJAX search engine that makes
searching your log files as easy as surfing the  web.  DOWNLOAD SPLUNK!
http://ads.osdn.com/?ad_id=7637&alloc_id=16865&op=click
_______________________________________________
Hardhats-members mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/hardhats-members

Reply via email to