On 7/4/07, Joseph E. Hoag <[EMAIL PROTECTED]> wrote:
>
>
>
> However, if we decide that human-readable communications (i.e., XML
> messages) need to be supported (in addition to byte streams), then it will
> probably become important to get align class names and parameter names
> between languages.  This is accomplished more easily if all languages
> generate off of a single class description file.
>
> Let's be realistic, though -- tweaking class names and method
> names/signatures in a library is more or less the same as replacing the
> entire library in terms of the pain that it will cause developers.  If the
> Java library is tweaked to look like the C++ library (in terms of
> class/method naming), then it essentially becomes a new library, and users
> (like TagCentric and Rifidi) will need to spend considerable time re-writing
> to conform to the new library.
>

I am not sure I agree with that... the actual names used when binding
to a library is not hard to change. It is much harder to switch
libraries altogether, since with a new library comes a different class
hierarchy, set of interfaces, memory management strategies,
synchronization issues, etc.  Changing names is a search and replace
operation. Changing libraries has an aspect of design to it.

I think it would be a mistake for the Java library to be forced to
"look like" the C++ library. Certainly the Perl library to be
contributed doesn't look anything like the Java or C++ library. Each
language is different so naturally the APIs will be different. That
said, the underlying concepts are invariant, since all these APIs end
up talking binary LLRP. So there can be some level of commonality.
Probably the names of enumerations messages, parameters and fields is
about all we can hope for. Even then, I can imagine issues of
camel-casing vs. lowercase, etc. problems where the culture for a
particular language will dictate. When that cannot be automatically
inferred from standard names, we could conceivably provide
alternatives in the schema.

> Do we really need to develop a de-facto XML LLRP messaging standard (with
> standard element names and formats) to go along with the byte-stream LLRP
> messaging standard?  This is debatable, and may cause more problems than it
> solves.  But if we decide that the answer is yes, then we probably want to
> merge the class description files sooner rather than later.
>

Having been using this format for the last several months, I find it
indispensible. Having a textual represenation of LLRP, along with
XPath, and some code to manage the TCP link, has served us like a
domain-specific programming language tailored to LLRP. And the social
aspect of being able to communicate about LLRP in an unambigous is
hard to overstate.

What kinds of issues do you foresee with the abstract schema?

-- John.

-------------------------------------------------------------------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourceforge.net/powerbar/db2/
_______________________________________________
llrp-toolkit-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/llrp-toolkit-devel

Reply via email to