Re: ANN: LuaXMLRPC 1.0 Alpha

[Jay Carlson <nop@...>]

Andre Carregal wrote:
> Jay,
>
> First of all we deeply apologize for the recent inconvenience. You are more
> than right about our announcement of the Kepler LuaXMLRPC over your original
> work.
>
> We also apologize and recognize we should have done our homework contacting
> you anyway before we began implementing our version. We wrongly assumed your
> implementation was stopped since 2001, our fault.

Well, it was---I only started working on again it last weekend.  So it 
was a reasonable assumption that it was dead.  However, I am allowed to 
tease you because you didn't send me mail. :-)  (Is there an etiquette 
of this somewhere?)

> Concerning the library name itself we will change ours immediately if you
> prefer, but we would also gladly join efforts if that interests you. 

I would much rather join efforts.  There are some things I work on for 
fun; there many infrastructure projects I work on to support them.  It 
was a little fun to write my XML-RPC binding, but I did it so I could 
write other things.  I just want there to be a good XML-RPC binding for 
Lua.  I would prefer somebody else to maintain it.

Put another way, I consider "convincing somebody else to write 
something" to be a good implementation strategy.  Especially if you're lazy.

> Our
> XMLRPC binding is going to be very integrated with Kepler and your
> experience with the protocol and the "Lua tables / XML structures" details
> would be a great help. Please let us know your position on the possible
> merging.

Kepler sounds interesting.  I'd like to merge.

I'll send out a copy of my Lua 5.0 XMLRPC system shortly.  Meanwhile, 
here's my type mapping:

====
XML-RPC's types don't map directly to the Lua types.  A table with the
key XMLRPC.type is a special table.  The value of the key describes
which XML-RPC type the table represents.  To construct these special
values, you should use their constructor functions rather than table
constructors.

<string>: Lua string
<i4>, <integer>: Lua number
<array>: a Lua table with a key of 1 present, or with ["n"]=0
<struct>: a Lua table (without the type key)

<boolean>: true or false

<double>: Build with XMLRPC.double(123.78); access with v.double
<base64>: Build with XMLRPC.base64("BaSe64--"); access with v.base64
<dateTime.iso8601>: Build with XMLRPC.dateTime(string); access with 
v.dateTime
====

Arrays are still a little tricky, as you can see.  This reminds me that 
I also need to add a "what XMLRPC type is this" function to the public 
interface.

> Please also note that the Kepler and the Lua team are separated. We cannot
> be "touchy about anybody else reusing the name 'lua'" because that name is
> not ours.

Actually I think it was Luiz who was mostly doing that; Roberto is 
listed on your credits page though.

Perhaps some of this is me longing for a benign dictator for projects 
outside the Lua core.  I think people may interpret your work that way 
no matter how you disclaim it.

> Once more, we are very sorry for the nuisance.

I don't think you need to apologize, but I will forgive you if you 
promise to never generate XML in application code ever again. :-)

I think Lua needs something like Perl's XML::Writer ( 
http://search.cpan.org/~edavis/XML-Writer-0.4.1/Writer.pm ) or the .NET 
XmlWriter ( 
http://msdn.microsoft.com/library/default.asp?url=/library/en-us/cpref/html/frlrfSystemXmlXmlWriterClassTopic.asp 
) to make it easy to do this right, regardless of what the internal form is.

Jay