Here are some notes to help you develop code for Xmlrpc-c.  I include
as "developing" debugging to figure out why Xmlrpc-c doesn't work for
you.

CODE LIBRARY
------------

The master Xmlrpc-c source code tree is the CVS repository on
Sourceforge.  Anybody can read it; only the maintainer can commit to
it.  If you're not the maintainer, simply use a 'cvs diff' command in
your CVS working directory to create a patch file that embodies your
changes and email that to the maintainer.  He can easily apply that
patch to his own CVS working directory and then commit the changes.


MAKE VARIABLES
--------------

You can pass make variable values to GNU Make to change the build.
There are two common ways to do this:

  1) Like this:

     $ make MYVAR=myvalue

  2) Via an environment variable, like this:

     $ MYVAR=myvalue make

     or
 
     $ export MYVAR=myvalue
     $ make

See GNU Make and shell documentation for details.

In Xmlrpc-c make files, there are two make variables that add
arbitrary options to every compile command: CADD and CFLAGS_PERSONAL.

They both do the same thing.  CADD is meant to be set on an individual
make command, whereas CFLAGS_PERSONAL is meant to be a long-lived
environment variable.  CFLAGS_PERSONAL is for flags you like on all
your compiles, but maybe others don't.

One of my favorite CADD settings is CADD=--save-temps.  To the GNU
Compiler, --save-temps means to create, in addition to the object
code, a file containing the intermediate preprocessed C code and a
file containing the intermediate assembler source code.  I can use
that to debug certain things.

The Xmlrpc-c build uses -g by default with Gcc, so you don't need to
use CADD to get debugging symbols in your object code.


There's also LADD for linker options.


CODE STYLE
----------

The maintainer is pretty particular about coding style, but doesn't
require anyone to submit code in any particular style.  He changes
what he thinks isn't maintainable enough as submitted.  You could do
him a big favor, though, and reduce the chance of him introducing bugs
into your code, but trying to copy the style you see in existing code.
The major theme is high level programming -- closer to English prose
and further from machine instructions.

Probably the most important thing is not to use tabs.  Tabs are
actually quite common in Unix C programming, but apart from tradition,
they are a bad idea.  They don't look the same to everyone.  A person
must suffer an additional configuration step -- setting up tab stops
in order to see the code the right way.  Spaces, on the other hand,
look the same to everyone.  Very old editors made it easier to compose
with tabs than with spaces, but with modern ones, there is no
difference.

The maintainer tries to catch all tabs in code submitted to him and
convert them to spaces, but this often leaves the code incorrectly
indented.  Better to give him code that already has the right number
of spaces explicitly.