Ophis/doc/tutor3.sgm

<chapter id="ch3-link">
  <title>Headers, Libraries, and Macros</title>

  <para>
    In this chapter we will split away parts of our <quote>Hello
    World</quote> program into reusable header files and libraries.
    We will also abstract away our string printing technique into a
    macro which may be invoked at will, on arbitrary strings.  We will
    then multiply the output of our program tenfold.
  </para>

  <section>
    <title>Header files and libraries</title>

    <para>
      The prelude to our program&mdash;the <filename>PRG</filename>
      information and the BASIC program&mdash;are going to be the same
      in many, many programs.  Thus, we should put them into a header
      file to be included later.  The <literal>.include</literal>
      directive will load a file and insert it as source at the
      designated point.
    </para>

    <para>
      A related directive, <literal>.require</literal>, will include
      the file as long as it hasn't been included yet elsewhere.  It
      is useful for ensuring a library is linked in.
    </para>

    <para>
      For pre-assembled code or raw binary data,
      the <literal>.incbin</literal> directive lets you include the
      contents of a binary file directly in the output.  This is handy
      for linking in pre-created graphics or sound data.
    </para>

    <para>
      If you only wish to include part of a binary
      file, <literal>.incbin</literal> takes up to two optional
      arguments, naming the file offset at which to start reading and
      the number of characters to read.
    </para>

    <para>
      As a sample library, we will expand the definition of
      the <literal>chrout</literal> routine to include the standard
      names for every KERNAL routine.  Our header file will
      then <literal>.require</literal> it.
    </para>

    <para>
      We'll also add some convenience aliases for things like reverse
      video, color changes, and shifting between upper case/graphics
      and mixed case text.  We'd feed those to
      the <literal>chrout</literal> routine to get their effects.
    </para>

    <para>
      Since there have been no interesting changes to the prelude, and
      the KERNAL values are standard, we do not reproduce them here.
      (The files in question are <xref linkend="c64-1-src"
      endterm="c64-1-fname"> and <xref linkend="kernal-src"
      endterm="kernal-fname">.) The <filename>c64kernal.oph</filename>
      header is likely to be useful in your own projects, and it is
      available in the <literal>platform/</literal> directory for easy
      inclusion.
    </para>
  </section>
  <section>
    <title>Macros</title>

    <para>
      A macro is a way of expressing a lot of code or data with a
      simple shorthand.  It's also usually configurable.  Traditional
      macro systems such as C's <literal>#define</literal> mechanic
      use <emphasis>textual replacement</emphasis>: a macro is
      expanded before any evaluation or even parsing occurs.
    </para>

    <para>
      In contrast, Ophis's macro system uses a <emphasis>call by
      value</emphasis> approach where the arguments to macros are
      evaluated to bytes or words before being inserted into the macro
      body.  This produces effects much closer to those of a
      traditional function call.  A more detailed discussion of the
      tradeoffs may be found in <xref linkend="ref-link">.
    </para>

    <section>
      <title>Macro definitions</title>

      <para>
        A macro definition is a set of statements between
        a <literal>.macro</literal> statement and
        a <literal>.macend</literal> statement.
        The <literal>.macro</literal> statement also names the macro
        being defined.
      </para>

      <para>
        No global or anonymous labels may be defined inside a macro:
        temporary labels only persist in the macro expansion itself.
        (Each macro body has its own scope.)
      </para>

      <para>
        Arguments to macros are referred to by number: the first is
        <literal>_1</literal>, the second <literal>_2</literal>, and so on.
      </para>

      <para>
        Here's a macro that encapsulates the printing routine in our
        <quote>Hello World</quote> program, with an argument being the
        address of the string to print:
      </para>

<programlisting>
.macro print
        ldx #0
_loop:  lda _1, x
        beq _done
        jsr chrout
        inx
        bne _loop
_done:
.macend
</programlisting>

    </section>
    <section>
      <title>Macro invocations</title>

      <para>
        Macros may be invoked in two ways: one that looks like a
        directive, and one that looks like an instruction.
      </para>

      <para>
        The most common way to invoke a macro is to backquote the name
        of the macro.  It is also possible to use
        the <literal>.invoke</literal> command.  These commands look
        like this:
      </para>

<programlisting>
`print msg
.invoke print msg
</programlisting>

      <para>
        Arguments are passed to the macro as a comma-separated list.
        They must all be expressions that evaluate to byte or word
        values&mdash;a mechanism similar to <literal>.alias</literal>
        is used to assign their values to the <literal>_n</literal>
        names.
      </para>
    </section>
  </section>
  <section>
    <title>Example code</title>

    <para>
      <xref linkend="tutor3-src" endterm="tutor3-fname"> expands our
      running example, including the code above and also defining a
      new macro <literal>greet</literal> that takes a string argument
      and prints a greeting to it.  It then greets far too many
      targets.
    </para>
  </section>
</chapter>