ljr/livejournal/doc/raw/ljp.book/bml/core.xml

36 lines
1.8 KiB
XML
Raw Normal View History

2019-02-05 21:49:12 +00:00
<chapter id="ljp.bml.core">
<chapterinfo>
<title>Core <abbrev>BML</abbrev> blocks</title>
</chapterinfo>
<title>Core <abbrev>BML</abbrev> blocks</title>
<para>
Core blocks are predefined blocks that are named with a leading underscore.
Most core blocks have a higher purpose than simple template use:
</para>
<variablelist>
<varlistentry>
<term><literal role="core.bml.block">_code</literal></term>
<listitem><para>
<literal>&lt;?_code _code?&gt;</literal> blocks are perhaps the most useful feature of
<abbrev>BML</abbrev> outside of the ability to have global site templates.
These blocks allow template authors to embed pieces of executable Perl code
within the bml page that get executed on the server.
</para><para>
</para><para>
The code you write gets executed in its own package (namespace) called
<computeroutput>BMLCodeBlock::</computeroutput>.
Any variables you declare in one code block on a page without using
<literal>my</literal> are carried on to the next <literal>_code</literal> block.
</para><para>
Because the BML parser must evaluate everything on the page before sending the
<abbrev>HTTP</abbrev> headers, make sure you don't print anything.
Any output printed to <literal>STDOUT</literal> will just be interpreted as
<abbrev>HTTP</abbrev> headers. How the <literal>_code</literal> blocks work is
that you need to return a value at the end.
Whatever value your code fragment returns is what the block evaluates to.
Usually what you end up doing is building a string, concatenating things to it
over and over, and then returning it at the end.
</para></listitem>
</varlistentry>
</variablelist>
</chapter>