What is iPerl?
Perl-programmes may contain bits of string to be output, whereas inversely iPerl-documents consist of any kind of text to be output, and bits of specially marked Perl that control the document with conditionals or loops spanning text, subdocument-includes and macro definitions. This is comparable to, but far more powerful than the C preprocessor or the m4 macro processor, but can also be seen as a template-mechanism.
What can iPerl do?
Various styles are provided by the library Text::iPerl and the command-line tool iperl, one resembling the C preprocessor, one the m4 macro processor and new ones for arbitrary documents and for example SGML-documents. The useful Perl-feature of embedding documentation into the programme (POD) is not only brought alive, but can be used in all kinds of documents.
These styles are also used by web-iPerl, a CGI-frontend to Perl-server-pages, which can generate (data-base-driven) tables in HTML- or XML-pages for example, or autoprocess forms.
Why is it called inverse Perl?
Perl programmes follow a well defined syntax, and may contain bits of arbitrary text called strings which might get output. Here it is the other way round: everything is arbitrary text to be output, unless it is specially marked up as being Perl.
What is markup?
Flat files are simpy a sequence of characters or bytes. While the human reader will likely class these into words, lines and paragraphs, a programme needs rules on how to distinguish various parts of the file. For example, to set off arbitrary text from the code, Perl uses string markup such as ', ", q! or qq!. The same thing happens here, only that the markup goes around bits of Perl instead and varies with the style.
Which style should I use?
This is partly a question of taste, since all styles are functionally equivalent. Almost at least, e.g. style cpp cannot embed bits of Perl in the middle of lines. There's a tradeoff between styles that interfere least with any character sequences that might occur in the host document and thus offer minimal comfort on the one hand, and those that give you flexible ways of embedding Perl on the other.
When it comes to interacting with other programmes, there might be some restrictions. An HTML-editor for example will be happiest with the verbose variant of style xml. A syntax-based editor for C++ will best accomodate style cpp.
Can I use iPerl in sendmail configuration?
All m4 files used for this would have to be converted to Perl-syntax. E.g.
ifdef(`confFROM_HEADER',, `define(`confFROM_HEADER', <$g>)')dnlbecomes
ifdef('confFROM_HEADER', '', q{define(confFROM_HEADER => '<\$g>')})dnlNote how q{} is suggestively used to quote a string containing a macro, i.e. Perl code, and how the $ is escaped in the definition since this string will be double quoted every time the macro gets expanded.
Can I use iPerl in C or C++?
The first problem is that modern compilers no longer call an external programme called cpp for preprocessing. So you have to call it yourself and pass the result off to the compiler.
Then, the preprocessor syntax has been arbitrarily used for the #pragma directive. So you need a function pragma that prints the corresponding directive back into the output.
Lastly all include files you use would also have to be converted to Perl-syntax, e.g. #ifdef LINUX becomes #if( defined macro 'LINUX' ) { and #endif becomes #}.
Why don't #define and #include work like in cpp?
The C preprocessor has two facets: markup consisting of a # in the first column for one thing and a fairly primitive language residing on lines thus marked up for another. Style cpp of iPerl uses only the markup of the C preprocessor. The language residing on these lines, however, is full-fledged Perl including cpp-similar functions provided by iPerl, like define and include. Hence the semantics are close to cpp, but the syntax is plain Perl.
Why don't define() and include() work like in m4?
Macros in style m4 are all mapped to calls of Perl functions, including m4-similar functions provided by iPerl, like define and include. Hence the semantics are close to m4, but the syntax of the arguments is plain Perl.
Where are my included functions?
Unlike cpp, which does simple one-pass text inclusion, include() is a run-time function in iPerl. The included document gets compiled to an individual Perl programme, which is then executed separately. Hence any functions in the inner document are only known after the outer document has been compiled.
Perl has solved the same problem with use instead of require, but this is magic in the interpreter inaccessible to us. The same effect can be achieved by explicitly placing the include() in a BEGIN { ... } block. Or, if it affects just the odd function call, you can mark it as such with an ampersand or parens.
How well tested is it?
Apart from being used variously, there is a quite extensive test-suite each release has to go through. For every style a test-document using each feature and each corrected bug is matched against a carefully analyzed reference output. And that is done with each of Perl 5.004_05, 5.005_03 and 5.6.0.