Here's a list of ideas and things that need to be done.
better/more robust implementation of xpt command parsing (see
dopi()
in xpt.c
). goal is
to make a more complete syntax in the processing-instruction
available, eg. allowing a semi-colon to separate multiple commands,
where currently multiple processing-instructions are needed. In this
way the interdependence of a group of PIs is fixed and clearly
visible.
maybe it's possible to use namespaced xpt elements making use of a DTD to check correctness of a document. Eg. use <xpt:date format="%Y-%m-%d"> instead of <?xpt date "%Y-%m-%d"?>
Investigate the possibility to validate the function calls make from xpt. Through processing instructions there is hardly any checking possible. The checking done by xpt is just syntax checking. The second level of checking could be done by the called function, but that's left to the implementor. It would be nice to have DTD features on this level. Maybe this could be accomplished by using namespaced xpt elements as shown above.
Maybe it's a nice feature to have validation on processing-instructions. This, may need extending the XML specification.
Maybe using libxslt extentions is a way of doing that, but i'm not certain. See http://xmlsoft.org/XSLT/extensions.html.
xpt function to read in external data with conversion and selection capability. reading not ony from the filesystem but also from commands and ports. watch out to make this a god-function.
sorting tables and lists.
define a logo for the project. Using Inkscape the following logo
is currently proposed. The idea behind it is that data is flowing
into the program from the left, gets processed and than is flowing
out on the right side. Note that the right arrow is in fact another
X
character from which the right side is removed.