Pep & Nom

Did you forget to walk barefoot on the grass? A Question

== The Status of the Pep & Nom System

This document attempts to set out in summary way the current status of the Pep (string register) Virtual Machine and the Nom parsing language.

The nom system is becoming a powerful and interesting way to recognise and transform text patterns. I have begun to use Nom in practical ways, such as writing text formatters and simple chatbots. A number of the nom translation scripts in the /tr/ folder are effective and useful. Several of the scripts in the example folder are practically useful (to me at least).

However, I have been working on this system for a number of years without any collaborators and many areas of the nom system are (ahem...) less than complete and certainly not bug-free (please help!) . Below I document some inadequacies of the system as far as I can remember them

1 dec 2025: "issues" with pep-nom

all the translation scripts need further debugging and testing
the translation scripts which are called translate.<lang>.pss are outmoded and should be converted to the grammar and quality of the scripts called nom.to<lang>.pss
The pep interpreter is written in plain c and sadly comes up with segmentation faults from time to time (although sometimes I may go 6 months without getting one). So this needs to be seriously debugged. It also uses byte characters which is a bit embarrassing
The /eg/sed.tojava.pss sed-to-java compiler needs to implement the sed 'jump' syntax and needs to be tested properly. It probably should be converted to a memory safe language that supports some kind of 'goto' syntax so that it can support the full range of gnu sed syntax.
the script /eg/toybnf.pss needs to be expanded so that it can become useful
the /eg/text.tohtml.pss is pretty useful (to me) but it should support nested lists, and it would be good for it or a clone of it to translate to man pages, markdown format, latex and epub.
the nom syntax checker /eg/nom.syntax.reference.pss is quite good at checking nom syntax and providing helpful error messages, but it needs to have it’s grammar changed slightly to reflect the grammar in the recent translation scripts such as /tr/nom.todart.pss
/eg/maths.tolatex.pss creates formulas from plain text, but handles operator associativity of the power operator ^ incorrectly.
the /eg/natural.language.pss “English” language phrase-grammar needs to be expanded greatly in order to become interesting (for example: by using the system command to look up word classifications)
the json parser /eg/json.check.pss needs to handle Unicode escape sequences
the c code in /object/machine.methods.c needs to be updated and debugged to reflect the code in /object/machine.interp.c (which is the interpreted version of this). The machine.methods.c is still useful for compiling standalone executables of nom scripts where other, more modern languages are not available (for example, on micro-controllers ?)
The Forth nom translator /eg/nom.toforth.pss needs to be written so that nom can run easily on micro-controllers, and provide micro-controllers with high(er) level languages (similar to lua on the nodemcu boards). But this involves writing memory management code for the tape array which will be an interesting challenge.
I need a better testing system for large scripts and for the translation of them.