Fill in envForLambda with errorObjects if not enough are provided.
Move addStructDef() into env.c
Add dedicated Dictionary struct to env.c
env->size -> env->capacity
Replace (pch) with (ch).
Fix error memory bug.
Add sugar for (? fil) => (? "fil").
Inline some of (fil).
Simplify (print) and allow for varargs.
Total up valgrind errors when testing.
Add basic allocation tracking.
Add extra error context
- more useful for debugging the language than programs written in it
Let reduce work on a single non-list item. May be removed.
Add more testing.
Add test filtering.
Fix triple-quotes.
Add `reduce` and `islist`.
- Reduce is broken with lambdas for some reason.
Remove `switch` from lib (breaks `penv`).
More detailed lambda to-stringing.
A bit of fixing for lambdas that return lambdas.
Re-org Object fields for simpler visual debugging.
Let `cat` work with one argument.
Add prompt option for `inp`.
Some more refinement of the webby.pl example.
Adjust global Environment implementation.
Example `webby.pl` is now a simple server program.
Only limit stringNObj outside of STANDALONE implementations.
Larger catObjects sizes.
Parsing files ignores semicolons inside of quotes.
Temporarily disable http due to buggy Ubuntu libmicrohttpd.
Add a way to access their fields with `'s`.
- Uses some fairly hacky parseAtom additions. Needs to be refined.
Add structCapacity to Environment.
Add global Environment pointer.
Use structDef indexes instead of direct pointers.
- Easier to increase capacity this way
strcpy struct field names
LOW_MEM drops some functionality (like `defe`) for very low memory devices like
aplite. SIMPLE_ERRORS is enabled when compiling with LOW_MEM, and overrides the
new error-with-context functionality with older, less intensive methods.
Also marked more code for standalone compilation only.
External:
Error changes.
User windows will show string reps of all object types
Internal:
Some re-org and renames in calc.
Add stringNObj for more controlled string-receiving.
Note: currently aplite has a hard time with larger scripts, possibly because of
recent changes. Not sure how much can be done about it, but I'm investigating.
Certain I/O functions (like `inp` or `repl`) do not need arguments, but
previously all functions required at least one argument to called correctly
(i.e. `(inp 0)` or `(repl "banana")`). Now, these can be called as intended:
`(inp)` or `(repl)`.
Fixed OOP error in allocating lambda environments. Add ability to disable test
groups. Disabled forbble tests and `switch` implementation (broken by `def`
change)
For example, running "./do_math.pl 5 + 5" would define args as a list of
( /usr/bin/pl ./do_math.pl 5 + 5 ), where each of the five elements is a
string object. The arguments will also be loaded when opening the repl,
but NOT when writing pl code directly as command-line arguments (i.e.
'./pl "(* 9 9)"').
Changed install dir.
General cleanup of Makefile.
Added `make uninstall` to Makefile.
Tests still require that lib scripts already be installed to pass.
Installs examples/* to /usr/share/pebblisp
A basic REPL is one of these examples, replacing the built-in C version
with one actually written in PebbLisp. It can be invoked the same way:
running `pl` without any arguments.
`iserr` is a basic type-check like `isnum`
`defe` adds a string to the environment as the equivalent symbol
`eval` can take in variables or functions, instead of just raw strings
Broke type-printing into its own function
Errors maxing out an environment. Likely extending incorrectly
Sage Vaillancourt
=