GitLab

Otherwise, some generator statements still may be spuriously executed on
subsequent calls to next()/send().

The problem manifests itself in make 4.0

I also fixed the LINK message when linking the final executable for unix and unix-cpy.

Addresses Issue #203.

TODO: Check lexer/parse/compile error path for leaks too.

mp_map_deinit() finalizes static map, mp_map_free() - dynamic.

This reverts commit a215b09c.

There can be multiple emitters allocated during compile (eg byte code
and native).

Want common qstrs to be first in the list so they have the lowest ids,
so that in the byte code they take up the least room.

This reuses as much str implementation as possible, from this we
can make them more separate as needed.

* Move gc_collect from main to py/gc.c
* Define GC's memory boundaries in linker script
* Issue #220

Also, add qstr's for string appearing in unix REPL loop, gross effect
being less allocations for each command run.

Fixes segfault on callable("string").

Addresses Issue #207.

STM port crashes without this re-init.  There should not be any state in
the core py/ code that relies on pre-initialised data.

Addresses Issue #199.

Also fixes a bug in the for-in-range optimiser.

I hope to remove break and continue byte codes in the future and just
use jump (if possible).