Skip to content

[RFE][patch] improve orcc reporting

Submitted by Francesco Romani

Link to original bug (#748334)

Description

Created attachment 302184
patches as tgz

I played with orcc some time ago, and hit a crash caused by incorrect input.
While debugging my issue, I found the orcc output too terse and not very friendly. So, I took the chance and spent some days hacking the parser sources (orcparse.c). It started with a few hacks, and I ended up rewriting a good part of it:

  • new richer parsing API
  • overhauled internal structure with massive modularization and cleanup
  • richer error reporting
  • more strict checking
  • new orcc option to do just the parsing
  • abort on parsing failed, instead of going ahead blindly and see the bytecode emitter crash

This code was rotting on my github account from some time, so I thought it would be of some good here, maybe.

Please find attach the resulting 36 patches (yes, thirty-six), made with git format-patch.
I intentionally made the small trying to make review easy.
Each patch should leave the code working as good practice mandates, but IMHO those are better merged as a whole, if that would be the case.

Commit message may be terse, maybe too much. Quite often the patch title self-explanatory, hopefully it actually is so.

github repo, for reference: https://github.com/mojaves/orc

Attachment 302184, "patches as tgz":
orcc_refactoring.tgz