README: describe how to use with external runners and builtin generators

[ligurio]

• describe usage with external test-runners (as followup for #29)
• describe builtin generators (I believe it would be useful to have these descriptions in README. Yes, I know about documentation generated by ldoc.)

[ligurio]

Left some minor suggestions, looks good for the rest.

Thanks for suggestions! Everything committed, squashed (also I have added Co-authored-by to mark you as co-author) and force-pushed. Seems change with descriptions for generators requires review too - I'm stuck with description for some of them.

[ligurio]

Unfortunately CI is broken, sorry for that. I'll fix it in a separate commit in the same PR (if you have no objections). right now it can start, but:

• Lua == 5.1 failed due to segmentation fault in busted (https://github.com/luc-tielen/lua-quickcheck/pull/30/checks?check_run_id=2333602586)
• Lua 5.2 failed with strange reason:

  Error: Failed installing dependency: https://luarocks.org/luassert-1.8.0-0.rockspec - Error fetching file: Failed downloading https://luarocks.org/luassert-1.8.0-0.rockspec - failed downloading https://luarocks.org/luassert-1.8.0-0.rockspec
  say 1.3-1 is now installed in /home/runner/work/lua-quickcheck/lua-quickcheck/lua_install (license: MIT <http://opensource.org/licenses/MIT>)

  although file https://luarocks.org/luassert-1.8.0-0.rockspec can be successfully downloaded on my personal machine.

• LuaJIT jobs failed to upload code coverage data:

  Uploading ...
  Upload error: Couldn't find a repository matching this job.
  Raw data:   Couldn't find a repository matching this job.
  --------------------
  Upload error: Couldn't find a repository matching this job.
  Raw data:   Couldn't find a repository matching this job.
  --------------------

[luc-tielen]

@ligurio Yeah, it never is that easy to switch over CI.. I'd look into the segfault first. If we can't fix that; we might have to go back to old way of CI? :thinking:

[ligurio]

Yeah, it never is that easy to switch over CI.. I'd look into the segfault first. If we can't fix that; we might have to go back to old way of CI?

I suspect the root cause is not in CI itself but in actions described by me. CI workflow runs on Ubuntu 20.04, I have the same OS on my laptop and everything works fine there. I'll try to investigate problem more and try to fix it.

[ligurio]

I have found a ticket about random luajit segfaults on running busted and also related conversation in luajit mailing list. Someone suggested a solution with small Lua snippet. I have applied it to lua-quickcheck and now CI is red but due to failed steps with upload code coverage data to coveralls service. I suspect lua-coveralls requires environment variable with coveralls token (see https://github.com/moteus/luacov-coveralls).

For me this solution looks like a dirty hack but I don't know better solution.

[luc-tielen]

@ligurio maybe we should split this PR into multiple parts? Just README docs in this 1, fixing CI in another? CI is broken anyway, so I will merge in when docs are updated.

[ligurio]

Agree. I have failed to fix CI quickly so it's better to split PRs. I'll do.

[ligurio]

@luc-tielen done