extended control of C/C++ validation

[barbieri]

Hi,

The current c_checker() and cpp_checker() uses a simple -I. -I.. which will get you some basic tests, but in real world we're often using things like pkg-config --cflags glib-2.0 and the likes to provide definitions and more include paths. Then a way to control INCLUDE_FLAGS would be required. I'm not sure how to best provide these extensions in a compatible way so we don't break sphinx, maybe using comments?

.. rstcheck-include-flags: -I.
.. rstcheck-pkg-config: glib-2.0
.. code-block: c
  regular code block here

Eventually allow these to be linked in addition to syntax check:

.. rstcheck-include-flags: -I.
.. rstcheck-pkg-config: glib-2.0
.. rstcheck-link: yes
.. code-block: c
  regular code block here

Likewise, sometimes in tutorial/documentation we split the code into multiple pieces that can't be validated at once. I could work-around that by using an external C/C++ file with include statements, however if we could add a argument to "append to previous block" it would be amazing, allowing things like rstcheck-mode with a choice of new, append and ignore:

.. code-block: c

  void on_event(void *ctx) {
   ...
  }

This provides a function to callback when event happens...

.. rstcheck-mode: append
.. code-block: c

  int main(void) {
   my_lib_init();
   my_lib_on_event_set(on_event, NULL);
   return 0;
  }

[myint]

Very interesting! Allowing the user to configure the include paths definitely makes sense. And the approach you describe makes sense and basically matches what is done for the ignoring languages feature.

The appending code blocks would be cool. I think that is a decent solution.

[barbieri]

great, this one is a bit trickier than the other issue I've opened in the sense it will need to accumulate from one statement to the other... the ignore of languages is a single step prior to start, it's not tied to the subsequent blocks.

Any preferences on how to tackle this? Maybe just a global stack where one pushes and then pops?

[myint]

No preferences from my side. I agree that it doesn't seem easy at all.

[Cielquan]

Does rstcheck/rstcheck#88 help here?