Initial definition for TSK_TRACE_ERRORS

[jeromekelleher]

This is a proposal to address the discussion in #3094.

Essentially, we add a function-like macro tsk_trace_error, and everywhere we normally do

ret = TSK_ERR_X;
goto out;

we instead do

ret = tsk_trace_error(TSK_ERR_X);
goto out;

We can then define tsk_trace_error to be a no-op for production code, and define a function which emits some error traces. For the changes I've made here, we get this on stderr:

tskit-trace-error: -701='Sequence length must be > 0. (TSK_ERR_BAD_SEQUENCE_LENGTH)' at line 11000 in ../tskit/tables.c

Implementation notes

• tsk_trace_error must be a macro, otherwise we can't use FILE and LINE to get locations
• This structure has no change on control flow, which is essential
• It should be easy enough to sed the changes to implement this for the vast majority of the library.

I can push this through and document if folks like it.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 92.90%. Comparing base (af33ed1) to head (87859e6).
Report is 4 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3095      +/-   ##
==========================================
+ Coverage   89.87%   92.90%   +3.03%     
==========================================
  Files          29       21       -8     
  Lines       32188    16509   -15679     
  Branches     5776     2791    -2985     
==========================================
- Hits        28928    15338   -13590     
+ Misses       1859      689    -1170     
+ Partials     1401      482     -919     

Flag	Coverage Δ
c-tests	?
lwt-tests	80.78% <ø> (ø)
python-c-tests	89.05% <ø> (ø)
python-tests	98.98% <ø> (-0.01%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

see 9 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[benjeffery]

Seems like a solid idea to me, should be able to hit most of the changepoints with a regex!

[jeromekelleher]

OK cool, I'll fill this out as I get a chance.