Fix lower incomplete gamma functions with x = 0

[cohomology]

In this case, the errno error handling did not work correctly, as internal functions where accidently setting it, although no overflow happens.

Fixes #1249.

[codecov]

Codecov Report

Attention: Patch coverage is 98.55072% with 1 line in your changes missing coverage. Please review.

Project coverage is 93.82%. Comparing base (a5c0625) to head (7503be4).

Files with missing lines	Patch %	Lines
test/git_issue_1249.cpp	98.38%	1 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #1251      +/-   ##
===========================================
- Coverage    93.83%   93.82%   -0.01%     
===========================================
  Files          657      658       +1     
  Lines        55244    55330      +86     
===========================================
+ Hits         51840    51916      +76     
- Misses        3404     3414      +10     

Files with missing lines	Coverage Δ
include/boost/math/special_functions/gamma.hpp	92.08% <100.00%> (-0.18%)	⬇️
test/git_issue_1249.cpp	98.38% <98.38%> (ø)

... and 7 files with indirect coverage changes

---

Continue to review full report in Codecov by Sentry.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update a5c0625...7503be4. Read the comment docs.

🚀 New features to boost your workflow:

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

[jzmaddock]

I think we need to let this line execute without the check for x > 0, that way our root finder will get a derivative back: in this particular case, any arbitrary large value will do.... ah but there should be an else before the *p_derivative /= x;.

[cohomology]

Hi, I adjusted the coding. Is it okay now?

I also corrected some wrong ternary operators in the file which prevent using some functions with non-standard float/double T template parameters.

[jzmaddock]

Other than my one small comment, this all looks good. Many thanks for this!

[jzmaddock]

Looks great, thanks, just running CI now...

[cohomology]

Where are these CI errors come from? I don't get them ...

[mborland]

Where are these CI errors come from? I don't get them ...

The Cauchy one was failing with the SYCL compiler, and the beta ones are new failure there that are unrelated. The last one is we need to increment the current version in Boost.Config. I would not worry about any of these failures. I need to address them, and make sure we haven't broken CUDA along the way

[jzmaddock]

Ah shucks, sorry, I hadn't spotted that the new beta tests were failing on CUDA. @mborland some of them are trivial failures (need to update expected error rates), but there are a few "gross" errors which might warrant investigation. Or it may be that the inputs are so extreme that there's not much we can do on that platform. If you have CUDA/Sycl set up can I let you investigate?

[mborland]

Ah shucks, sorry, I hadn't spotted that the new beta tests were failing on CUDA. @mborland some of them are trivial failures (need to update expected error rates), but there are a few "gross" errors which might warrant investigation. Or it may be that the inputs are so extreme that there's not much we can do on that platform. If you have CUDA/Sycl set up can I let you investigate?

Yes, next week I'll look into these. The Scipy guys also are in a position now to use the CUDA stuff so I'll put some effort into the rest of the library too.