Closed MartinDelille closed 5 years ago
Merging #136 into master will not change coverage. The diff coverage is
n/a
.
@@ Coverage Diff @@
## master #136 +/- ##
=======================================
Coverage 98.64% 98.64%
=======================================
Files 34 34
Lines 956 956
=======================================
Hits 943 943
Misses 13 13
Continue to review full report at Codecov.
Legend - Click here to learn more
Δ = absolute <relative> (impact)
,ø = not affected
,? = missing data
Powered by Codecov. Last update bfdb8a3...5350809. Read the comment docs.
Thank you very much. I checked your changes and they are fine. Note that the code you've changed is the external header-only Lean Mean C++ Option Parser library. You might want to submit your changes there, too.
You're welcome!
Unfortunately, I didn't find an easy way to contribute to optionparser.
@mbenkmann You seems to be the author. Would you mind to mirror it to github?
optionparser could become a git submodule then.
The easy way to contribute to optionparser is an email to the mail address listed on the front page
http://optionparser.sourceforge.net/
Given that I have the script all set up to push new releases to SourceForge, including uploading the doxygen docs, I feel no particular desire to mirror it on github. Just check the file into your repository.
As for the change itself: New style casts are ugly.
@mbenkmann Ok if you won't accept the change I might not push them. Tell me if you have a better idea to fix the warning in a nicer way!
I'm sure the compiler you use has #pragmas to disable certain warnings for individual files.
That is not what I call a nice way in my opinion. Let's keep it like that!
Oh, it's funny that you discuss it in this merged pull request. ;-)
@MartinDelille I will definitely keep the new version using explicit casts. Often I prefer type(var)
casts (i.e., calling the conversion constructor), however, in this case the explicit casts are much better since, for example, const_cast
show what part of the cast is currently important (the const-ness change), and reinterpret_cast
for a pointer-to-unsigned-long-long cast seems also much more clearer to me.
By the way, for the next major release I have to replace the lean mean option parser by something different, which allows registering new options after the initialization. I want that bandit users can define their own options (see #92), with use-cases such as --fast-tests-only
or --intense-testing
or --seed=12345
or whatever the user might think of. While the lean mean option parser is great at what it is written for, it does not directly offer such flexibility. (This is an old item on my todo list but currently other things have higher priority, like finishing my PhD.)
Your code was triggering
old-style-cast
warning. Would you accept this fix?