> On Dec. 9, 2019, 3:34 p.m., Benno Evers wrote: > > 3rdparty/stout/include/stout/path.hpp > > Lines 388 (patched) > > <https://reviews.apache.org/r/71878/diff/1/?file=2182150#file2182150line388> > > > > This feels a bit more "common" to me: > > > > struct PathComponentIterator { > > // .. > > }; > > > > typedef const PathComponentIterator const_iterator; > > > > What do you think?
I'd vote to not do this as we do not need it. > On Dec. 9, 2019, 3:34 p.m., Benno Evers wrote: > > 3rdparty/stout/include/stout/path.hpp > > Lines 389 (patched) > > <https://reviews.apache.org/r/71878/diff/1/?file=2182150#file2182150line389> > > > > Using `std::iterator` is going to be deprecated in C++17, apparently > > the preferred alternative is to just create typedefs for `value_type`, etc. > > in the iterator class directly. I'll do that since it also documents more what is happening in a more obvious way. > On Dec. 9, 2019, 3:34 p.m., Benno Evers wrote: > > 3rdparty/stout/include/stout/path.hpp > > Lines 437 (patched) > > <https://reviews.apache.org/r/71878/diff/1/?file=2182150#file2182150line437> > > > > I think this needs to be a loop, consider e.g. `/home/foo////bar` That would be path normalization which I didn't want to do in this class as it has a number of other complications. For a path like `/home/foo//bar` we would like to generate elements `{"home", "foo", "", "bar"}`. I'll clean up remnants of my attempts in this class here and in tests. > On Dec. 9, 2019, 3:34 p.m., Benno Evers wrote: > > 3rdparty/stout/include/stout/path.hpp > > Lines 454 (patched) > > <https://reviews.apache.org/r/71878/diff/1/?file=2182150#file2182150line454> > > > > Should we also check `path == other.path`? Or maybe even `CHECK(path == > > other.path)`? I'll end the hard `CHECK` once I have gotten rid of the end sentinel. > On Dec. 9, 2019, 3:34 p.m., Benno Evers wrote: > > 3rdparty/stout/include/stout/path.hpp > > Lines 469 (patched) > > <https://reviews.apache.org/r/71878/diff/1/?file=2182150#file2182150line469> > > > > I don't completely understand why we need a sentinel, isn't > > `path->end()` suitable for everything we need to do? I removed as per our offline discussion. - Benjamin ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/71878/#review218978 ----------------------------------------------------------- On Dec. 9, 2019, 5:38 p.m., Benjamin Bannier wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/71878/ > ----------------------------------------------------------- > > (Updated Dec. 9, 2019, 5:38 p.m.) > > > Review request for mesos and Benno Evers. > > > Bugs: MESOS-10062 > https://issues.apache.org/jira/browse/MESOS-10062 > > > Repository: mesos > > > Description > ------- > > Added iteration support to stout's Path. > > > Diffs > ----- > > 3rdparty/stout/include/stout/path.hpp > ba1f665ce94b9636d88a7ecce8643c56758f7b5c > 3rdparty/stout/tests/path_tests.cpp > 19dd910a534040468aeb48f15ebdf56dff32bc15 > > > Diff: https://reviews.apache.org/r/71878/diff/2/ > > > Testing > ------- > > `make check` > > > Thanks, > > Benjamin Bannier > >