Warnings during cmake build-test

[unbtorsten]

Not strictly a bug, but could turn into one or more bugs, if ignored for too long:

To Reproduce
Steps to reproduce the behavior:

1. Clone HighFive repository and cd into local repo
2. $ cmake -B build-test .
3. $ cmake --build build-test --parallel

This yields the following warnings:

HighFive/tests/unit/tests_high_five_base.cpp:326:13: warning: ‘HighFive::DataSet::DataSet()’ is deprecated: "Default constructor creates unsafe uninitialized objects" [-Wdeprecated-declarations]
  326 |     DataSet d2;  // deprecated as it constructs unsafe objects
      |             ^~
In file included from HighFive/tests/unit/tests_high_five_base.cpp:21:
HighFive/include/highfive/H5DataSet.hpp:103:5: note: declared here
  103 |     DataSet() = default;
      |     ^~~~~~~
HighFive/tests/unit/tests_high_five_base.cpp: In function ‘void CATCH2_INTERNAL_TEST_44()’:
HighFive/tests/unit/tests_high_five_base.cpp:758:45: warning: ‘haddr_t HighFive::ObjectInfo::getAddress() const’ is deprecated: "Deprecated since HighFive 2.2. Soon supporting VOL tokens" [-Wdeprecated-declarations]
  758 |     auto addr = dataset.getInfo().getAddress();
      |                 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~
In file included from HighFive/include/highfive/H5Object.hpp:155,
                 from HighFive/include/highfive/H5DataSpace.hpp:17,
                 from HighFive/include/highfive/H5DataSet.hpp:13:
HighFive/include/highfive/bits/H5Object_misc.hpp:109:16: note: declared here
  109 | inline haddr_t ObjectInfo::getAddress() const noexcept {
      |                ^~~~~~~~~~

Expected behavior
No warnings

Desktop:

• OS: Debian 12
• CMake 3.25.1
• gcc 12.2.0

I may be able to look into this further in the next few days. For now, I didn't want to just ignore it.

[1uc]

The first one is benign, for historical reasons DataSet used to have a default ctor. Then someone decided that it made little sense to have a DataSet object that doesn't actually refer to a dataset in some file. Hence, they deprecated the default ctor. However, to ensure that we don't accidentally remove it; and thereby break users code, there's this test.

One could probably push and pop -Wdeprecated; but one might need to do so for clang, GCC and MSVC.

The other one predates me and I don't understand why it's deprecated. There's certainly no active push towards VOL in HighFive right now. I used to assume that it's because something that's deprecated in HDF5, but a quick search:
https://docs.hdfgroup.org/hdf5/v1_10/struct_h5_o__info__t.html
https://docs.hdfgroup.org/hdf5/develop/struct_h5_o__info1__t.html
doesn't lead to anything obviously deprecated in HDF5.

[1uc]

Yes, both warnings are annoying.

[unbtorsten]

Understood. This touches on a broader context of when and how to deprecate functionality/interfaces. I suggest adding some more comments for future reference and wouldn't pursue this at this time: #832