If I understand your question correctly: since ntlmclient is piggybacking on the https configuration for libgit2 itself, we don't need any additional headers or libraries-libgit2's cmake rules pulled them in already

In particular, they're getting defined in https://github.com/libgit2/libgit2/blob/main/cmake/SelectHTTPSBackend.cmake - the ntlmclient project is a bit odd. I wrote it, basically for libgit2's use, but I wanted to make it able to standalone as well. That's why it's in deps/ but also just uses our configuration.

But I still don't understand which headers it uses.
I see a lot of includes in ntlm.c ( in Xcode project navigator ).
However, headers are missing.
I'm a bit confused why it doesn't contain headers in target. ( They are visible, of course, via HEADERS_SEARCH_PATH flag ), but they are not listed as headers in target.

For example, http-parser contains both http_parser.c and http_parser.h.
I expect the same for ntlmclient. ( All mandatory headers are included ).

Also I don't see a flag which switches unicode_builtin and unicode_iconv implementations in ntlmclient.

ntlm.c includes

#include "ntlm.h"
#include "unicode.h"
#include "utf8.h"
#include "crypt.h"
#include "compat.h"
#include "util.h"

so these headers should be added to a Xcode target.

This change is a try to understand which headers are required by ntlmclient since it contains different implementations of crypto-backends. ( and different implementations for unicode-backend? )
Also this change will not affect ( should not affect ) a build and only a cosmetic change ( add proper headers to a target in CMakeLists.txt )

If I'm understanding what you're saying, the lack of .h files in ntlmclient's cmake definition is probably an oversight. If we're not listing the .h files in ADD_LIBRARY or TARGET_SOURCES, then that should still compile properly, but it may not list the files properly in the IDE definitions. From this helpful blog post:

In the above example, note that .h header files were specified as sources too, not just the .cpp implementation files. Headers listed as sources don’t get compiled directly on their own, but the effect of adding them is for the benefit of IDE generators like Visual Studio, Xcode, Qt Creator, etc. This causes those headers to be listed in the project’s file list within the IDE, even if no source file refers to it via #include. This can make those headers easier to find during development and potentially aid things like refactoring functionality, etc.

I was typing at the same time you were. 😁 Yes, I agree that these should probably be included in the Xcode target.

Interesting, I've found another part of deprecated source?

No chance that unicode_iconv.c will be chosen.

No chance that unicode_iconv.c will be chosen.

Ah, interesting. If I were to do it over again, I would probably look at the GIT_USE_ICONV flag that we use elsewhere. iconv is basically a requirement on macOS to deal with Unicode path canonicalization (core.precomposeunicode). But that's probably not a huge deal. 🤔

Maybe it is better to separate parameters and pass them from libgit2 CMakeLists.txt to internal ntlmclient CMakeLists.txt? ( If CMake supports parameters for CMakeLists-intercommunication ).

Changes are merged in #5974.
Closing issue.