encfs @1.9.5_2: fails to compile on Sonoma: NullCipher.cpp:81:36: error: no matching constructor for initialization of 'std::shared_ptr<AbstractCipherKey>'

[vidakris]

see description. It wasn't working earlier either, but I keep trying every few months for some reason.

[kencu (Ken)]

info:build /opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_macports_release_tarballs_ports_fuse_encfs/encfs/work/encfs-1.9.5/encfs/NullCipher.cpp:81:36: error: no matching constructor for initialization of 'std::shared_ptr<AbstractCipherKey>'
1347	:info:build std::shared_ptr<AbstractCipherKey> gNullKey(new NullKey(), NullDestructor());
1348	:info:build                                    ^        ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

[kencu (Ken)]

looks like something to take upstream to me. Not a MacPorts build issue.

ChatGPT has a suggestion for the c++ fix, but not sure what upstream wants to do about it.

[kencu (Ken)]

ChatGPT:

It seems like the error is related to the constructor of std::shared_ptr<AbstractCipherKey>. The code is attempting to create a shared pointer with a custom deleter, NullDestructor(), but it seems there's an issue with the constructor parameters.

One possible solution could be to use std::make_shared instead of the direct constructor, and provide the custom deleter as a second argument to std::make_shared. Here's how you can modify the line:

`cpp :info:build std::shared_ptr<AbstractCipherKey> gNullKey = std::make_shared<NullKey>(); `

This assumes that NullKey is a class derived from AbstractCipherKey. The std::make_shared function automatically deduces the type and allocates memory for both the object and the control block, which should help resolve the issue.

If NullKey does not take any arguments in its constructor, this should work. If it does, you may need to provide the necessary arguments within the std::make_shared call.

[mouse07410 (Mouse)]

Yes it is an upstream issue. No, upstream is not maintaining this package anymore. Yes, we probably should apply the patch, because I don't expect any more upstream updates coming for encfs.

[kencu (Ken)]

We should probably apply the ChatGPT suggestion?

OK -- wanna test that and see if it is a good fix?

[vidakris]

output of build.sh

[vidakris]

output of unittest

[vidakris]

I tried the suggestion, it indeed solves the compiling issue. build.sh, however, still isn't successful, it fails at running the unittest stage. build.sh runs make -- make unittest -- make test, so at the last step, see attachments