Issues with new Fmt support about graylog-logger HOT 7 CLOSED

The CMake issue turned out to be a bit more complex than I thought which is partially to blame why I did not see this issue in my testing. The FMT library (when installed from source) will also install the correct "FindFMT"-files for CMake to use. This is also true for some FMT Conan packages. However, it seems that some newer FMT Conan packages (including the "conan-center" ones) has lost the files that makes "find_package(fmt)" work properly. I will have to give it a bit more thought to figure out the best solution to this issue.

from graylog-logger.

Hi, sorry for not getting back to you earlier. Give me until tomorrow to investigate the issues that you are having and I will try to formulate a plan for how they can be resolved.

from graylog-logger.

It took a little while for me to remember why I put the arguments in a tuple and thus had to use minimal::apply. The reason being that if the user of the library supplies a reference to a variable as an argument, we will have a problem with dangling references. We have to make copies of all the used variables (using std::make_tuple and [=] when creating the lambda) and then convert that tuple back to template arguments using minimal::apply as std::apply is not a part of C++14. I am pretty sure that minimal::apply is implemented only using C++14 features. Can you post the compile errors that you are seeing?

from graylog-logger.

OK, I figured there was probably a good reason you were doing things that way!

It turned out the real issue is I have this at the top of my cpp source file.

#include <graylog_logger/Log.hpp>
...
using namespace std;

The compiler was trying to use std::invoke rather than your minimal::invoke (even with std=gnu++14).

I'm probably going to refactor to avoid using namespace now that I know it can interact with templates like this (lesson learned - Google style guide says not to do it for a reason). But perhaps qualifying this line as minimal::invoke(...) would help stop others falling into the same trap I did

from graylog-logger.

Sure, I can do that.

from graylog-logger.

This was somewhat more complicated than I thought and thus took much longer than I thought it would. Still, I think it is fixed now in PR #38.

from graylog-logger.

Brilliant, both fixes work in my setup! I thought you resolved this really quickly, many thanks.

from graylog-logger.