Use fmtlib for Output logger

[fmatthew5876]

Fix: #2098

In order to avoid rebase hell, I will add commits to change all of our log statements only after this PR gets approved.

Looks like linux packaging for this library is not great. In Ubuntu eoan for example it only has ancient version 5.1.2 which doesn't have pkg-config and breaks on trying to print char32_t.

Focal has 6.1.2 which should be fine.

vcpkg is up to date with latest 6.2.0

[Ghabry]

So the minimal supported version should be 5.3.

Is for Debian in "Debian Stable Backports".
Fedora starting with Fedora 31.
OpenBSD Ports has 2.0 ??? o.O
Ubuntu starting from Ubuntu 20.04.
OpenSuse Leap 15.0

Other part are the console ports. I wonder how much frustration compiling this with newlib will give >.>

[Ghabry]

Library building looks good. Havn't tested Player linking yet.

You also have to add "-lfmt" to all our Makefiles. (fortunately this will solve when the CMake migration is done...)

[fmatthew5876]

I tested and this compiles and links successfully with

• Ubuntu eoan cmake
• Ubuntu focal autotools
• Windows vcpkg

[Ghabry]

Build fails with

tests/output.cpp:13:10: error: 'Post' is not a member of 'Output'
   13 |  Output::Post("Test %s", "post");

[fmatthew5876]

Thanks for catching that, fixed the test.

[Ghabry]

Could youdo a full text search for Output::Post ? Also see a failure in the vita build.

[fmatthew5876]

Hmm I did do a text search, but must have missed that one. Anyway, updated it.

[Ghabry]

Everything compiled succesfully.
Needs toolchain recompile before merge.

[fmatthew5876]

Ok I've added commits to convert all log messages over. I also cleaned up and tested Window_NumberInput.

I did this by grepping for all files with Output::. Then opening them all in my editor and using search and replace in the edtior for the basic cases and manually fixing the complicated ones.

I didn't want to do something fully systematic like sed here because we have some snprintf usage. strftime() and other things like removing .c_str() which would take longer to get right in a script and probably still end up with mistakes.

This is tricky because any mistake here will unfortunately show up as a runtime exception (or just an assert() if exceptions are disabled on the platform). So let's review the diffs carefully.

[fmatthew5876]

Self reviewed this, fixed a few minor things. From my perspective this is ready to go.

[Ghabry]

Does fmtlib support %s and %d in case we miss something?

[fmatthew5876]

Does fmtlib support %s and %d in case we miss something?

It'll just print the %s and %d literally and ignore the arguments.

I did a ag Output:: | grep % and it returns me nothing. Also grepped for %. It's only comments, mods, strftime, and printfs. Pretty sure we got them all

[Ghabry]

Jenkins: Test this please <3

[Ghabry]

(Note that many new commits - At least 10 - were added since the last review)

[fmatthew5876]

Addressed review comments

[fmatthew5876]

Rebased, and fixed a few in input from the rebase.

Did another set of greps to verify no stray % leaked in.

[carstene1ns]

I hope this works correctly with our newlib plaforms.