Apologies for forgetting the phabricator review in the commit message.

Submitted without the Differential message. My bad.

Moved function definitions outside of the anonymous namespace

It is the attribute we decided to use for now. But the plan is to rename it at some point.

Looks good.
Thank you.

Why do you say "Restore compile time code" in the review title? EnumToString never was compite-ime until now.

Consider this program:

Moving the table initialization into a Meyers singleton.

There was a typo in the commit message, so the review was not closed automatically

No, it is not Meyers singleton. It uses the same C++ guarantee that is the basis for the Meyers singleton.
It is is not thread safe. I only intended for the table to be unique. If we need thread safety then maybe a constexpr table is best.

Peter has resigned from the review. He says, all his concerns have been addressed but he would like someone else to approve the change.

Applied clang-format

Applying clang-format

1. Returning a reference to string since string are now held permanently in memory
2. Adding #include<array>
3. No changes in idioms.cpp because it turns out that it is not used by the runtime and could never have been since it returned strings by value

This pattern is based on the Meyers singleton. There is only one instance of the lookup table.
It was not possible in the past to return a reference to string since the string was computed on the fly. It did not seem to bother anyone so I did not change that.
Though this would be possible now. I gave it a quick try and I am only getting a few compilation errors. So, I can look into these and add to these changes.
I could also use good ol' strtok if regex is going to be a problem.

Needed to fix test because of change https://reviews.llvm.org/D134900

Using MLIR_MAIN_SRC_DIR instead of CMAKE_SOURCE_DIR because

1. It's definition is now fixed
2. It works for both in and out-of-tree builds

Added lines at end of files.
Corrected typo

Needed to run clang-format on a couple of files.

I am not going to submit this patch as I have found out that it is not legal FIR to have branch based CFG operations inside of block based CFG operations (fir.do_loop, fir.if and fir.iterate_while)

Renamed test files.

I had mistakenly removed lit.cfg.py instead of undoing the changes in it.

Split tests into 4 files. Just can't figure how to get split-file to work with the pre-check tests.
Added constructor with options to CfgConversion

Added split-file as a tool. Seems, it was necessary for the pre-check tests.

Fixed issue this way: https://github.com/flang-compiler/f18-llvm-project/pull/822

I have been looking some more into how the flang-new driver is being built.
I see that flangFrontend and flangFrontendTool libraries have been introduced and include dependencies with the Fortran libraries. It makes things look elegant. When we link, we just have to say flangFrontend and we get all the Fortran libraries. However when building with shared libs , you can see that the flangFrontEnd libraries are very small, they no longer include any of the code they depend on. So the attempt to have an umbrella library does not work because the linker does not link with the dependent libraries. In a shared lib flow, you have to make them explicit.

The change does NOT appear to be necessary for main of llvm-project, however I could reproduce it with fir-dev branch of f18-llvm-project. I was told that changes to this file should go through phabricator and percolate to fir-dev. I am not too sure what to do now.

The link is about building flang with an external build/install of LLVM/MLIR.

My apologies. I should have blown away my build tree prior to testing this, otherwise cmake still somehow finds its way to the *.h.inc files

A use-case would be an out-of-tree build with a Jenkins build for instance, which only keep the install tree around. MLIR enables this already.

reated a full diff with "git show HEAD -U999999" instead of just "git show HEAD"