A majority of the actions/backends expect to find certain classes and
definitions in the source code. There's no canonical reference for this
so it's a combination of incomplete docs and the actual uses in
llvm-project.
(most of the time I expect people will come from llvm-project to
experiment, rather than the other way around)
I've added an example that uses the Searchable Tables backend.
<!-- THIS COMMENT IS INVISIBLE IN THE FINAL PR, BUT FEEL FREE TO REMOVE
IT
Thanks for taking the time to improve CE. We really appreciate it.
Before opening the PR, please make sure that the tests & linter pass
their checks,
by running `make check`.
In the best case scenario, you are also adding tests to back up your
changes,
but don't sweat it if you don't. We can discuss them at a later date.
Feel free to append your name to the CONTRIBUTORS.md file
Thanks again, we really appreciate this!
-->
LLVM TableGen is used to generate complex output files in the llvm
project. A generic description of classes and definitions produces
"records" that can then be walked by a TableGen backend to produce
things like C++ code, configuration files, etc.
The biggest examples are LLVM's assembler and disassembler where all the
targets' instructions are defined in TableGen.
https://llvm.org/docs/TableGen/ProgRef.html
An example:
```
class Register<int _size, string _alias=""> {
int size = _size;
string alias = _alias;
}
def X0: Register<8> {}
def X29: Register<8, "frame pointer"> {}
```
```
------------- Classes -----------------
class Register<int Register:_size = ?, string Register:_alias = ""> {
int size = Register:_size;
string alias = Register:_alias;
}
------------- Defs -----------------
def X0 { // Register
int size = 8;
string alias = "";
}
def X29 { // Register
int size = 8;
string alias = "frame pointer";
}
```
It's often a pain point for people new to LLVM so having a quick way to
experiment would be a great benefit for the community (we have a Jupyter
kernel which is good but not as simple as Compiler Explorer).
The compiler for TableGen is `llvm-tblgen`. This comes with most release
builds of LLVM in /bin along with clang and the others. Its default is
to output a text dump of the records defined so that's what I've used
here. This is not executable code so I've disabled the features related
to that.
A user could pass options to `llvm-tblgen` to produce text in a format
other than this, C++ code or JSON for example. However this text dump is
the main use case.
I've re-used an existing clang install, since that includes
`llvm-tblgen` in `bin/`. I just added the 17.01 version for this first
change.
Syntax highlighting is a mix of the Fortran and Ada configuration,
following the language spec I linked above. Though I am very new to that
so it is likely incomplete.
