[Lldb-commits] [PATCH] D142926: [lldb][RFC] Replace SB swig interfaces with API headers

Fri Feb 3 11:03:21 PST 2023

bulbazord added a comment.

In D142926#4102287 <https://reviews.llvm.org/D142926#4102287>, @labath wrote:

> I like this.
>
> I'm not sure what it would take, but I think it'd be nice if the de-swig-ification was not specific to the framework build. Ideally, I'd make it controlled by a separate cmake variable, (autodetected to on if the relevant tool is found, but also overridable by the user).

Yeah we could search for unified earlier in the process and use it when we distribute headers. The de-swig-ification of the headers is primarily for distribution, do any other platforms distribute the headers? I'm sure that happens, but I'm unsure if they do anything other than just copying the headers outright.

================
Comment at: lldb/bindings/interface/SBAddressDocstrings.i:1-2
+%feature("docstring",
+"A section + offset based address class.
+
----------------
mib wrote:
> kastiglione wrote:
> > How much of these header docs are specific to the API, but not specific to Swig bindings? Should/can we move these to the SB headers and have Swig make use of those? In other words, can/should files like `SBAddressDocstrings.I` contain only docs that are specific to bindings?
> IIUC, @bulbazord feel free to correct me on this, most doc will be generate from doxygen in the header file and only things that are specific to the bindings will be in the interface.  
What Ismail is saying is what I'd like us to move to. Some of the docstrings are not specific to python at all and can be moved to the header files and generated from doxygen. I think that would best be done in a follow-up change in order to make this one more manageable as there is quite a lot to change already.

Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D142926/new/

https://reviews.llvm.org/D142926