remacs
Port char-width
DEFUN ("char-width", Fchar_width, Schar_width, 1, 1, 0,
doc: /* Return width of CHAR when displayed in the current buffer.
The width is measured by how many columns it occupies on the screen.
Tab is taken to occupy `tab-width' columns.
usage: (char-width CHAR) */)
(Lisp_Object ch)
{
int c;
ptrdiff_t width;
CHECK_CHARACTER (ch);
c = XINT (ch);
width = char_width (c, buffer_display_table ());
return make_number (width);
}
I'd like to port this function. Presumably, I'll also need to port the function char_width, but I'm not sure what to do with buffer_display_table, i.e. I'm not sure where it's defined, or whether I can access it from Rust. I'd appreciate a little help.
You can start by calling them both from C.
https://github.com/emacs-mirror/emacs/blob/78ec68e18f07a90a9ad400683b973ff51baa80e1/src/indent.c#L60
All C functions can be imported from remacs_sys.
A simple port to Rust would call the C functions. Which is why this is labelled as a good first issue. A more interesting port would look at the C functions called and bring them along too.
I'm looking at the C function buffer_display_table, which references macros BVAR, XCHAR_TABLE, and DISP_TABLE_P. I found the definition for BVAR and DISP_TABLE_P, but not XCHAR_TABLE or at least one of the macros used by DISP_TABLE_P. I'm using rg to try and find original definitions in the C code. Are those macros I'm missing defined in external C libraries?
Trying to delve into the C for a better understanding, but in the meantime I'll go for the simple port option and open a WIP PR.
INLINE struct Lisp_Char_Table *
XCHAR_TABLE (Lisp_Object a)
{
eassert (CHAR_TABLE_P (a));
return XUNTAG (a, Lisp_Vectorlike);
}
Defined around line 1818 of lisp.h.
In general XFOO means "hard cast to FOO, do no checks". This is an exception.
BTW, check out https://github.com/Wilfred/deadgrep I have it bound to C-c g.
So my rg didn't find it because it's not a #define macro. Is it a macro at all? Is INLINE a macro that contains a nested #define? What's a better regex to search for definitions of X___() conversions?
I guess by using into() I shouldn't necessarily need to understand what types the C functions are using?
I see from the most recent commit that make_number is no longer a thing? Should I replace the usage of make_number in the C char_width() function with a match val.into() expression like is used here?
If that's the case, is it an idiom that we could DRY somehow, either with a function or a macro?
Probably it's ok if you use into when the return type is EmacsInt. But I don't like it when you can't see the resulting type.
INLINE in C works like the inline directive in Rust. It causes the compiler to replace every instance where the function would be called with the code of the function instead. This reduces oveerhead of building call stacks, jumping in and out of functions, etc. Mentally it behaves like a macro but you have to reason about it as if it is a function.
As for finding the XFOO items look for the define, then look for a line starting with XFOO. That will catch inlines. Or simply look at headers which is all you need to learn about the type.
We consider the into and from usage reasonably DRY. We have been adding force_foo methods lately to mimic the XFOO behavior of not performing checking where it makes sense. Essentially the X routines in C are equivalent to thing.into().unwrap() most of the time.
These are all really good questions @NQNStudios. Thanks for taking the time.
Is this fine?
use crate::{
// ...
remacs_sys::{buffer_display_table, char_width, CHECK_CHARACTER},
};
/// Return width of CHAR when displayed in the current buffer.
/// The width is measured by how many columns it occupies on the screen.
/// Tab is taken to occupy `tab-width' columns.
#[lisp_fn(c_name = "char_width", name = "char-width")]
pub fn char_width_lisp(ch: LispObject) -> EmacsInt {
CHECK_CHARACTER(ch);
let c: usize = ch.into().unwrap();
let width = char_width(c, buffer_display_table());
width.into().unwrap()
}
EDIT: this is NOT fine.
@A6GibKm why is this not fine? would you like to elaborate? would be nice to finish this task, seems like you almost got it ;)
@A6GibKm please post a PR, then we can comment on the code as you make progress.