[dy-tea]

See #25063.

This will probably cover most methods. Might be better to have a map of method names or something? Not sure how significant the performance hit of doing these checks is.

[huly-for-github]

Connected to Huly®: V_0.6-24208

[spytheman]

Not sure how significant the performance hit of doing these checks is.

Run ./v run .github/workflows/compare_pr_to_master.v to find out.

[dy-tea]

Looks like at worst about 30% slower. Not exactly amazing. I will think about a more efficient way to do these checks.

[gechandesu]

Just as an idea. Can we prefix the builtin methods to avoid conflicts instead of doing checks? It shouldn't have such a performance impact, but the side effects are unknown

[dy-tea]

Technically you could take the approach of prefixing built-in methods with something and output an error if the user tries naming a variable with that prefix.
That would prevent name collisions entirely with little performance hit.

[dy-tea]

New impl is at worst around 4% slower than master, at lot better than before.

[Delta456]

You would need to cover a lot more cases than you think 🥲

[dy-tea]

Indeed there are a lot more cases of this than I expected (let me know if I am missing any obvious ones). Notably I added a lot more string, array and map methods. The second check I added doesn't appear to significantly affect performance over the one I had previously but there might be a better way to combine the checks?
Will look at it more tomorrow if I have time.

[dy-tea]

My newest implementation seems to have basically no performance impact. I'm covering a lot more cases (although likely still missing a few).

[dy-tea]

Noticed that I broke autofree so I added handling for that too. Let me know if I'm missing testcases, otherwise the functionality is there.

[spytheman]

The code from the issue #25063 , still leads to a cgen error 🤔.

#1 05:40:19 ^ master ~/v>git switch -
Switched to branch 'pr_25178'
#0 05:40:23 ^ pr_25178 ~/v>./v self
V self compiling ...
V built successfully as executable "v".
#0 05:40:27 ^ pr_25178 ~/v>./v vlib/v/tests/clash_var_fn_name_test.v
#0 05:40:37 ^ pr_25178 ~/v>cat /home/delian/code/misc/2025_09_03__05/xx.v
fn main() {
        mut r := rune(0)
        rune_bytes := r.bytes()
        _ = rune_bytes
}

#0 05:40:44 ^ pr_25178 ~/v>./v /home/delian/code/misc/2025_09_03__05/xx.v
================== C compilation error (from tcc): ==============
cc: /tmp/v_1000/xx.01K46Q5KDGB27TBYVVGWBQ5RKA.tmp.c:1544: error: '{' expected (got ";")
=================================================================
(You can pass `-cg`, or `-show-c-output` as well, to print all the C error messages).
#1 05:40:51 ^ pr_25178 ~/v>./v -cc clang-18 /home/delian/code/misc/2025_09_03__05/xx.v
================== C compilation error (from clang-18): ==============
cc: /tmp/v_1000/xx.01K46Q5V6Z5PE53QK72MZC7690.tmp.c:1965:12: error: initializing 'Array_u8' (aka 'struct array') with an expression of incompatible type 'Array_u8 (rune)' (aka 'struct array (unsigned int)')
cc:  1965 |         {Array_u8 _ = rune_bytes;}
cc:       |                   ^   ~~~~~~~~~~
cc: 1 error generated.
======================================================================
(You can pass `-cg`, or `-show-c-output` as well, to print all the C error messages).
#1 05:40:59 ^ pr_25178 ~/v>

V generated:

VV_LOC void main__main(void) {
    rune r = ((rune)(0));
    Array_u8 _var_rune_bytes = rune_bytes(r);
    {Array_u8 _ = rune_bytes;}
    ;
}

[spytheman]

On master, the generated code was:

VV_LOC void main__main(void) {
    rune r = ((rune)(0));
    Array_u8 rune_bytes = rune_bytes(r);
    {Array_u8 _ = rune_bytes;}
    ;
}

[dy-tea]

Can't believe I missed that 🥲. I will add some extra tests.

Edit: yes the fix was quite simple, I'm also adding a test for assignment.

[dy-tea]

CI seems to be timing out on installing dependencies for some checks. Should I rerun?

[spytheman]

CI seems to be timing out on installing dependencies for some checks. Should I rerun?

Yes. Github Actions runners are flaky in that way unfortunately :-| .

[spytheman]

Excellent work.

[spytheman]

Sorry, but after merging it, the performance is much worse:

I've reverted the change for now.