Issue 14541 - "duplicate COMDAT" linker error with the template forward reference in Tuple.opAssign

[9rnsr]

https://issues.dlang.org/show_bug.cgi?id=14541

The "error reproduction instantiation" might succeed to finish semantic analysis by the forward reference resolution. When it happens, the cached error instance needs to be overridden by the succeeded instance.

[rainers]

s/hasElablrateAssign/hasElaborateAssign/g
s/alredy/already/g

[9rnsr]

Thanks, fixed.

[rainers]

I'm not really qualified to comment sensibly, but getting a different result while trying to reproduce an error message seems strange. Should the retrying be part of the actual instantiation?
Could the same happen with a third invocation, i.e. the first two produce an error, but the third does not?

[9rnsr]

getting a different result while trying to reproduce an error message seems strange.

Thanks for your understanding and question. In short, the two instantiations are not exactly same things. (Sorry my old comment was not good.)
Please read the updated comments explaining the point of issue.

[rainers]

Thanks for explanations. I agree that caching the successful result is an improvement, and I'm going to merge this if there are no other objections.
But the existing design seems to be pretty brittle:

enum hasElaborateAssign(T) = is(typeof(foo!T())) || true;

//static assert(hasElaborateAssign!int);
static assert(is(typeof(foo!int())));

void foo(T)()
{
    static assert(hasElaborateAssign!T);
}

This compiles, but if you insert the static assert in the third line, you get an error test.d(4): Error: static assert (is(typeof(__error))) is false, i.e. the assertion that passed before, suddenly fails.

[rainers]

Auto-merge toggled on

[9rnsr]

But the existing design seems to be pretty brittle:

Oh, that's brilliant small snippet. That's just an opposite pattern of issue 14541.
I have a quick fix for that, but it would also need more thought for the performance.

Could you please file the issue in bugzilla in order not to forget it.

[rainers]

Could you please file the issue in bugzilla in order not to forget it.

Done: https://issues.dlang.org/show_bug.cgi?id=14803

[9rnsr]

Thanks!

[rainers]

Not sure why the auto-merge was disabled, retrying...

[rainers]

Auto-merge toggled on

[9rnsr]

Oh, I didn't notice that... thanks again.