[question] Does diamond graphs configure() methods that remove options shared/fpic could be broken?

[memsharded]

It is possible if evaluated more than once, with different branches values.

[SSE4]

I spotted a serious issue in validate implementation: #8383

[memsharded]

This is not an issue at all in the validate(), it is working as expected. It is an issue in the options processing: #8383 (review)

[memsharded]

The new graph resolution model in Conan 2.0 will try not to re-evaluate same recipe when diamonds, just check consistency. So that might solve this issue, which is otherwise simply too risky to solve for Conan 1.X

[memsharded]

Solved in 2.0, closing.