Delyan Angelov

The generated code is invalid with `clang` (which the Android NDK uses), but compiles with `tcc`. With `gcc`, it compiles with just a warning. => for reproduction on Linux/Windows, we...

On one hand, in my mind, the way tests worked before, by requiring `?`, if the user wanted to propagate the error further, just like ordinary functions, was *more consistent*,...

Great work! I especially liked how adding proxy support on windows, actually removed stuff from `vschannel.c` 🥳 .

(`git pull --rebase origin master` to avoid the unrelated CI failures)

> @spytheman Tests are not passing since there are workarounds in `vlib/v/tests/inout/tmpl_parse_html.vv` for `@` replacements that are no longer necessary. @ was written as @@ when you wanted the literal...

Note that I did put tmpl_parse_html.vv in an explicit skip list in `vlib/v/tests/inout/compiler_test.v` in c70e18e, so another option may be to just rebase the code, instead of fixing `tmpl_parse_html.vv` 🤔...

Thanks for the benchmarks.

I've made some simplifications to the `extract-int.v` and `extract-opt.v` files, to reduce the additional operations, that the loops perform (s.substr especially since it does allocation, b.step, b.ok), and added a...

> Also this PR doesn't endorse safe-by-default behavior - i.e. the syntactically shorter variant shall be the one returning int | none and the syntactically longer variant shall return int....

> using the shorter names for the ?int versions will be a breaking change Actually, I am wrong about this - currently on master, .index() does return an `?int`, so...