WIP: Drop Phobos dependency for compiler tests

[kinke]

The goal here is to render make dmd-test totally independent from Phobos, only depending on druntime in the same repo here.

[dlang-bot]

Thanks for your pull request and interest in making D better, @kinke! We are looking forward to reviewing it, and you should be hearing from a maintainer soon.
Please verify that your PR follows this checklist:

• My PR is fully covered with tests (you can see the coverage diff by visiting the details link of the codecov check)
• My PR is as minimal as possible (smaller, focused PRs are easier to review than big ones)
• I have provided a detailed rationale explaining my changes
• New or modified functions have Ddoc comments (with Params: and Returns:)

Please see CONTRIBUTING.md for more information.

---

If you have addressed all reviews or aren't sure how to proceed, don't hesitate to ping us with a simple comment.

Bugzilla references

Your PR doesn't reference any Bugzilla issue.

If your PR contains non-trivial changes, please reference a Bugzilla issue or create a manual changelog.

Testing this PR locally

If you don't have a local development environment setup, you can use Digger to test this PR:

dub run digger -- build "master + dmd#16587"

[kinke]

Currently just useful to get a CI-generated list of problematic tests that would need to be adapted (incl. handling the ^^ operator). E.g., from https://github.com/dlang/dmd/actions/runs/9528085481/job/26265410251?pr=16587:

FAILED targets:
- runnable/xtest46_gc.d
- runnable/argufilem.d
- runnable/xtest46.d
- runnable/arrayop.d
- runnable/paranoia.d
- runnable/link11069b.d
- runnable/testassign.d
- runnable/testabi.d
- runnable/stress.d
- runnable/test5943.d
- runnable/test13117.d
- runnable/ctorpowtests.d
- runnable/test12197.d
- runnable/wc3.d
- runnable/wc.d
- runnable/test13117b.d
- runnable/testaa2.d
- runnable/wc2.d
- runnable/builtin.d
- compilable/compile1.d
- compilable/issue17167.sh
- compilable/test15019.d
- compilable/json2.d
- compilable/paranoia_ctfe.d
- compilable/test5227.d
- compilable/ctfe_math.d
- compilable/b19294.d
- fail_compilation/ice7782.d
- fail_compilation/diag10327.d
- fail_compilation/test22361.d
- fail_compilation/fail21091b.d
- fail_compilation/b3841.d
- fail_compilation/fail21091a.d
- fail_compilation/fail18938.d
- fail_compilation/fail22366.d

The dshell tests have been hacked out for now, as their prebuilt-part (using the fresh DMD, not the host compiler) unfortunately depends on Phobos, and that early build failure hid the actual test results.

[kinke]

It's not just the pow operator, but std.math is special in general - the compiler recognizes std.math function calls during CTFE and handles them in https://github.com/dlang/dmd/blob/master/compiler/src/dmd/builtin.d via hardcoded identifiers. There are multiple compiler tests for these CTFE builtins.

So maybe std.math should be migrated to core.math in druntime.

[rikkimax]

I looked at the power function in std.math, it's a pretty involved affair to move that out, if I remember right it needs exp.

It's not just std.math that has to be moved either, it's multiple modules.

It doesn't work in -betterC because not all of it is templated, it's horribly coupled to having a binary.

https://issues.dlang.org/show_bug.cgi?id=23538

[kinke]

I looked at the power function in std.math, it's a pretty involved affair to move that out, if I remember right it needs exp.

It's not just std.math that has to be moved either, it's multiple modules.

The entire std.math package of course. IIRC, it's pretty self-contained, shouldn't depend on lots of other Phobos stuff.

It doesn't work in -betterC because not all of it is templated, it's horribly coupled to having a binary.

Please keep the discussion on this topic, that's entirely unrelated. FWIW, I've always been against dummy-templatizing druntime/Phobos functions just for the sake of providing symbols for the crappy -betterC semantics. If builds are supposed to remain fast, then why would I need to analyze and codegen some common math functions for every debug build, if the 3 float variants can be handled via prebuilt overloads to be linked from a library? If we manage to untangle the druntime and Phobos deps, then -betterC semantics could at one point include linking a pay-as-you-go druntime, where such math functions would be taken from, without dragging in all the rest.

[rikkimax]

If we manage to untangle the druntime and Phobos deps, then -betterC semantics could at one point include linking a pay-as-you-go druntime, where such math functions would be taken from, without dragging in all the rest.

Actually I've been thinking about this very issue, we already link in per binary an object file, why not include some other stuff with it with GC'able sections?

Note: I mentioned -betterC as that is how I found it. Great way to find issues such as this.