Alright, it finished in a bit less than 24 hours (on a Ryzen 5 with 12
threads).

It failed with

--8<---------------cut here---------------start------------->8---
...
phobos-ldc-v1.14.0/win64.mak
patching file std/socket.d
Hunk #1 FAILED at 448.
1 out of 2 hunks FAILED -- saving rejects to file std/socket.d.rej
source is under 'phobos-ldc-v1.14.0'
applying '/gnu/store/72h2qglkyvhcajiwwh2km7k46zrb7yry-ldc-disable-phobos-tests.patch'...
Backtrace:
           3 (primitive-load "/gnu/store/c6m8yq9rx0hlab7wbjl7523xhpk…")
In ice-9/eval.scm:
    619:8  2 (_ #(#<directory (guile-user) 5cf140> "phobos-ldc-v1.…"))
In srfi/srfi-1.scm:
    640:9  1 (for-each #<procedure apply-patch (a)> ("/gnu/store/72…"))
In guix/build/utils.scm:
    616:6  0 (invoke _ . _)

guix/build/utils.scm:616:6: In procedure invoke:
Throw to key `srfi-34' with args `(#<condition &invoke-error [program: "/gnu/store/frpnp3d1ggybwmfhpy7hqimmrndndxa9-patch-2.7.6/bin/patch" arguments: ("--force" "--no-backup-if-mismatch" "-p1" "--input" "/gnu/store/72h2qglkyvhcajiwwh2km7k46zrb7yry-ldc-disable-phobos-tests.patch") exit-status: 1 term-signal: #f stop-signal: #f] 56dd00>)'.
--8<---------------cut here---------------end--------------->8---

Looks like the patch is out of date.  Did it every work for you, Pjotr?

I've tried fixing it but somehow Guix is recompiling everything again.
I'll let it run for a bit, but I can't proceed if it takes 24 hours to
iterate, so I'd like to understand why it rebuilds the world every
time.

-- 
Pierre Neidhardt
https://ambrevar.xyz/