Fix ghci being launched before other sources are built

[lukel97]

The following case previously failed

$ cat foo.c
int foo() { return 42; }
$ cat Lib.hs
module Lib where

foreign import ccall "foo" foo :: Int

bar = foo
$ cat cabal-csrc-repl.cabal
cabal-version:      2.4
name:               cabal-csrc-repl
version:            0.1.0.0
library
    exposed-modules:  Lib
    build-depends:    base ^>=4.14.0.0
    C-sources:        foo.c
    default-language: Haskell2010
$ cabal v2-repl
Resolving dependencies...
Build profile: -w ghc-8.10.1 -O1
In order, the following will be built (use -v for more details):
 - cabal-csrc-repl-0.1.0.0 (lib) (first run)
Configuring library for cabal-csrc-repl-0.1.0.0..
Preprocessing library for cabal-csrc-repl-0.1.0.0..
GHCi, version 8.10.1: https://www.haskell.org/ghc/  :? for help
[1 of 1] Compiling Lib              ( Lib.hs, interpreted )
Ok, one module loaded.
*Lib> foo
ghc: ^^ Could not load '_foo', dependency unresolved. See top entry above.

Looks like an accident from a6e427a

[lukel97]

ping @phadej

[DanielG]

LGTM

[phadej]

I have troubles writing a changelog entry for this fix.

The title of this PR doesn't make sense standalone.

[lukel97]

@phadej maybe something like

Fix an issue where cabal repl would fail to build C sources before launching the GHCI session

[phadej]

I'm suspecting changes in run cause haskell/network#474 regression.

If it so, I'll will revert this PR.

EDIT: see https://lists.gforge.inria.fr/pipermail/cado-nfs-commits/2015-August/005301.html

turns out configure can conceivably be called with fd 0 closed, e.g. in
an unattended cluster script. For example this groans.
$ (test -n "$DJDIR" || exec 7<&0) 0<&-
bash: 0: Bad file descriptor

fd 0 is stdin, IIRC

[phadej]

Update, it wasn't that. With cabal-install-3.2.0.0 the

cabal install --lib network 0<&-

also fails.

I.e. inheriting stdin makes cabal fail configure.

[phadej]

Backported to 3.2 in #7134