[#4040] Align output for parallel builds

[chshersh]

Resolves #4040

• Patches conform to the coding conventions.
• Any changes that could be relevant to users have been recorded in the changelog.
• The documentation has been updated, if necessary.
• If the change is docs-only, [ci skip] is used to avoid triggering the build bots.

I've mostly copy-pasted this commit and tweaked it a little bit:

• hvr@d1e3ff7

Also I've noticed some similar output inside cabal-install/Distribution/Client/Install.hs module but I'm not sure whether it's in scope of this issue.

I've tested it by building cabal-install and manually running generated executable over some of my projects. Here is how output looks like:

Downloading  parser-combinators-1.0.0
Downloaded   parser-combinators-1.0.0
Downloading  integer-logarithms-1.0.2.1
Starting     parser-combinators-1.0.0 (lib)
Building     parser-combinators-1.0.0 (lib)
Installing   parser-combinators-1.0.0 (lib)
ignoring (possibly broken) abi-depends field for packages
Completed    parser-combinators-1.0.0 (lib)
Downloaded   integer-logarithms-1.0.2.1
Downloading  case-insensitive-1.2.0.11
Starting     integer-logarithms-1.0.2.1 (lib)
Building     integer-logarithms-1.0.2.1 (lib)
Installing   integer-logarithms-1.0.2.1 (lib)
ignoring (possibly broken) abi-depends field for packages
Completed    integer-logarithms-1.0.2.1 (lib)
Downloaded   case-insensitive-1.2.0.11
Downloading  unordered-containers-0.2.9.0
Starting     case-insensitive-1.2.0.11 (lib)
Building     case-insensitive-1.2.0.11 (lib)
Downloaded   unordered-containers-0.2.9.0
Downloading  scientific-0.3.6.2
Starting     unordered-containers-0.2.9.0 (lib)
Installing   case-insensitive-1.2.0.11 (lib)
ignoring (possibly broken) abi-depends field for packages
Completed    case-insensitive-1.2.0.11 (lib)
Building     unordered-containers-0.2.9.0 (lib)
Downloaded   scientific-0.3.6.2
Downloading  megaparsec-6.5.0
Starting     scientific-0.3.6.2 (lib)
Downloaded   megaparsec-6.5.0
Building     scientific-0.3.6.2 (lib)
Installing   scientific-0.3.6.2 (lib)
ignoring (possibly broken) abi-depends field for packages
Completed    scientific-0.3.6.2 (lib)
Starting     megaparsec-6.5.0 (lib)
Building     megaparsec-6.5.0 (lib)
Installing   unordered-containers-0.2.9.0 (lib)
ignoring (possibly broken) abi-depends field for packages
Completed    unordered-containers-0.2.9.0 (lib)
Installing   megaparsec-6.5.0 (lib)
ignoring (possibly broken) abi-depends field for packages
Completed    megaparsec-6.5.0 (lib)
Configuring library for tomland-0.3..
Preprocessing library for tomland-0.3..
Building library for tomland-0.3..
[ 1 of 15] Compiling Toml.Bi.Monad ...
...

[23Skidoo]

Code in D.C.Install is for the old-install command. Since you changed FetchUtils, which is used by both old and new install, it'd be nice if you updated the remaining messages for old install too. Otherwise LGTM.

[23Skidoo]

Looks like there are some test failures due to changed output, otherwise I think this is ready to go in.

[chshersh]

@hvr @23Skidoo I tried to fix expected output. Now locally I see only this error (so let's way what Travis CI will say).

UNEXPECTED FAIL: cabal-testsuite/PackageTests/CustomDep/sandbox.test.hs cabal-testsuite/PackageTests/NewBuild/T3827/cabal.test.hs

It took some time to figure out how to run tests and how to fix output and where is the problem. But it wasn't that difficult actually. Changes in expected output were required because I replaced Installed string with Completed. Without this change output of cabal install was like this:

Resolving dependencies...
Downloading  pretty-simple-2.1.0.1
Downloaded   pretty-simple-2.1.0.1
Starting     pretty-simple-2.1.0.1
Building     pretty-simple-2.1.0.1
Installed pretty-simple-2.1.0.1

And now it's like this:

Resolving dependencies...
Downloading  pretty-simple-2.1.0.1
Downloaded   pretty-simple-2.1.0.1
Starting     pretty-simple-2.1.0.1
Building     pretty-simple-2.1.0.1
Completed    pretty-simple-2.1.0.1

Doesn't affect new-install command.

P.S. I'm actually using cabal from my branch as a build tool on both Ubuntu and Mac OS machines and it is extremely nice that everything works from raw branch and I even have nice and aligned output!

[hvr]

@chshersh This looks good to me; CI basically passed (except for an unrelated out-of-memory glitch during compilation), so I'm merging this. Thanks!