Ian Lynagh authored 13 years ago and pcapriotti committed 13 years ago

And we don't build them by default, as they shouldn't go into the
release.

It's less precise now, but still good enough for what we need, and
this way we don't need to have literal tabs in the file.

I've also renamed the bindisttest install directory to contain multiple
spaces, so we are also testing that multiple spaces aren't getting
collapsed into a single space.

This makes "make" keep working after a "./validate --fast".

The --fast option now disables the following:
  - dynamic libs
  - bindist and bindisttest

Which knocks several minutes off validate for me, but it's still over
30 minutes using 5 cores on 64-bit Linux.

Usual caveats apply: if you're using --fast, then make sure you aren't
doing anything that might destabilise dynamic libs or binary dists.

We don't actually enable the test yet, but when we do validate will
check whether it fails.

This reverts commit e44c591c.

for debugging the build system.

This fixes the ghcpkg05 test failure.
(ghc-pkg check was complaining because the mtl docs didn't exist).

some more tests to run (those with reqlib('mtl')) and serves as an
extra sanity check on Cabal and the bindist.

If we aren't validating and mk/build.mk doesn't exist then boot
prints a warning, pointing at instructions for setting up mk/build.mk.

rather then the bootstrapping compiler. This fixes problems where the
bootstrapping compiler dynamically links against libraries not on the
target machine.

You can now run "make test" in the root, and the fast testsuite will be
run with cleaning enabled. It will also put the summary in
testsuite_summary.txt.

This stops us having to worry about sh/sed/... portability.

on Windows, as they are now in-tree

Cleaning should always succeed now

We only need the forwardslash mode now, so always use that mode and
don't accept any arguments.