See cabal-testsuite/README.md for a detailed description
of the new architecture.

Signed-off-by: Edward Z. Yang <ezyang@cs.stanford.edu>

Edward Z. Yang authored 8 years ago and Edward Z. Yang committed 8 years ago

Previously, the code would unconditionally report that internal
dependencies were satisfiable, even if --exact-configuration was
provided, we're in per-component mode, and the internal lib was
NOT supplied via a --dependency parameter.  Now ./Setup configure
correctly fails at the right stage.

There is a bit of wibbling to do with compatibility package names.
I think I've gotten it right.

Signed-off-by: Edward Z. Yang <ezyang@cs.stanford.edu>

Edward Z. Yang authored 8 years ago and Edward Z. Yang committed 8 years ago

Signed-off-by: Edward Z. Yang <ezyang@cs.stanford.edu>

Add `--index-state` support to `cabal get`

[skip ci]

Move functions for creating solver HUnit test cases into a new module.

This commit moves the utility functions from
UnitTests.Distribution.Solver.Modular.Solver to a new module,
UnitTests.Distribution.Solver.Modular.DSL.TestCaseUtils.

With this `cabal get` gains the ability to

 a) unpack and update with the .cabal revision which was
    active at the given index-state, as well as

 b) allow to unpack the most recent package version
    (while satisfying imposed version constraints) as of
    the given index-state

The `--index-state` flag can be combined with `--pristine` if only b) is
desired.

Handle undefined MIN_VERSION_base in Distribution.Utils.BinaryWithFin…

Fix copy-paste mistake in error message

Fix implementation of addBuildableCondition

Oops! Forget to delete old PkgconfigDependency during refactor

Dedicated modules for `*Dependency` types and `UnqualComponentName`

...into Distribution.Utils.Generic. Distribution.Simple.Utils moved items.

Also gets rid of `Distribution.Packages.TextClass`

Deduplicate library directory arguments when testing libraries

Edward Z. Yang authored 8 years ago and Edward Z. Yang committed 8 years ago

Now that cabal-testsuite uses a Custom setup, we can assume
that we are able to read the LBI from our build.  This is good
news, because the old algorithm didn't really work at all
(and stopped working with cabal-install-1.24.0.1)

Fixes #4108.

Signed-off-by: Edward Z. Yang <ezyang@cs.stanford.edu>

Since this fixes #3858

The addBuildableCondition function was added to solve the problem with
"buildable: False". The problem was that we would solve or check
dependencies on the basis of the component in question being needed, and
then at the end discover that the component is actually not buildable at
all, and if we'd known that up front we would not have solved for the
component's dependencies.

The trick that addBuildableCondition does is a syntactic transformation,
from components like:

executable blah
  buildable: False
  build-depends: foo >= 1, bar < 2
  something-else: whatever

to:

executable blah
  -- empty!

Or at least, that's the intention. In the above situation the
implementation of addBuildableCondition returns an empty CondNode:

CondNode mempty mempty []

The type at which mempty is used is important here. This transformation
is used in two places: one in the solver and the other in finalizePD.
In the solver the mempty is used at types from the PackageDescription:
Library, Executable, TestSuite etc. So in this case the transformation
works fine we end up with empty executables, test suites etc.

In finalizePD however the mempty gets used at type PDTagged (which is
sort of a union of Library, Executable etc plus none/null) and the
mempty for PDTagged is PDNull which means it does not even specify
which component we're referring to. So effectively that means instead of
ending up with an empty executable in the above example, we end up
deleting the executable entirely!

This was a change in behaviour. Prior to adding addBuildableCondition
the result of finalizePD would include non-buildable components and the
rest of the build system infrastructure was set up to skip over them
when building. The change was not noticed precisely because the rest of
the system was already set up to ignore non-buildable components.

This is not however a benign change in behaviour. In particular in
cabal-install in the install plan we end up completley forgetting about
all the non-buildable components. This means we cannot even report that
components are non-buildable when users ask to build them, because we've
completely forgotten that they exist.

So this patch keeps the original addBuildableCondition for use by the
solver since the solver uses it at sensible monoid types. The patch adds
a special version for the PDTagged type which changes the transformation
so that in the above example we end up with:

executable blah
  buildable: False
  something-else: whatever

So we've stripped out all the build-depends but we keep everything else,
including of course the "buildable: False".

Fixes #4092.

Signed-off-by: Edward Z. Yang <ezyang@cs.stanford.edu>

fix comment typo of security

UnitId was already opaque but missed the ShortText treatment;
this allows for simple re-casting from ComponentIds (which are
internally ShortTexts as well)

D.C.Compat.Process.readProcessWithExitCode: also catch permission errors

Closes #1613.

Implement +nowrap verbosity flag to suppress log-msg line-wrapping

Unfortunately, propagating the `Verbosity` to the `topHandler`
proves to be more complicated.

This is mostly a problem for `die` invocations which are translated
into user-error exceptions (which in turn then cause callstacks
to be printed; are we abusing `die`?)

Improve error output when package fails to build.

Fix redundant import warning from Parsec.

Add more fields to plan.json schema

[skip ci] to reuse ci-status from PR

Signed-off-by: Edward Z. Yang <ezyang@cs.stanford.edu>

This adds the following new fields:

 - `.compiler-id`
 - `.os`
 - `.arch`
 - `.install-plan[].pkg-name`
 - `.install-plan[].pkg-version`
 - `.install-plan[].pkg-src-sha256`

Having separate "pkg-{name,version}" fields is desirable because the "id"-field
is actually a UnitId which is supposed to be an opaque id which may not
always be easily convertible into a PackageId via string operations.

The "pkg-src-sha256" is generally useful to uniquely identify a
source-tarball (when the package-id is not exact enough).

Finally, the compiler-id/os/arch information is needed to determine the
values of the respective impl()/os()/arch() predicates used in
conditionals in .cabal files.

Here's an example demonstrating the new fields:

    {
      "cabal-version": "1.25.0.0",
      "cabal-lib-version": "1.25.0.0",
      "compiler-id": "ghc-8.0.1",
      "os": "linux",
      "arch": "x86_64",
      "install-plan": [
	{
	  "type": "configured",
	  "id": "StateVar-1.1.0.4-048d5d25d5813f17f152f9e766fe3ea0ef6f5317439aa115f3fc24e53a0e3c17",
	  "pkg-name": "StateVar",
	  "pkg-version": "1.1.0.4",
	  "flags": {},
	  "style": "global",
	  "pkg-src-sha256": "7ad68decb5c9a76f83c95ece5fa13d1b053e4fb1079bd2d3538f6b05014dffb7",
	  "depends": [
	    "base-4.9.0.0",
	    "stm-2.4.4.1-2c7789312d2396a91db173e1ae8cf3a9af3dd18de42bb1021c4978fbab49c191",
	    "transformers-0.5.2.0"
	  ],
	  "exe-depends": [],
	  "component-name": "lib"
	},
        {
          "type": "pre-existing",
          "id": "rts",
          "pkg-name": "rts",
          "pkg-version": "1.0",
          "depends": []
        },
	{
	  "type": "configured",
	  "id": "hackage-matrix-builder3-0.3-inplace-matrix-lib",
	  "pkg-name": "hackage-matrix-builder3",
	  "pkg-version": "0.3",
	  "component-name": "lib:matrix-lib"
	  ...
	}
	...
    }

New error message is:

cabal: Failed to build p-1.0 (which is required by exe:e4 from p-1.0, exe:e3
from p-1.0 and others).

Fixes #4081.

TODO: Use these utility functions in other appropriate places.

Signed-off-by: Edward Z. Yang <ezyang@cs.stanford.edu>

Edward Z. Yang authored 8 years ago and Edward Z. Yang committed 8 years ago

The idea is we can use Rep to get a full, structural representation
of a type, and the fingerprint it using Typeable.  This gives
us a very concise way of fingerprinting our Binary representation.

This patch is not completely correct; the fingerprint needs
to be overridable when someone writes a custom Binary instance.
But this should be "good enough" in practice; we're not using
these fingerprints to check anything security critical.

TODO: Not sure if I have tagged all the call-sites which could
profit from this.

Fixes #4059.

Signed-off-by: Edward Z. Yang <ezyang@cs.stanford.edu>

Edward Z. Yang authored 8 years ago and Edward Z. Yang committed 8 years ago

Signed-off-by: Edward Z. Yang <ezyang@cs.stanford.edu>

No LambdaCase, resolve #4079