This is what users want, and expect: as demonstrated by the test failure
when not under valgrind!
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

While investigating the previous patch, a bug caused poll to return
POLLHUP on the listening socket, which caused us to spin.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

We were closing before calling del_fd, which also closed.

The shutdown() logic applies when a child and parent are using the
*same* socket fd to communicate to each other.  That's really unusual
(who would you connect to?), and should probably be done by the user.

Generally, you'd use socketpair() for this child-parent case.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Helps with the common case of wanting to NULL out a pointer when the object
freed.  We could also track it if resized, but that's TODO.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

There are several times I've wanted an extra arg to the destructor, and had
to embed it in the thing destroyed.  It's more efficient to put it into
tal itself (since it allocates space anyway), but we make it conditional
on a flag to avoid bloating every destructor.

The infrastructure makes it easier to add an extra arg to the general
notifiers later if we want.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

io_close() currently marks the io_conn for freeing, but doesn't
actually do it.  This is a problem for tal() users, because we can't
just call it in the parent's constructor.

Make io_close() just tal_free() + return &io_conn_freed (a magic
io_plan pointer).
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

It seemed like a good idea, but it complicates things and I never used
it (since I never really trusted that the alternate paths would be
equivalent).
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

So the errno when you call tal_free() is handed to all the notifiers,
independent of what the others do.

This makes sense, but also helps for the upcoming ccan/io change.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Currently, _info files can specify options, or note expected failures, for
ccanlint checks in the _info file with specially structured comments. That
differs from most other things ccanlint gets from _info, where it instead
executes the info file with certain parameters.

This changes ccanlint and existing _info files to use the normal method for
the ccanlint test options as well.  This also has the advantage that an
info file can alter its test options based on things from config.h - in
some cases whether a test can work or not might depend on various things.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

Changing the _info file can change how ccanlint assesses the module.
Therefore, if the _info file changes, we should re-run ccanlint module
tests with make check.  We didn't previously have a dependency for that,
though, so this adds it.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Two copies of this map meant only the tests would ever work.

Use the new coroutine module to abstract away our dependence on
ucontext.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

This is essentially a wrapper around ucontext.h, but the idea is that
alternative back end implementations could be used in future.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

lightning uses 48 bits, so this provides a quick utility for our test
vectors, and provides a nice example.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Making arbitrary bytes a uchar* is always a bit nasty for the caller.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

The new Makefile system, via the helper script in tools/gen_deps.sh, when
generating the targets to test a module, inserts dependencies meaning it
must first check modules this one depends on, whether via 'depends' or
'testdepends' in _info.

Although it seems logical, including 'testdepends' is actually incorrect.
If ccan/a testepends on ccan/b then ccan/b must be *built* in order to test
ccan/a, but it doesn't need to be tested.  testepends are explicitly
permitted to contain loops - it's quite common for two complementary
modules to be used to test each other.  This is one of the reasons
testdepends exists separate from depends.

So, remove testdepends from the generated check dependencies, removing the
circular dependency that Make complains about.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

The previous name was misleading, since it does not define the number of
elements (ed_elem) on the stack, but rather the number of distance
values (ed_dist).  Rename to make this more clear and add more
documentation about what it does and how best to define it.

Note:  This is an API change for custom-compiled versions, but since the
module has only been included for a couple days I don't think it's worth
a back-compat #ifdef at this point.
Signed-off-by: Kevin Locke <kevin@kevinlocke.name>

Currently Travis builds fail on the edit_distance module, because that
module uses C99 extensions, which aren't enabled by default on the
Ubuntu Precise compiler version.  Force it to allow these, by adding the
-std=gnu99 option to the compiler.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

This commit creates the edit_distance module for calculating various
edit distance measures.  The edit distance quantifies the similarity
between two strings based on the number of modifications necessary to
turn one string into the other.  There are several edit distance
measures which differ in the operations which are permitted and the
cost (aka weight) of the operations.  This module provides functions
for calculating the Longest Common Subsequence (LCS), Levenshtein, and
Damerau-Levenshtein (restricted and unrestricted) distances.  Weighted
versions of these functions can be created by defining cost functions
as preprocessor macros when compiling this module.  Distances over
other array types (e.g. wide strings, integers, structs) can be
accomplished by defining the element type and equality test macros.
Signed-off-by: Kevin Locke <kevin@kevinlocke.name>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

The Makefile provides for "fast" and "full" variants of make check.
However, nothing seems to use or implement the "full" variant at all.
Further, the %.check and %.fastcheck targets are not actually used, with
other targets instead using the actual .ok tag file as the target.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

split() was deprecated, is now removed. explode() is the new hotness.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

Particularly since it's used by a gitolite hook on ozlabs.org to update
the webpages.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>

clang version 3.4, for whatever reason, expands a few of the testcases
in cppmagic with slightly different whitespace from what we expect.  Since
this was changed in later clang versions, it was probably a compiler bug,
but it turns out an extra space in the testcase can make it pass on clang
3.4 as well.  This means we finally have clang builds working in Travis.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

Generally, ccan config variables are always defined as either 0 or 1 and
are tested with #if.  That's instead of being being either defined or
undefined and tested with #ifdef.

TIME_HAVE_MONOTONIC breaks that convention.  This can cause warnings in
ccan/timer which uses it assuming the 0/1 convention.  Change it to remove
that warning.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

Remove a warning caused by the unused 'explain' variable in
headers_idempotent.c.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

There's no real reason to exclude 'generator' from the build or check
targets.  Add it back in.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

The new makefile excludes all modules with external library dependencies
from the usual make check, meaning the libraries are not needed for the
Travis build.

Although it would be nice to include these modules, there are a bunch of
problems with sorting out a correct environment for them.  For now, just
leave them out.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

# Conflicts:
#	.travis.yml
#	Makefile
#	Makefile-ccan

The example code in wwviaudio doesn't compile as is.  Mark it as an
expected failure for ccanlint.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

This fixes a warning.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

This removes the parallel make (-j) option when running make check under
Travis CI.  This will slow down the build somewhat, but makes it easier to
match errors to the modules which caused them.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

--summary (used by make check) causes ccanlint to only ever print a
one-line summary of the results.  When one of the tests fails, this isn't
very useful, because it doesn't say why it failed, or even which test
failed.  This is particularly frustrating for failures in Travis CI builds
where it's not always straightforward to reproduce a similar enough build
environment to check the failure manually for a particular module.

--summary does have one purpose which is vital for make check, though: it
suppresses any attempt to interactively correct / improve the module
printing only results.

This changes --summary to only have the second effect, not the first.  This
won't change the output in the case of passing modules, but will give more
useful information in the case of failing modules.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>