Replace the current brute force implementation of
rfc822_next_header_of_name() with one using a hash table to (lazily) index
the header fields by name.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

This patch adds functions to the rfc822 module to retrieve header fields
of a given name.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

Currently, calling bytestring_from_string(NULL) will result in a SEGV
within strlen().  This patch makes this construct safe, returning
bytestring_NULL, which seems a less surprising result.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

Currently the test infrastructure for constructing example messages then
parsing them assumes that constructed headers never have errors of any
kind.  That's true so far, but it limits the versatility of this test
apparatus.  This patch extends the infrastructure to allow minor errors
(that is things other than a missing colon) to be handled.  The existing
test data is also extended to include cases which use this.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

David Gibson reports a new format, meaning we didn't show the gcov files
with -vv.

This error constant is actually more specific than the name suggests - it
indicates that a header field name contains characters which are not
permitted in a header field name.  Rename the constant to better reflect
this, likewise rename the testcase based around it.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

The run-bad-header.c test in the rfc822 module is actually testing handling
of one specific sort of bad header - a header field which contains no
colon character.  Rename the test and its internal variables to better
reflect this.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

The check_header() test function will be useful to more than one testcase
so this patch moves it into the common helper.c file.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

Add a simple full-blown example program for the rfc822 module.
'headernames' simply dumps the header field name for every header field in
each message file given on the command line to stdout.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

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

Approved by all contributors:
https://lists.ozlabs.org/pipermail/ccan/2012-September/000490.html
https://lists.ozlabs.org/pipermail/ccan/2012-September/000488.html
https://lists.ozlabs.org/pipermail/ccan/2012-September/000489.htmlSigned-off-by: Rusty Russell <rusty@rustcorp.com.au>

And enhance the documentation a little.

Times can be invalid: non-normalized, or negative.  

It's a superset of struct timeval.  It's also a bit trickier, since we
need -lrt on Linux to use clock_gettime(), and also we can't just cram the
time into a u64 for multiply and divide.

If we do have it, and we redefine it, compilation breaks.

In particular, HAVE_CLOCK_GETTIME_IN_LIBRT.

We intuit the dependencies of embedded examples by looking for #include.
But these extra dependencies may require extra libraries, so we need to
add those, too.

If a dependent module needs -l options, we don't notice.  Fix that, now
we have one.

We let the get_deps() caller hand us the compiled _info filename, but what
about recursive dependencies?  These we re-generate every time.

So fix this: hand a generator callback to get_deps(), expose the one
which simply compiles it, and add a ccanlint one which looks up the
manifest to see if we have one already.

Before:
$ ccanlint -vvvv ccan/failtest | grep -c 'Creating.*_info'
31

After:
$ ccanlint -vvvv ccan/failtest | grep -c 'Creating.*_info'
17

This fixes some format string related bugs that show up on 64-bit builds.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

For convenience, check in a copy of the full text of RFC5322 (the latest
revision of RFC822).
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

In particular, don't complain on conditional lines.  That's less thorough,
but removes false negatives we have now.

next_header_parse() had a bug which would mis-parse an extra header if the
body of the message began with a linear whitespace character (space or
tab).  This patch fixes it and adds a suitable testcase.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

It's only used for testing, but it's LGPLv3 (vs LGPLv2).  But unless we
suppress it, ccanlint exits with status 1 and my sanity checks stop me from
uploading.

Fixes this error:
	/hash.h:304:13: error: two or more data types in
	declaration specifiers

Seems it's not available on Solaris.

This is where you see the genius of the originals: you can't implement them
without horrible hacks to get the program name :(

Seems like Solaris doesn't have err.h, as discovered by Samba.

This adds a "waf" buildtype to create-ccan-tree. When passed it
generates a basic wscript for building everything into a libccan.a

It doesn't support running the generator, so you need to provide
config.h
Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Jeremy Kerr <jeremy.kerr@canonical.com>

Add a new build type to create an appropriate Makefile.am for the
newly-created ccan tree.
Signed-off-by: Jeremy Kerr <jeremy.kerr@canonical.com>

Replace the --exclude-configurator argument with a
single --build-type=TYPE argument, allowing future expansion of build
system types (and allowing source-tree-only creation). We currently
support:

 * 'make' (just the Makefile, equivalent to --exclude-configurator)
 * 'make+config' (current default)
Signed-off-by: Jeremy Kerr <jeremy.kerr@canonical.com>

Rather than explicitly excluding the tests, just copy the source files
by default. Add an option (--copy-all) to include all of the non-source
stuff in each module.
Signed-off-by: Jeremy Kerr <jeremy.kerr@canonical.com>