Log message:
update to 2.2.9
changes:
-Fixed a memory leak
-Oops, fseek() should have been using SEEK_SET, not SEEK_END

Log message:
Update to 2.2.8:

2007-04-25  Jeffrey Stedfast

	* README: Bumped version

	* configure.in: Bumped version to 2.2.8

	* tests/test-pgp.c: Test exporting of keys.

	* gmime/gmime-utils.c (rfc2047_decode_word): Fixed compile
	warnings.

	* gmime/gmime-stream-file.c (stream_reset): Removed an unused
	variable.

	* gmime/gmime-charset.c (g_mime_charset_can_encode):
	s/if (mask->level = 1)/if (mask->level == 1)/

2007-04-23  Jeffrey Stedfast

	* README: Bumped version

	* configure.in: Bumped version to 2.2.7

2007-04-14  Jeffrey Stedfast

	* gmime/*.c (g_mime_*_get_type): Set n_preallocs to 0.

2007-04-12  Jeffrey Stedfast

	* gmime/*.c: no need for a second NULL argument to g_object_new()

	* util/cache.c (cache_new): Change max_size and node_size to be of
	type size_t.

	* gmime/gmime-multipart-encrypted.c
	(g_mime_multipart_encrypted_new): g_object_new() doesn't need a
	second NULL argument.

	* gmime/gmime-utils.c (decode_8bit): Close the iconv descriptor
	and since we are using is_ascii() now, we don't need to use
	unsigned char *'s.

2007-04-12  Jeffrey Stedfast

	* gmime/gmime-utils.c (decode_8bit): Use is_ascii().
	(g_mime_utils_header_decode_text): Same.
	(g_mime_utils_header_decode_phrase): Here too.

	* gmime/gen-table.c: Added a is_ascii() macro for use instead of
	the ctype isascii() so that I don't have to worry about casting.

2007-04-11  Jeffrey Stedfast

	Revision 1119 (previous commit) made the following 2 functions
	even less attractive than they already were, so I decided to
	rewrite them especially since it wasn't hard to find a far cleaner
	approach.

	* gmime/gmime-utils.c (g_mime_utils_header_decode_text): Rewritten
	to be cleaner, faster, and more elegant.
	(g_mime_utils_header_decode_phrase): Same.

2007-04-11  Jeffrey Stedfast

	Fixes for bug #423760 and bug #342196

	* gmime/gmime-charset.c (g_mime_charset_can_encode): New
	convenience function to check whether a length of UTF-8 text can
	be converted into the specified charset.
	(g_mime_set_user_charsets): New function allowing an application
	to provide GMime with a list of user-preferred charsets to use for
	encoding and decoding headers.
	(g_mime_user_charsets): New function to get the list of
	user-preferred charsets.

	* gmime/gmime-utils.c (decode_8bit): New function to convert
	arbitrary 8bit text into UTF-8 using the charset list provided by
	g_mime_user_charsets().
	(rfc2047_decode_word): Don't assume that just because the declared
	charset is UTF-8 that it actually is in UTF-8.
	(rfc2047_decode_word): If we can't open a converter for the
	declared charset to UTF-8 or if we can't convert from the declared
	charset into UTF-8, fall back to using decode_8bit().
	(g_mime_utils_header_decode_text): Convert 8bit word tokens into
	UTF-8 using decode_8bit().
	(g_mime_utils_header_decode_phrase): Same.
	(rfc2047_encode_word): Be a little more efficient about removing
	'\n' chars...
	(rfc2047_encode): When encoding a level-2 word cluster, attempt to
	fit the cluster within a charset provided by
	g_mime_user_charsets() rather than using GMime's best-fit charset
	table (unless, of course, it doesn't fit within any of the
	user-specified charsets).

2007-03-28  Jeffrey Stedfast

	* gmime/gmime-iconv-utils.c (g_mime_iconv_strndup): No need to
	cast out to a char *, it already is.

	* gmime/gmime-stream-mem.c (g_mime_stream_mem_set_byte_array):
	Only free the previous memory buffer if we were the owner.

Log message:
Update to 2.2.6:

2007-03-28  Jeffrey Stedfast

	* README: Bumped version

	* configure.in: Bumped version to 2.2.6 - critical release for the
	gmime-stream-fs and gmime-stream-file fixes below.

	* gmime/gmime-stream-filter.c (stream_reset): Try resetting source
	stream first, this way if that fails we don't reset our state
	either.

	* gmime/gmime-stream-fs.c (stream_reset): Same as below.

	* gmime/gmime-stream-file.c (stream_reset): Make sure to always
	return 0 or -1, not the value from fseek()ing back to the
	beginning of the stream (which might not be position 0).

2007-03-27  Jeffrey Stedfast

	* gmime/gmime-gpg-context.c (gpg_ctx_parse_status): If we aren't
	in a UTF-8 locale, convert status messages into UTF-8 when using
	the status line in the GError.

2007-03-26  Jeffrey Stedfast

	* gmime/gmime-utils.c (g_mime_utils_8bit_header_decode)
	(g_mime_utils_8bit_header_encode)
	(g_mime_utils_8bit_header_encode_phrase): Changed all of these
	functions to take `const unsigned char *' like they used to. Fixes
	bug #423147.

	* gmime/gmime-gpg-context.c (gpg_ctx_new): Initialize flushed and
	nodata state variables. Thanks to valgrind for finding these...

	* gmime/gmime-content-type.c
	(g_mime_content_type_new_from_string): Shouldn't need to
	g_strstrip() the type, is_ttoken() won't match against lwsp
	chars. Also, instead of simply scanning past lwsp between some
	tokens, use decode_lwsp() between *all* tokens so that we handle
	comments as well (not that there should be any...).

	* gmime-2.0.pc.in: Got rid of the redundant @LIBS@ which contained
	the same value as @GMIME_LIBS@.

	* configure.in: Added configure checks for large file support.

2007-03-26  Jeffrey Stedfast

	* gmime-param.c: #include "gmime-parse-utils.h" and got rid of
	this file's own decode_lwsp() implementation.

	* internet-address.c: #include "gmime-parse-utils.h" which is
	where a bunch of the parser functions have been moved from here.

	* gmime-utils.c: #include "gmime-parse-utils.h" for decode_lwsp()
	and decode_domain()
	(decode_addrspec): Moved here from internet-address.c which had
	been exporting the symbol previously (so that decode_domain didn't
	have to be copy/pasted). Now this function can be made static -
	yay! No more exporting un-namespaced symbols!

	* gmime/gmime-message.c: #include "gmime-parse-utils.h" for
	decode_lwsp()

	* gmime/gmime-parse-utils.[c,h]: New source files containing some
	parse utility functions that had been copy/pasted between several
	source files previously.

2007-03-25  Jeffrey Stedfast

	* README: Bumped version

	* configure.in: Bumped version to 2.2.5

2007-03-25  Jeffrey Stedfast

	* tests/Makefile.am (local-check): Simplified.

	* tests/test-streams.c: Generate our own test data if we aren't
	given any.

	* tests/test-cat.c: Updated for the new Makefile.am rule.

	* gmime/gmime-stream-mmap.c (stream_length): Fixed to return the
	correct length when bound_end == -1 but bound_start != 0.
	Discovered this thanks to my new testsuite.

	* gmime/internet-address.c (decode_mailbox): If we can't find a
	local-part and inptr is ',', set *in = inptr; rather than *in =
	inptr + 1; Discovered this bug thanks to my new testsuite.

	* tests/test-mime.c: Rewritten to use testsuite library.

2007-03-24  Jeffrey Stedfast

	* tests/Makefile.am (check-local): Rewritten to actually run the
	automated tests.

	* tests/test-mbox.c: Rewritten to use testsuite library.

	* tests/test-pgp.c: Rewritten to use testsuite library.

	* tests/test-pgpmime.c: Rewritten to use testsuite library.

	* tests/test-cat.c: Rewritten to use testsuite library.

	* tests/test-streams.c: Rewritten to use testsuite library.

	* tests/test-iconv.c: Rewritten to use testsuite library.

	* tests/testsuite.[c,h]: New automated test-suite helper library.

2007-03-20  Jeffrey Stedfast

	* configure.in: Check for the existance of _timezone and nfds_t.

	* gmime/gmime-utils.c (mktime_utc): Use _timezone if
	HAVE__TIMEZONE is set.

	* gmime/gmime.c (g_mime_init): If HAVE_TIMEZONE or HAVE__TIMEZONE
	is set, call tzset() to initialise the timezone variable.

2007-03-12  Jeffrey Stedfast

	* gmime/gmime-param.c (decode_param): Instead of only checking
	that the param value begins with "=?", check if that sequence
	exists anywhere in the value. Fixes bug #407522.

	* src/uudecode.c (uudecode): Use a strcmp instead of comparing
	against a string literal (which was technically correct in this
	case, but the prettyification would be nice if "-" was specified
	in argv too). Fixes bug #408613.

2007-03-10  Jeffrey Stedfast

	* gmime/gmime-message.c (g_mime_message_get_all_recipients): New
	convenience function to get To, Cc, and Bcc recipients all in a
	single list.

2007-02-19  Jeffrey Stedfast

	* gmime/gmime-message.h: Moved g_mime_message_get_body() into the
	deprecated block.

2007-02-11  Jeffrey Stedfast

	* gmime/gmime-utils.c: Removed the need for the ctype isblank()
	function.

	* gmime/gmime-param.c (decode_int): Need to use (unsigned char *)
	when passing indexes to ctype functions.

2007-02-10  Jeffrey Stedfast

	* gmime/gmime-charset.c: Changed the names of the charset format
	macros to coincide with the new names output by iconv-charset.c.
	(g_mime_charset_iconv_name): Use the new macro names.

	* iconv-detect.c: Use %u instead of %d, also changed the names of
	the format #defines that we print out.

	* gmime/gmime-utils.c (g_mime_utils_quote_string): Optimise a
	smidgen.
	(g_mime_utils_unquote_string): Fixed, this wasn't working right...

2007-02-09  Jeffrey Stedfast

	* gmime/gmime-stream-filter.c (stream_write): Instead of comparing
	the return value of the source stream write() to our input length
	to write, compare to -1. We can't assume the the value the source
	stream will return will be what we expected, all we can do is
	check for error.

	* gmime/gmime-part.c: Same as below.

	* gmime/gmime-filter-basic.c: Fixes to signedness compile warnings
	and updated to reflect gmime-utils.h changes.

	* gmime/gmime-multipart.c (multipart_set_boundary): Updated to
	reflect gmime-utils.h changes.

	* gmime/gmime-stream.c (g_mime_stream_writev): If any of the
	writes fail, return -1.

	* gmime/gmime-utils.c: header_encode_*() and header_decode_*() now
	take const char * rather than const unsigned char *
	arguments. This has been done in an effort to fix compile warnings
	about signed vs unsigned char * pointers. The 'save' argument to
	all of the stream encoders/decoders have been changed to a
	guint32, again for signed vs unsigned warning fixes.

Log message:
Fix build on DragonFly.

Log message:
Update to 2.2.4:

2007-02-08  Jeffrey Stedfast

	* README: Bumped version

	* configure.in: Bumped version to 2.2.4

2007-02-06  Jeffrey Stedfast

	Fix for bug #394433

	* configure.in: Detect if the system has GNU's getopt
	implementation available for us to use.

	* Makefile.am: Don't build our own copy of the GNU getopt library
	if the system has it available.

	* src/uuencode.c: If the system has getopt.h, use it instead of
	our own getopt.h.

	* src/uudecode.c: If the system has getopt.h, use it instead of
	our own getopt.h.

2007-02-06  Jeffrey Stedfast

	More fixes on account of my test suite...

	* tests/test-cat.c (test_cat_substream): end = start + random
	amount... duh. Fixes a bug where sometimes the end bound was < the
	start bound.

	* gmime/gmime-stream-cat.c (stream_substream): Calculate the
	length of each of the source streams correctly. D'oh.
	(stream_substream): Fixed the calculation for s->start.
	(stream_substream): Fixed the s->end calculation for when end
	within the source stream's bounds.
	(stream_substream): Keep track of how long our substream is going
	to be for use with setting absolute bound_end on a multi-stream
	substream.

2007-02-05  Jeffrey Stedfast

	Fixes on account of the awesome test suite I wrote earlier... still
	need to fix ::substream() tho (or maybe it's just the test that's
	broke?).

	* gmime/gmime-stream-cat.c: Added an id member to struct
	_cat_node, for use with debugging...
	(stream_read): Seek in the source stream, don't call
	Cat::stream_seek() to do it.
	(stream_seek): Swapped the logic of the "within bounds" check, had
	it backwards. When seeking past a stream, set the node->position
	to the length of the stream (technically, it's like we read() thru
	all that data, right?). Instead of resetting all streams starting
	at n->next, reset them all starting at current->next... since 'n'
	technically might be before 'current'.
	(g_mime_stream_cat_add_source): Assign each node an id for easier
	debugging...

2007-02-05  Jeffrey Stedfast

	* tests/test-cat.c: The beginnings of a test suite for
	GMimeStreamCat.

	* gmime/gmime-stream-cat.c (stream_read): Go to the next stream if
	nread <= 0, not just nread == 0.

2007-02-04  Jeffrey Stedfast

	* gmime/gmime-stream-mem.c (stream_reset): No longer need to
	update stream->position.

	* gmime/gmime-stream-mmap.c (stream_reset): Same g_return_if_fail
	change. Also don't update stream->position.
	(stream_seek): Added similar sanity checking/eos resetting as
	StreamFs code.

	* gmime/gmime-stream-file.c (stream_read): Style changes.
	(stream_write): Style changes.
	(stream_close): Fixed compiler warning.
	(stream_reset): Changed g_return_if_fail to a true if-then.
	(stream_seek): Changed to mimic the StreamFs changes.
	(g_mime_stream_file_new): Similar to fs_new() change.

	* gmime/gmime-stream-fs.c (stream_close): Loop the close()
	ourselves.
	(stream_reset): Changed g_return_if_fail to a true if-then. Also
	be better about resetting eos.
	(stream_seek): Rewritten to be more correct/robust (at least I
	hope). Also properly reset eos when appropriate.
	(g_mime_stream_fs_new): If lseek() fails, pretend start offset is
	0.

	* gmime/gmime-stream.c (stream_reset): No longer needs to update
	stream->position.
	(g_mime_stream_reset): Update stream->position if everything reset
	smoothly. This is just a convenience change to subclass
	implementations.

	* gmime/gmime-stream-cat.c (stream_read): Rewritten. Hopefully
	correct now? Ugh. We can hope...
	(stream_write): Fixed to work better.
	(stream_close): Rewritten.
	(stream_reset): Rewritten to reset the streams, don't seek.
	(stream_seek): Rewritten... still not correct, but should be ok
	assuming our bound_start is 0.
	(stream_length): Rewritten to not depend on a pre-calculated
	length value... this Does Not Work (tm) if the source streams are
	unbound and we've written to them.
	(stream_substream): Rewritten... because ::seek() is so complex
	for this type of stream, I've tried to eliminate a lot of the
	headaches by making substreams only slurp up the streams within
	the bounds of the start/end requested. If the entire contents
	within the requested bounds are contained within a single source
	stream, we return a substream of said source stream instead.
	(g_mime_stream_cat_add_source): Don't precalculate the length here
	anymore.

	* gmime/gmime-stream-buffer.c (stream_read): Fixed to never allow
	buflen to be negative if the read() of our source stream fails.
	(stream_close): Handle the case where we've already been closed.
	(stream_eos): Simplified.
	(stream_reset): Stylistic changes.
	(stream_seek): Don't allow seeks under our bound_start.

2007-02-03  Jeffrey Stedfast

	* gmime/gmime-stream-cat.c (stream_flush): Flush all streams up to
	and including the current stream, not just the current stream.
	(stream_write): break if we don't write any data to
	current->stream or get an error so we can try the next stream.

	* gmime/gmime-stream-file.c (stream_close): Same as below.

	* gmime/gmime-stream-fs.c (stream_close): Allow closing multiple
	times... makes this consistant with the other streams.
	(stream_write): If a system write fails with EFBIG or ENOSPC, set
	eos to TRUE.

2007-01-20  Jeffrey Stedfast

	* tests/test-streams.c (test_stream_gets): printf formatter fixes.

	* examples/imap-example.c: #include <glib/gstdio.h>, fixes bug
	#394434.

2007-01-20  Jeffrey Stedfast

	Fixes bug #394419

	* gmime/gmime-filter-charset.c (filter_filter): Do iconv const
	casting.
	(filter_complete): Same.

2006-11-02  Jeffrey Stedfast

	* gmime/gmime-utils.c (g_mime_utils_header_format_date): Changed
	the names of the parameters to date and tz_offset to prevent
	warnings from -Wshadow.

	* gmime/gmime-filter-html.c (writeln): Fixed another possible
	buffer overflow condition pointed out by hpj.

	* gmime/gmime-filter-crlf.c (filter_filter): Fixed a possible
	buffer overflow condition pointed out by hpj.

	* gmime/gmime-gpg-context.c (gpg_ctx_op_step): Modified to use
	poll() rather than select()

2006-09-18  Joe Shaw

	* mono/Makefile.am: Use SOURCES_XML instead of XML_SOURCES to
	specify the sources.xml file.  XML_SOURCES breaks with newer
	automakes.

2006-08-29  Pawel Salek

	* gmime/gmime-message-partial.c: set buf just before use - since
	the mem stream buffer may get reallocated in the meantime leading
	to segfaults.

2006-08-29  Jeffrey Stedfast

	* gmime/gmime-param.c (decode_quoted_string): Unescape escape
	sequences. Fixes bug #352771.

	* gmime/gmime-message.c (g_mime_message_set_subject): Updated the
	docs.
	(g_mime_message_get_subject): Same.

2006-08-21  Peter Bloomfield

	* gmime/gmime-utils.c (g_mime_utils_unquote_string): handle an
	arbitrary sequence of quoted and unquoted sections.

2006-08-02  Jeffrey Stedfast

	* README: Bumped version

	* configure.in: Bumped version to 2.2.3

	* gmime/gmime-message.c (process_header): Decode the values for
	Subject, From, Reply-To so that g_mime_message_get_subject/etc all
	return decoded strings like they were meant to.

2006-06-23  Jeffrey Stedfast

	* gmime/gmime-gpg-context.c (swrite): Use g_build_filename() and
	g_get_tmp_dir() rather than hard-coding the tmpfile to be in /tmp.

2006-06-23  Jeffrey Stedfast

	* gmime/gmime-part.c: Use g_ascii_* versions of str[n]casecmp.

	* gmime/gmime-parser.c: Use g_ascii_* versions of str[n]casecmp.

	* gmime/gmime-param.c: Use g_ascii_* versions of str[n]casecmp.

	* gmime/gmime-multipart-signed.c: Use g_ascii_* versions of
	str[n]casecmp.

	* gmime/gmime-multipart-encrypted.c: Use g_ascii_* versions of
	str[n]casecmp.

	* gmime/gmime-multipart.c: Use g_ascii_* versions of
	str[n]casecmp.

	* gmime/gmime-message-partial.c: Use g_ascii_* versions of
	str[n]casecmp.

	* gmime/gmime-message-part.c: Use g_ascii_* versions of
	str[n]casecmp.

	* gmime/gmime-message.c: Use g_ascii_* versions of str[n]casecmp.

	* gmime/gmime-iconv.c (g_mime_iconv_open): Use g_ascii_* versions
	of str[n]casecmp.

	* gmime/gmime-header.c (g_mime_header_new): Use
	g_mime_strcase_[hash,equal] rather than param_[hash,equal] to
	reuse code.

	* gmime/gmime-gpg-context.c: Use g_ascii_* versions of
	str[n]casecmp.

	* gmime/gmime-filter-enriched.c: Use g_ascii_* versions of
	str[n]casecmp.

	* gmime/gmime-disposition.c (g_mime_disposition_add_parameter):
	Use g_mime_strcase_[hash,equal] rather than param_[hash,equal] to
	reuse code.
	(g_mime_disposition_new): Same.
	(param_equal): Removed.
	(param_hash): Removed.

	* gmime/gmime-content-type.c: Use g_ascii_* versions of
	str[n]casecmp.
	(g_mime_content_type_new_from_string): Use
	g_mime_strcase_[hash,equal] rather than param_[hash,equal] to
	reuse code.
	(g_mime_content_type_set_parameter): Same.
	(param_equal): Removed.
	(param_hash): Removed.

	* gmime/gmime-charset.c: Use g_ascii_* versions of str[n]casecmp.

2006-06-23  Jeffrey Stedfast

	* gmime/gmime-stream-cat.c (stream_seek): Fixed to only seek if
	necessary and also to use ::reset() if seeking back to the
	beginning of the stream so that it works for non-seekable streams
	too. Fixes bug #345503 - again thanks to Charles for reporting
	this bug.

	* gmime/gmime-object.c (subtype_bucket_foreach): Fixed to also
	free the bucket memory. Fixes bug #345768, thanks to Charles Kerr
	for pointing this out.

2006-06-22  Jeffrey Stedfast

	* gmime/gmime-gpg-context.c: Removed #define _POSIX_SOURCE to fix
	bug #172179.

2006-06-15  Jeffrey Stedfast

	* README: Bumped version

	* configure.in: Bumped version to 2.2.2

2006-06-01  Jeffrey Stedfast

	* gmime/gmime-utils.c (parse_broken_date): Fixed to handle
	MM-DD-YY properly and also to fallback to DD-MM-YY if the first
	pair of digits isn't within range for a proper month.

2006-05-30  Jeffrey Stedfast

	* gmime.h.in: Removed.

	* gmime.h: Now static, no longer auto-generated.

	* gmime/gmime.c: Set the gmime_[major,minor,micro]_version
	variables.
	(g_mime_check_version): New function to check the required
	version.

	* gmime/gmime-stream-fs.c: Implement our own no-op version of
	fsync() for systems that don't have fsync().

2006-05-24  Jeffrey Stedfast

	* examples/imap-example.c (main): Use g_mkdir() to make this
	example portable to Win32.

2006-05-01  Jeffrey Stedfast

	* gmime/gmime-gpg-context.c (gpg_ctx_parse_status): Handle prompts
	from GnuPG asking for the user's PIN for their SmartCard.

2006-04-21  Jeffrey Stedfast

	* gmime/gmime-stream.h: #include <stdio.h> for SEEK_SET/CUR/END
	enum values.

2006-03-16  Jeffrey Stedfast

	* README: Bumped version

	* configure.in: Bumped version to 2.2.1

	* gmime/gmime-parser.c (parser_step): No-op if we are in an ERROR
	state.
	(parser_construct_message): Don't bail if we don't start in state
	INIT, this is valid.
	(parser_step_from): We need to bail even if parser_fill() returns
	> 0 if the line is incomplete (e.g. a forced refill was needed but
	no additional data could be read).

2006-03-16  Jeffrey Stedfast

	* README: Bumped version

	* configure.in: Bumped version to 2.2.0, it's about time I made a
	stable release.

2006-03-15  Jeffrey Stedfast

	* gmime/gmime-parser.c (parser_step_from): If we fail to find a
	From line, set an error state. Also fixed the End-Of-Data check to
	prevent an infinite loop if the stream ends with \n\n.
	(parser_construct_message): Make sure our state starts off at
	INIT. Also check for the ERROR state in our parser_step() loop and
	return NULL if we encounter an error.

2006-03-03  Jeffrey Stedfast

	* util/url-scanner.c (g_url_web_end): Allow urls such as
	"http://www.novell.com./path" (note the '.' before the path
	component).
	(g_url_addrspec_start): If the '@' is the first character in the
	match, then it isn't a valid email address.

Log message:
Change the format of BUILDLINK_ORDER to contain depth information as well,
and add a new helper target and script, "show-buildlink3", that outputs
a listing of the buildlink3.mk files included as well as the depth at
which they are included.

For example, "make show-buildlink3" in fonts/Xft2 displays:

	zlib
	fontconfig
	    iconv
	    zlib
	    freetype2
	    expat
	freetype2
	Xrender
	    renderproto

Log message:
Track information in a new variable BUILDLINK_ORDER that informs us
of the order in which buildlink3.mk files are (recursively) included
by a package Makefile.

Log message:
Update to gmime 2.1.19 and drop maintainership.
Based upon the 2.1.17 update by Fredrik Carlsson in PR 32487

Changes:
The usual: fixes, new features.

Log message:
Aligned the last line of the buildlink3.mk files with the first line, so
that they look nicer.

Log message:
Over 1200 files touched but no revisions bumped :)

RECOMMENDED is removed. It becomes ABI_DEPENDS.

BUILDLINK_RECOMMENDED.foo becomes BUILDLINK_ABI_DEPENDS.foo.

BUILDLINK_DEPENDS.foo becomes BUILDLINK_API_DEPENDS.foo.

BUILDLINK_DEPENDS does not change.

IGNORE_RECOMMENDED (which defaulted to "no") becomes USE_ABI_DEPENDS
which defaults to "yes".

Added to obsolete.mk checking for IGNORE_RECOMMENDED.

I did not manually go through and fix any aesthetic tab/spacing issues.

I have tested the above patch on DragonFly building and packaging
subversion and pkglint and their many dependencies.

I have also tested USE_ABI_DEPENDS=no on my NetBSD workstation (where I
have used IGNORE_RECOMMENDED for a long time). I have been an active user
of IGNORE_RECOMMENDED since it was available.

As suggested, I removed the documentation sentences suggesting bumping for
"security" issues.

As discussed on tech-pkg.

I will commit to revbump, pkglint, pkg_install, createbuildlink separately.

Note that if you use wip, it will fail!  I will commit to pkgsrc-wip
later (within day).