NEWS for ttn-pers-scheme


- 0.50 | 2008-03-16

  - License now GPLv3+ (see COPYING)

  - Dropped (ttn html-world) procs: render-html-world, html-world

	These were announced in release 0.48 (2007-04-07).

  - Dropped module: (ttn METAINFO)

	This was introduced in release 0.33 (2004-01-30), but
	proved to be useless.  Sorry, no pre-announcement (see
	"Imminent retirement" below).

  - Bugfix: (ttn listener) make-listener regression

	A bug was introduced in ttn-pers-scheme 0.18 (2001-04-09)
	whereby the variable holding the current connection socket was
	accessed out of scope.  Caught by "guile-tools lint".

  - New configure check: (ice-9 optargs-kw) proc `remove-keys'

	This proc originated from module (ttn optargs-kw-utils).

  - Benchmarks back

	It's been several years since benchmarks were run, partly due to
	slow hardware (200MHz Pentium Pro, 300MHz Openbrick) and twitchy
	release fingers.  Now the infrastructure has been spruced up and
	the files benchmarks/tea-{0.1,2.0} are again being distributed.
	Default iterations bumped 10x for 10 years of progress... :-/

	Note: These files will probably not be updated on every release.

  - Imminent retirement

	As of this release, ttn-pers-scheme is heading towards
	retirement.  This means useful functionality has been migrated
	elsewhere and that after 2008-12-31, the package will no longer
	be available.  See "Future" in the documentation for details.


- 0.49 | 2007-04-09

  - Changes to module (ttn markup)

    - New proc: symbol<-kw/sym
    - New proc: string<-kw/sym
    - New proc: css-tree

	Might as well put these common operations here.

    - New parameter to `expand': #:xbsc

	Aka "XMLish (blech) start close".  Really: "blech!"

  - New (ttn flatten) proc: flatten-to

  - New module: (ttn html-data2)

	For XHTML 1.0 generation.


- 0.48 | 2007-04-07

  - After DATE, THING will be removed; DO NOT rely on it

    - 2007-10-15: proc `(ttn html-world) render-html-world'
    - 2008-04-15: proc `(ttn html-world) html-world'

	You can use `html-world.html-data' (cleaner, more general).

  - Dropped module: (ttn echo)
  - Dropped module: (ttn stringutils)

	These were announced in release 0.44 (2005-09-04).

  - Dropped (ttn abbrev-tree) proc: scrub
  - Dropped (ttn abbrev-tree) proc: split-first-word
  - Dropped (ttn abbrev-tree) proc: list->sep-string

	These were announced (in the manual but not in NEWS -- oops!)
	in release 0.43 (2005-08-20).

  - Dropped module: (ttn moduleutils)
  - Dropped feature: quoting support in `read-text-db-table'

	These were announced in release 0.42 (2005-06-05).

  - Changes to module (ttn html-world)

    - External program no longer needed: mogrify(1)

	This is part of the package `ImageMagick' and was used to invert
	a ppm image .  Now, this is done with another external program,
	pnminvert(1), part of the package `Netpbm'.  See modules.af for
	a full list.

    - New #:gen method: #:ppmforge

	This uses ppmforge(1) with options `-quiet -stars 0' to create
	the ppm image.  A large majority of the time, the image does not
	correspond to any landmass found on Earth, but sometimes a vague
	similarity occurs.  The #:ppmforge method is now the default for
	`html-world' and `html-world.html-data'.

    - New keyword arg for `html-world.html-data': #:link-tree

	Normally, the proc `href' from module (ttn html-data) is used to
	construct the tree (nested list of strings) that represents a
	chunk of text that is to be marked up w/ an HTML <A HREF...></A>
	tag.  This option lets you use another proc.

    - New proc: string<-image-command
    - New proc: randomly-hyperlinked

	These are the underlying mechanisms in `html-world.html-data',
	now generalized and documented.  Let the madness begin...

  - Optimization: `flatten' does nothing for the null string

  - New (ttn html-data) procs: link tt, pre, div, span

  - New keyword for `listener': #:catch-tag SYMBOL

  - New (ttn dirutils) macro: with-cwd

  - New (ttn shell-command-to-string) proc: file-lines

  - New (ttn sanity) proc: ppsrc

  - New module: (ttn markup)

	Procedures in this module support definition of markup languages
	and related tree (nested list of strings) construction.  Flatten
	the trees to get the grotty <ELEM ATTR=VALUE>BODY</ELEM> stuff.

  - New homepage: http://www.gnuvola.org/software/ttn-pers-scheme/

	Please update your bookmarks.  Old glug.org urls can be
	munged to get the new url by passing them through:

	sed 's,/people/ttn/,/,;s,/ps/,/ttn-pers-scheme/,'

	Note: old urls in NEWS will remain unchanged, intentionally.


- 0.47 | 2006-07-22

  - Dropped (ttn rcsutils) proc: group-files-by-rcs-state
  - Dropped (ttn cvsutils) proc: group-files-by-cvs-state

	This was announced in release 0.46 (2006-02-23).  You can
	replace these procs, both of which take list FILES, with:
	  (group-logs-by 'state (map rlog FILES))
	and
	  (group-logs-by 'state (map cvslog FILES))
	respectively.

  - Dropped module: (ttn split-mbox)
  - Dropped module: (ttn flatten-to-buffer)
  - Dropped module: (ttn find-file)
  - Dropped module: (ttn write-buffer)
  - Dropped module: (ttn write-buffer-if-changed)
  - Dropped module: (ttn save-buffer)

	This was announced in release 0.46 (2006-02-23).  Except for the
	first one, functionality has moved to (ttn bufutils).

  - Dropped module: (ttn slice)
  - Dropped module: (ttn edit)

	This was announced in release 0.42 (2005-06-05).  You can get
	(ttn edit) functionality from module (ice-9 editing-buffer).

  - Dropped module: (ttn defvar)

	This was announced in release 0.38 (2004-12-03).

  - Bugfix: (ttn rw-vcg) procs can handle `classname' and `infoname'

	These attributes are now transformed so that the integer between
	the attribute name and the colon is made into the first element
	of the attribute value (a list).  For example:

	vcg file text:  infoname 2 : "good stuff"
	internal rep:   #:infoname (2 "good stuff")

	Previously, parsing would simply fail (with error) on such text.
	Also, reading can now handle a space between the attribute name
	and the colon.  The module still needs better testing (as well
	as a proper implementation that doesn't rely on sed(1)!).


- 0.46 | 2006-02-23

  - No longer maintained, to be dropped after 2006-05-31

    - module (ttn split-mbox)
    - module (ttn flatten-to-buffer)       ;; see (ttn bufutils)
    - module (ttn find-file)               ;; likewise
    - module (ttn write-buffer)            ;; likewise
    - module (ttn write-buffer-if-changed) ;; likewise
    - module (ttn save-buffer)             ;; likewise
    - `group-files-by-rcs-state' from module (ttn rcsutils)
    - `group-files-by-cvs-state' from module (ttn cvsutils)

  - More slackful (ttn html-data) procs

	The following procs now take any number of args, not just
	a single arg: h1, h2, h3, h4, h5, h6, b, li, table, tr, td.
	This means you can write, e.g., `(table (tr (td ...) ...))'
	instead of `(table (list (tr (list (td ...) ...)) ...))'.

  - New (ttn rlog-scan-proc) proc: group-logs-by

  - New module: (ttn bufutils)

  - Info page index unified (from three to one)


- 0.45 | 2006-01-30

  - Dropped module: (ttn collect-files-sorted-by-mtime)
  - Dropped module: (ttn display-table)

	These were declared "no longer maintained" in release 0.44
	(2005-09-04) and 0.38 (2004-12-03), respectively.

  - Changes to (ttn cvsutils)

    - Bugfix: Export `group-files-by-cvs-state'

    - Update parsing to handle date format from recentish "cvs log"

	With CVS 1.12.9, the dates output from a "cvs log" command
	differ slightly from the classic RCS "YYYY/MM/DD HH:MM:SS"
	format.  Parsing now handles the new date format, i.e.,
	"YYYY-MM-DD HH:MM:SS ZZZZZ", as well as the old.  The timezone
	portion ZZZZZ is discarded.


- 0.44 | 2005-09-04

  - Dropped module: (ttn pgtype) -- see NEWS 0.37 (2004-08-05) entry
  - Dropped module: (ttn pgtable) -- likewise
  - Dropped module: (ttn pgmeta) -- likewise

  - Module no longer maintained: (ttn collect-files-sorted-by-mtime)
  - Module no longer maintained: (ttn echo)
  - Module no longer maintained: (ttn stringutils)

  - New (ttn html-data) proc: smhdwy-since

  - Extension to proc `read-text-db-table': flags

	Flag #:list means return each record as a list instead of as an
	alist (the default).  Flag #:closure means return each record as
	a closure (procedure) that accepts one arg SEL.  If #:list is
	also specified, SEL is a 0-based integer to index into the
	record's data.  Otherwise, SEL names a field in the record's data.

  - New (ttn email-log-if-fail) proc: set-system-mail-program!


- 0.43 | 2005-08-20

  - Dropped module: (ttn ftw) -- see NEWS 0.30 (2002-08-13) entry
  - Dropped module: (ttn eformat) -- 0.38 (2004-12-03)
  - Dropped module: (ttn gap-buffer) -- 0.37 (2004-08-05)

  - Configuration change

	The "configure" script no longer checks that module (guile-user)
	exports proc `eval-in-module'.  This proc is now accessed via
	loading module (ice-9 gumm).

  - Extension to proc `filtered-files-in-vicinity'

	This proc now takes options, in the form of additional keyword
	flags, that modify the default behavior:

	#:filter-prefixed -- Pass to the filter proc each filename with
			     the directory already prefixed.
	#:collect-nodir -- When the filter proc returns #t, collect the
			   filename without the directory prefix.

	These options encapsulate common post-processing operations and
	can be used singly or together (or omitted entirely).

  - Extension to proc `read-text-db-table'
  - New (ttn read-text-db-table) proc: text-db-table-meta

	The text db table format now can include meta info:

	(text-db-table-config
	  (meta . META)     ; optional; META is opaque, default #f
	  (delim . "\f")
	  (fields (NAME1 TYPE1)
	          (NAME2 TYPE2)
	          ...))

	The new object property `text-db-table-meta' (procedure with
	setter) can be used to access/modify the META associated with
	the list of records returned by `read-text-db-table'.

  - Extensions to proc `make-listener'

	First arg can now be a list of `bind' arguments, instead of an
	integer.  For backward compatability, an integer is taken to
	specify a TCP/IP port, in which case the family is `AF_INET' and
	the address is `INADDR_ANY'.  For example, these are identical:

	(make-listener 23456)
	===
	(make-listener (list AF_INET INADDR_ANY 23456))

	Here is an example of specifying a socket on the filesystem:

	(make-listener (list AF_UNIX "/tmp/sockfile"))

	Additionally, `make-listener' now accepts keyword option
	#:concurrency with value #:coop-thread, which means dispatch
	each child in its own thread; or #f (the default), which means
	dispatch and wait for each child to finish serially.

  - Extension to `grep' and `grep-matches'

	The flags that modify the default behavior for these procs may
	now be keywords.  (Symbols are ok, too, as before.)

  - New (ttn dirutils) proc: file-newer-than-file?


- 0.42 | 2005-06-05

  - Module no longer maintained: (ttn moduleutils)
  - Module no longer maintained: (ttn slice)
  - Module no longer maintained: (ttn edit)

  - Quoting no longer required for text db table specifcation

	Formerly, a text db table spec required quotes, like so:

	  (text-db-table-config
	    '(delim . "\f")
	    '(fields (NAME1 TYPE1)
	             (NAME2 TYPE2)
	             ...))

	Now, the quotes are no longer required:

	  (text-db-table-config
	    (delim . "\f")
	    (fields (NAME1 TYPE1)
	            (NAME2 TYPE2)
	            ...))

	For backward compatability, quotes are still accepted for the
	time being; that support will be dropped at some point.

  - New (ttn cron) macro: cron!


- 0.41 | 2005-04-03

  - Changes to (ttn sgf)

    - Dropped proc: display-in-emacs-syntax

	This was a provisionary proc useful in the period before SGF
	read/write support appeared in gnugo.el 2.2.10:

	http://www.glug.org/people/ttn/software/personal-elisp/standalone/

    - Property `GM' range extended to 20

	The `GM' property specifies the type of game tree(s) encoded in
	an SGF file.  The SGF[4] specification says:

	Valid numbers are: Go = 1, Othello = 2, chess = 3,
	Gomoku+Renju = 4, Nine Men's Morris = 5, Backgammon = 6,
	Chinese chess = 7, Shogi = 8, Lines of Action = 9,
	Ataxx = 10, Hex = 11, Jungle = 12, Neutron = 13,
	Philosopher's Football = 14, Quadrature = 15, Trax = 16,
	Tantrix = 17, Amazons = 18, Octi = 19, Gess = 20.

  - More chars trasliterated in (ttn html-world) procs' output

	Previously, only "<" was changed to "L".  Now, "&" is changed to
	"8" and ">" is changed to "7", as well.

  - Change to (ttn html-data) proc: `p'

	This proc now takes an optional arg, any value of which causes
	begin and end ("<P>" and "</P>", respectively) style rendering
	around the arg.

  - New field type for (ttn read-text-db-table) proc: `read-text-db-table'

	A field may now be declared as having type `rest-lines-trim',
	which arranges for the reading procedure to extract the rest of
	the lines of the record before the delimiter (exactly like field
	type `rest-lines'), and additionally to apply `string-trim-both'
	(from SRFI-13) to the result before collecting it.

  - New module: (ttn rw-vcg)

	VCG is a program to layout and visualize graphs in 2D.  This
	module allows you to read and write .vcg files (which specify
	"nodes" and "edges" and their attributes as well as global
	attributes) produced by or for VCG, from a Scheme program.


- 0.40 | 2005-01-04

  - Bugfix to (ttn spewutils) proc `eval-html-data-file'

	ttn-pers-scheme 0.39 introduced a bug whereby this proc
	evaluated the data file in the wrong environment.

  - New procs for (ttn html-data)

	Specifically:
	q, em, strong, dfn, code, samp, kbd, var, cite, abbr.


- 0.39 | 2005-01-03

  - New (ttn grep) proc: grep-l

  - New (ttn expand-file-name) proc: substitute-env-vars
  - New (ttn expand-file-name) proc: expand-file-name-substituting-env-vars

  - Experimental read-hash extension

	Proc `give-me-a-sane-environment-please!' (for interactive
	sessions) now extends the read-hash table to handle `#^Ffoo'
	as essentially `(expand-file-name-substituting-env-vars "foo")',
	a string.  Here is some elisp that uses this feature:

	(local-set-key "\C-cf"
	  (lambda (file) (interactive "fFile: ")
	    (insert "#\C-f" file)))

  - New (ttn sanity) proc: pp

  - New (ttn html-data) proc: blockquote

  - Module (ttn edit) now simply re-exports (ice-9 editing-buffer)

	The functionality and the exported macro `editing-buffer' of
	module (ttn edit) has been moved to Guile 1.4.x.  At some point
	(ttn edit) will no longer be maintained, although the interface
	is still provided for the time being.

	Along w/ this change, module (ttn gpgutils) uses module (ice-9
	editing-buffer) directly now, instead of (ttn edit).


- 0.38 | 2004-12-03

  - Bugfix to (ttn edit) expansion of `case'

	The expander for `editing-buffer' used to transform `case'
	expressions incorrectly: adding an extra set of parens for
	each element in the choices list.

  - Module no longer maintained: (ttn display-table)

	This functionality has migrated to Guile-PG (see link below).

  - Module no longer maintained: (ttn eformat)
  - Module no longer maintained: (ttn defvar)

	These will be dropped in some future release.

  - New module: (ttn sgf)

  - Extensions to (ttn html-data) proc `href'

	This now takes an optional third arg `name'.
	For example: (flatten (href "x" "y" "z"))
	          =| <A NAME="z" HREF="x">y</A>

	Additionally, `url' can be #f, in which case it is omitted.
	For example: (flatten (href #f "y" "z"))
	          =| <A NAME="z">y</A>

  - Extension to `collect-files-sorted-by-mtime'

	This proc now takes keyword args: #:key PROC, applying PROC to
	each element in `files' to get the actual filename to stat.

  - Info entry now under category "Guile modules"

	It used to be under "Software Libraries".


- 0.37 | 2004-08-05

  - Module no longer maintained: (ttn pgtype)
  - Module no longer maintained: (ttn pgtable)
  - Module no longer maintained: (ttn pgmeta)

	The functionality provided by these modules has migrated to
	Guile-PG, with homepage:

	  http://www.glug.org/people/ttn/software/guile-pg/

	Thus, although these modules are still provided here for the
	time being, they are no longer being maintained.  This means,
	among other things, that the configure script no longer
	recognizes option "--enable-pgtable-check-db=NAME".

  - Extension to (ttn edit) "e:" proc `insert'

	If OBJECT is a gap buffer, its contents are inserted.
	Point moves forward as usual.

  - Bugfixes to (ttn edit) buffer traversal "e:" procs

	New tests for this module revealed some discrepencies w/ Emacs
	for return values and behavior of `forward-line', and behavior
	of `beginning-of-line' and `end-of-line'.  In particular (i.e.,
	as expected ;-), edge and corner cases for buffers w/ and w/o
	newlines proved most useful for exposing bugs.  See edit.test in
	the tests/ subdirectory for details (more tests welcome).

  - New (ttn dirutils) proc: mkdir-p

  - New (ttn fileutils) proc: filename-absolute-or-in-vicinity

  - New (ttn expand-file-name) proc: reset-tilde-cache!

  - Changes to (ttn html-world) procs

	The procs `html-world' and `html-world.html-data'
	now also accept keywords for the #:gen value.

  - Start of switch to (ice-9 gap-buffer)

	In preparation for migration of (ttn gap-buffer) functionality
	to module (ice-9 gap-buffer), modules that used the former have
	been changed to use the latter.  The complete list is:

	(ttn call-process-to-buffers)
	(ttn edit)
	(ttn find-file)
	(ttn flatten-to-buffer)
	(ttn gpgutils)
	(ttn make-buffered-caller)
	(ttn parse-rfc822)
	(ttn save-buffer)
	(ttn write-buffer-if-changed)
	(ttn write-buffer)

	At some point (ttn gap-buffer) will no longer be maintained,
	although it is still provided here for the time being.


- 0.36 | 2004-07-02

  - Change to (ttn personal-pgtable) proc: `all-defs'

	This proc now understands the two-form format of the meta file,
	where the first form is a list of (KEYWORD EXPANSION) specs, and
	the second form is the database table-spec alist, as before.

	For example, before the change -- one form:

	(("mydb" ("table1" (a text "NOT NULL" "UNIQUE")
	                   (b text "NOT NULL" "UNIQUE"))
	         ("t2" (a text "NOT NULL" "REFERENCES table1 (a)"
	                       "MATCH FULL" "ON UPDATE CASCADE")
	               (a text "NOT NULL" "REFERENCES table1 (b)"
	                       "MATCH FULL" "ON UPDATE CASCADE"))))

	After the change -- two forms:

	((#:nn "NOT NULL")
	 (#:u  "UNIQUE")
	 (#:r  "REFERENCES")
	 (#:mf "MATCH FULL")
	 (#:uc "ON UPDATE CASCADE"))

	(("mydb" ("table1" (a text #:nn #:u)
	                   (b text #:nn #:u))
	         ("t2" (a text #:nn #:r "table1 (a)" #:mf #:uc)
	               (b text #:nn #:r "table1 (b)" #:mf #:uc))))

	The table specs after the change are easier to read and less
	prone to typing errors, especially in an environment that
	fontifies Scheme keywords (like Emacs), than before the change.


- 0.35 | 2004-06-21

  - New (ttn edit) "e:" procs

	These procs are now available:

	  (forward-line &optional n)
	  (beginning-of-line &optional n)
	  (end-of-line &optional n)
	  (bolp)
	  (eolp)

	They are valid within the `editing-buffer' macro.  Please
	refer to Emacs docstrings for functions by the same name,
	for more info.

  - Performance improvements in (ttn edit) search/replace

	Some unnecessary string consing has been eliminated in the
	search/replace operations; they are now a bit quicker.


- 0.34 | 2004-06-10

  - Changes to functionality

    - Bugfix to (ttn filesystem-tree-to-list)

	This proc used to improperly nest empty sibling subdirectories
	due to mis-{understanding,handling} of level information.  The
	DFS collection algorithm has been rewritten to use base (prefix)
	comparisons.

    - `read-text-db-table' handles CRLF for eol in addition to LF

    - `email-log-if-fail' changes

	This now supports multiple semicolon-separated command lists
	(not only a single list).  Also, the return value is now
	specified: #t means the command was successful, #f means not
	(implies mail was sent out).

    - `call-process-to-buffers' no longer requires external program touch(1)

    - `slackful-concat-proc' now accepts a keyword as well as a symbol

  - New procs

    - In (ttn personal-pgtable): `personal-pgtable-worker'

    - In (ttn dirutils): `extract-stem-proc'

    - In (ttn html-data): `br' and `center'

  - Dropped module: (ttn psql)

  - Dropped scripts

    - display-rcs-state-summary

    - generate-autoload


- 0.33 | 2004-01-30

  - Info page installed

	At long last, there is documentation.  Also distributed, but
	not installed, is the same documentation in HTML format.  The
	availability of docs somewhat explains the brevity of the New
	Foo items below.

  - Installation updates module catalog (if supported by system)

  - Bugfix to (ttn edit) proc: `looking-at'

	A bug was fixed whereby regular expressions beginning with
	"^" were not handled correctly: `looking-at' would return #f
	when point was in fact at the beginning of the line.  It would
	only return #t if point was at the beginning of the buffer.

  - Input pipe explicitly closed

	The following procs now explicitly do a `close-pipe' when done
	reading input, allowing memory to be collected:

	  (ttn shell-command-to-string)  shell-command-to-string
					 shell-command-to-list
					 shell-command->string
					 shell-command->list

  - Internal (ttn gap-buffer) values bumped up

	(define default-initial-allocation (* 2 1024))
	(define default-chunk-size 1024)
	(define default-realloc-threshold 32)

  - New (ttn dirutils) proc: `filtered-files-in-vicinity'

  - New module: (ttn METAINFO)

  - New module: (ttn cvsutils)

	This is analogous to (ttn rcsutils), for CVS.

  - New module: (ttn read-text-db-table)

  - New module: (ttn personal-pgtable)

  - New module: (ttn rlog-scan-proc)

  - New scripts

	These are not installed but are mentioned here since they
	play a part in the methodology.

    - steal-this-package

	Actually, running this program results in an error at this
	time.  The adventurous may wish to pursue the matter anyway.

    - collect-lessons

	This is used to help maintain the "Learning Scheme Programming"
	node in the Guide.  The mechanisms are in place but the actual
	recording of lessons is still very much in the future.

  - Makefile oriented to CVS

	Source has moved from RCS to CVS and the top-level Makefile now
	reflects that (notably for "make dist").


- 0.32 | 2003-08-05

  - "make check" improvements

	The test for (ttn gpgutils) does a "gpg --verify" and compares
	expected values against obtained results.  The time component
	of the test has been localized to work outside of the PST (UTC
	offset "+0800" or 28800 seconds) time zone.

  - New module: (ttn optargs-kw-utils)

	At the moment, this just exports `remove-keys', which is used
	to remove keyword/value pairs from an "argument rest" list,
	since those are not automatically removed by `define*' et al.

  - Changes to (ttn call-process)

	The `remove-keys' proc is no longer available in this module.
	See new module (ttn optargs-kw-utils) above, instead.

  - Changes to (ttn html-world)

    - proc `html-world' now takes keywords #:size and #:gen
    - proc `html-world.html-data' likewise

	The original image as generated by xearth(1) has a default size
	of 150x150 pixels (which after ASCIIfication fits comfortably in
	an 80-column screen).  You can now pass #:size N to `html-world'
	to specify an NxN-pixel originating image.

	Additionally, the default backend program has been changed from
	xearth(1) to xplanet(1), the former having been stigmatized as
	non-free software, at least by Debian.  As it turns out, by the
	time you read this, the LZW patent will have expired, so now
	it's ok to use xearth after all!  In any case, you can pass
	#:gen BACKEND, where BACKEND is a symbol `xplanet' or `xearth',
	to `html-world' to select the preferred image generator.

    - procs `refresh' and `textarea' now handle numerical args

	These now detect if the `sec' and `row'/`col' args,
	respectively, are numbers and do a `number->string' for you.
	Strings continue to be supported.

    - bugfix for proc `input'

	Elements are now properly interspersed with spaces.



- 0.31 | 2002-09-27

  - Bugfix to (ttn spewutils)

	Use `eval-in-module' to be explicit.

  - New configure option

	Previously, enabling (ttn pgtable) testing was a hairy process.
	Now this is handled by passing "--enable-pgtable-check-db=NAME"
	to configure.  See README.

  - Distribution changes

    - Directory ttn/testing/benchmarks/ is now at top-level

	Benchmarking is back; i.e., you can now do "make tea" again.
	See Makefile for BENCHMARK_ITERATION_MULTIPLIER usage.

    - Directory ttn/testing/ttn/ is now renamed to tests/ at top-level

	Additionally, the tests are broken out into individual files, and
	the Makefile now uses "guile-tools run-all-tests" to run them.
	You can choose specific tests to run like so:

	  make check ALLTESTS="ftw.test gap-buffer.test" DEBUG=1

	Omit DEBUG=1 for a more silent running.  The log is written to
	tests/CHECK.LOG.  You can do "make check" in either top-level or
	directly in the tests/ subdirectory.


- 0.30 | 2002-08-13

  - Bugfix to (ttn html-data)

	Proc `copyright' no longer hardcodes "2001".  Instead, the value
	is taken from the current year.

  - Bugfix to (ttn expand-file-name)

	The new algorithm has been implemented that now handles "../.."
	in the presence of a "~USER" and other funky cases properly.

  - Deprecated module: (ttn ftw)

	The `ftw' and `nftw' procs have now been folded into Guile in
	its (ice-9 ftw) module.  (ttn ftw) just loads (ice-9 ftw) for
	now but it will probably go away in the future.

  - New (ttn gap-buffer) proc: gb->port!

	This proc sends GB contents to PORT efficiently.

  - `save-buffer' now verifies operation

	This proc, in (ttn save-buffer), now checks the written file to
	see that it has the same size as the buffer contents, returning
	#t if so.


- 0.29 | 2002-01-12

  - Functionality news

    - Change to (ttn html-data)

	Proc `html' now includes the "!DOCTYPE" jazz, woo hoo!

    - New module: (ttn split-mbox)

	Exported proc `split-mobx' doc:
	Read lines from PORT, splitting at "^From " (note space), and
	returning a list of messages, each a list of lines.

    - Change to (ttn parse-rfc822)

	Proc `parse-rfc822' now also accepts a list of lines as arg
	`mailmsg', in addition to gap-buffer or string.  The returned
	closure now accepts key `body-lines', which returns the lines
	comprising the body.  The intent is to allow easy line-oriented
	processing.

    - Change to (ttn pgtable)

	New proc `pg-result-symbol' can be used instead of
	`pg-result-status' from module (database postgres), to return a
	symbol (e.g., `PGRES_COMMAND_OK').  This change allows (ttn
	pgtable) to be used w/ any installed version of guile-pg 0.07,
	instead of only those hacked by ttn (e.g., 0.07ttn4) to do the
	symbolic translation in C.  See README for updated dependency
	info.

  - Administrative news

    - New files: modules.af, autogen.sh

	Configuration now makes use of some Guile facilities for
	automatically maintaining module dependency info.  The file
	modules.af is processed by "guile-tools autofrisk" to create
	modules.af.m4, which defines AUTOFRISK_CHECKS now referenced by
	./configure.in.

	Following widespread packaging convention, there is now the
	top-level script autogen.sh, which invokes various tools to set
	things up.  This script accepts option "--verbose".

	Lastly, (ttn gpgutils) and (ttn pgtable) testing configuration
	is now handled in part by ./configure, although there is still
	some manual component.  This will continue to lessen in the
	future.  See README for details.

    - Testing output less verbose

	Output of "make check" is now more of summary (by module).
	You can do "make DEBUG=1 check" to see details.

    - "make dist" no longer requires full tag

	The full tag is computed in the same way used in "make
	newtag=FRAG tag".  Thus, you can use FRAG everywhere.  The
	varname is changed as a reminder: "make dtag=FRAG dist" (instead
	of "disttag=").

    - No more autoload file

	This means the benchmarks are not available this release.


- 0.28 | 2001-12-17

  - (ttn ftw) benchmark now uses "/etc" instead of "/"

	/ can be humongous, /etc is big enough anyway.
	(Actually, /etc isn't very deep, maybe /usr is better.)

  - Changes to (ttn html-data)

    - New procs: `name', `dl', `dll'

	These are <A NAME="x"></A> and <DL>...</DL> (last one takes a
	single list, similar to `ul' and `ull').

    - `copyright-since' arg can now be either string or number

  - New procs in (ttn spewutils)

	(update-all-pages! source-page pages)
	(update-all-html-data-pages! pages)

	SOURCE-PAGE is proc that is mapped over PAGES, a list of
	filenames.  The latter proc calls the former w/ the standard
	filename transformer: FOO.html -> FOO.html-data.

  - Timestamp type conversion in (ttn pgtype) more restricted

	When converting the PostgreSQL "timestamp" type to Scheme, the
	value is now an integer, the number of second since epoch.  This
	loses for times before the epoch, so this restriction may be
	(re-)lifted in the future.

  - New module: (ttn display-table)

	This module exports `display-table', which has been moved out of
	(ttn pgtable), although it is still available there for now.
	`display-table' now also takes optional arg STYLE, which is a
	symbol, one of:
	   space		fat-space
	   h-only		fat-h-only
	   v-only
	   +-only
	   no-h
	   no-v			fat-no-v
	   no-+
	If omitted, the default is to use "-", "|" and "+" for the
	horizontal, vertical and intersection decorations, respectively.

  - `pgtable-manager' in (ttn pgtable) now supports alists

	The table manager now also supports `table->alists' and
	`tuples-result->alists'.  In both cases, the return value is a
	list of alists, with each alist corresponding to a tuple in the
	table (keys are column names).

  - Benchmarking results updated

	The release process now (again) updates files tea-1.0 and
	tea-0.2 -- dev machine specs follow:
	  Linux 2.2.19pre17 (gcc version 2.7.2.3) i686 unknown
	  Detected 199434 kHz processor.
	  Calibrating delay loop... 398.13 BogoMIPS
	  Memory: 29784k/32768k available
	  (1744k kernel code, 412k reserved, 672k data, 156k init)
	  Dentry hash table entries: 4096 (order 3, 32k)
	  Buffer cache hash table entries: 32768 (order 5, 128k)
	  Page cache hash table entries: 8192 (order 3, 32k)


- 0.27 | 2001-11-28

  - New module: (ttn flatten)

	This module exports the proc `flatten':
	  (flatten tree . out)

	Walk TREE depth-first, displaying elements if they are strings.
	Signal "bad type" error for non-string, non-list elements.
	Optional 2nd arg OUT is a procedure to use instead of `display'.

  - New module: (ttn write-buffer-if-changed)

	This module exports the proc `write-buffer-if-changed':
	  (write-buffer-if-changed buffer filename)

	Write BUFFER into file FILENAME but only if the contents are
	different.  FILENAME is passed through `expand-file-name', q.v.
	FILENAME cannot be a directory.

  - New module: (ttn collect-files-sorted-by-mtime)

	This module exports the proc `collect-files-sorted-by-mtime':
	  (collect-files-sorted-by-mtime files collect)

	Sort FILES by modification time, and map COLLECT over the
	result.  FILES is a list of filenames.  COLLECT is a proc that
	accepts two args, the filename and its modification time.

  - New module: (ttn flatten-to-buffer)

	This module exports the proc `flatten-to-buffer' and alias
	`flatten->buffer':
	  (flatten-to-buffer tree . gb)
	  (flatten->buffer   tree . gb)

	Call `flatten' on TREE, saving output to new gap-buffer, and
	return it.  Optional arg GB specifies a gap-buffer to use
	instead of newly allocating.

  - New module: (ttn spewutils)

	This module exports the procs:
	   (eval-html-data-file file)
	   (spew-html! html-data outfile . log)
	   (update-page! template-file outfile . log)

	This module's interface is likely to change and should not be
	considered stable just yet.

  - Changes to (ttn dirutils) `save-cwd'

	A bug was fixed in this macro where the body's return
	value was being discarded.

  - Changes to (ttn html-world)

    - Size reset to ttn-pers-scheme-0.09 value

	The initial image used as a basis for the output now has size
	150 by 150 pixels (it used to be 200 by 200).  This is basically
	because the author is now using an 80-column web browser again.

    - More external dependencies

	The programs "ppmtopgm" and "pgmtopbm" now also need to be in
	the PATH for `html-world' to work correctly.  Older versions of
	pbmtoascii used to be able to take ppm files directly, but that
	apparently is no longer the case.  The README has been updated.

    - New proc: html-world.html-data

	This proc is the preferred interface; `html-world' and
	`render-html-world' are deprecated.  Docstring follows:

	Call `html-world' w/ arg LINKS, and reformat the output as
	html-data tree.  See `html-world' for more info on LINKS.

  - Changes to (ttn html-data)

    - New proc: tt-pre

	This wraps its args w/ "<TT><PRE>" and "</PRE></TT>".

    - Bug fix: `html' no longer includes "Content-type: text/html\n\n"


- 0.26 | 2001-11-14

  - "New" modules

	The following modules were created by pulling out one or two
	procs from either (ttn fileutils) or (ttn shellutils).  This was
	done to reduce memory overhead in the case of unused definitions
	in the same module as the definition you are actually interested
	in.

	(ttn email-log-if-fail)			from (ttn fileutils)
	(ttn expand-file-name)
	(ttn save-buffer)
	(ttn write-buffer)
	(ttn find-file)

	(ttn call-process)			from (ttn shellutils)
	(ttn call-process-to-buffers)
	(ttn make-buffered-caller)

	In the future, other (ttn *utils) modules will probably be
	similarly factored.

  - Other changes to (ttn fileutils)

	Three new procs are exported: `filename-sans-end-sep-proc',
	`filename-components' and `filename-components-append'.  The
	latter two are inverses.  Here are the docstrings:

    - filename-sans-end-slash NAME

	Return NAME, a string, stripping the terminating "/" character.
	If there is no "/", just return NAME.

    - filename-components STRING

	Return a list of filename components parsed from STRING.
	Components are delimited by "/", which is discarded.

    - filename-components-append LS

	Return a string composed by interspersing LS elements w/ "/".

  - Other changes to (ttn shellutils)

	A new proc is exported: `(sysfmt . args)'.  Docstring:
	Apply `format' to ARGS and pass to `system'.

  - Changes to (ttn html-data)

	Three new procs are exported: `ull', `copyright' and
	`copyright-since'.  Signatures are:

	(ull x)		; variant of `ul' that takes single list
	(copyright . prefix)
	(copyright-since year)


- 0.25 | 2001-11-02

  - New distributed file: BUGS

	This file is formatted similarly to the guile BUGS file.

  - New module: (ttn html-data)

	This module exports the following procs:
	 (html . x)
	 (href url text)
	 (head . x)
	 (title x)
	 (body . x)
	 (h1 x) (h2 x) ... (h6 x)
	 (hr)
	 (b x)
	 (p)
	 (textarea rows cols name x)
	 (input type name value)
	 (form method action . x)
	 (refresh sec url)
	 (li x)
	 (ul . x)
	 (table x)
	 (tr x)
	 (td x)

	All procs return lists (possibly nested), suitable for walking
	w/ a "flattening" procedure, something like:

	(define (output-html-data x)
	  (cond ((string? x) (actual-output x))
	        ((list? x) (for-each output-html-data x))
	        (else (error "unknown type:" x))))

	ACTUAL-OUTPUT might send X to a port, for example.


- 0.24 | 2001-10-30

  - (ttn gap-buffer) `gb?' is now exported

  - New module: (ttn parse-rfc822)

	This module exports one procedure `parse-rfc822', which has the
	following docstring:

	Parse MAILMSG, return a query proc.
	MAILMSG may be a gap-buffer, or something that `make-gap-buffer'
	takes.  The query proc takes a symbol COMP, and returns the
	message component COMP.  Supported values for COMP are:
	  from    -- first whitespace-delimited token
                     after "^From " on first line
	  headers -- the headers as an alist (keys are
                     strings), order maintained
	  body    -- rest of the mail message
	Any other query results in a "bad component" error.


- 0.23 | 2001-10-04

  - Configuration accepts guile versions 1.5.x, 1.6.x, 1.7.x

  - New distributed file: .release.scm

	This is basically a program for doing a release, included here
	for completeness.  (At this time, it is evaluated completely by
	the human maintainer, but the goal is to eventually mechanize
	the release process entirely.)

  - Bugfixes

    - (ttn stringutils) `string-append/separator' handles null 2nd arg

    - (ttn pgtable) `tuples-result->table' handles zero-tuples case

	Previously, `tuples-result->table' would try to create a table
	of size 0, causing an error.  This has been changed to instead
	create a table of size 1.  (In fact, should some result imply a
	negative number of tuples, the returned table still has size 1.)

  - Throughout, `write-line' replaced w/ `display' and `newline'

  - New platform: benchmarks fubar for now


- 0.22 | 2001-08-26

  - "make check" for `call-process' no longer looks for exact string match

	Part of the testing for (ttn shellutils) proc `call-process' is
	to check the buffer caputuring standard error for non-successful
	calls.  Previously, the stderr output was subject to direct
	string comparison w/ an expected value.  This methodology fails,
	however, when the locale differs from that of the author's,
	since a localized error message will be unlikely to match
	directly, even though the behavior is correct.  (In this case,
	/bin/ls stderr in French is definitely not the same as that in
	English.)

	Thus, these tests now check that the stderr buffer is non-empty
	instead of doing a string match.  This is less rigorous than
	before, but the alternative is to dig deep into localization
	issues, which is somewhat of a pain.


- 0.21 | 2001-08-14

  - Changes to (ttn scat)

    - Macro `define-sl' no longer provided
    - New proc: `sl->bytes'

  - More procedure documentation: now 88/128 (69%) complete

  - "make tea" now uses ttn/autoload.scm directly

	The benchmarking process used to process the autoload file to
	replace "guile-user" with "guile" because Guile used to set the
	default current module differently for scripts and for repls.
	Since Guile has been changed to use "guile-user" consistently
	for all situations, the substitution is no longer necessary.

  - Dependencies updated

    - Guile 2001-08-13 (from cvs repo)
    - guile-pg 0.07ttn3


- 0.20 | 2001-07-14

  - "make check" and "make tea" now allow choice of Guile interpreter

	By default, the interpreter "guile" is used (should be on your
	path).  However, you can specify another Guile like so:

	make GUILE=/some/path/to/alternate/guile check

	This is useful to identify (in)compatibilities.  At this time,
	no effort is made to infer the proper `%load-path' given an
	arbitrary Guile interpreter; you will need to set env var
	`GUILE_LOAD_PATH' manually, either in the shell or passing it
	through `make' similar to how `GUILE' is handled.

  - Changes to (ttn scat)

	This module, not under any active development since 1998, has
	undergone an overhaul.  A bug was fixed in `make-scat' whereby
	omission of DEFAULT-VALUE caused an error.  The semantics of
	`scat-concat' are now reversed: first arg is shifted instead of
	second arg.  `make-sl' now accepts field defs w/ numeric values
	instead of hex strings, as long as the bit length is also
	specified.  User var `sl-field-def-check-func' is no longer
	available and `make-sl' no longer calls it.  `copy-sl' has been
	renamed to `sl-copy'.  `sl-edit' has been renamed to `sl-edit!',
	and a copying (non destructive) version named `sl-from-template'
	is now available.

	Additionally, we now do some (ttn scat) testing on "make check".


- 0.19 | 2001-05-19

  - Benchmarking news

	A traversal benchmark for (ttn gap-buffer) was added.  The poor
	results prompted a complete implementation redesign (interface
	not changed).  Traversal performance is now linear w/ buffer
	size instead of being higher-powered (it was so very VERY *VERY*
	inefficient we had to kill the benchmark for fear of burning out
	the CPU on the dev box -- how's that for "higher-power"? ;-).

	Redesign helped all operations.  Insert performance is now
	linear w/ length of inserted string.  Delete performance is now
	O(1).  Intimate module (ttn edit) was also modified with
	expected (though unquantified) performance gain.

  - Administrative news

    - Sync w/ development Guile (1.5.x)

	There was previously some attempt at remaining compatible w/
	guile-1.4, but that has now been abandoned, due primarily to
	these features / changes available only in the Guile development
	version:

	- (ice-9 rdelim) required for `read-line' and friends, but
	  `use-modules' no longer can be conditionalized, so the old
	  `(or (defined? 'read-line) (use-modules (ice-9 rdelim)))'
	  workaround no longer works.

	- new `use-modules' functionality -- actually, we have
	  sufficient functionality in (ttn moduleutils) (where the new
	  `use-modules' stuff originated :-) but `use-modules' is more
	  standard.

	- `identity' makes a return -- this used to be provided in
	  (ice-9 boot-9), then someone renamed it `id' then someone
	  re-renamed it `identity' again; welcome back, old friend!

	- (ice-9 documentation) and (ice-9 session) now work together
	  to allow files following the Guile Documentation Format
	  Version 2 to be added to the search list and searched using
	  "(help 'NAME)".

	- (scripts read-scheme-source) is useful for bin/snarf-docs.

    - README dependencies more explicit

	Starting w/ this release, the Dependencies section in the README
	specifies a development version (usually by date) of guile-core
	that works well with ttn-pers-scheme (has been tested with "make
	check").

	Sorry for the extra hassle.  Once the next version of Guile is
	released (supposedly Real Soon Now), we will sync to that and
	try to remain there.

	Aside from the overall Guile dependency, we also list packages
	or programs which are required for specific modules.

    - Documentation methodology starting to be defined

	Doing "make" now also produces ttn-procedures-docs.txt, which
	follows the Guile Documentation Format Version 2, and doing
	"make install" copies this file to the site directory (see
	README).  The goal is to have all exported procedures documented
	by next release; this is approximately 59% (76/129) complete.

	Procedure documentation is snarfed using bin/snarf-docs, a work
	in progress (significant bits of which will probably make it
	into Guile's (scripts read-scheme-source) module at some point).
	So far, the recognized format is simply: contiguous comments
	before a top-level form that share the same number of leading
	semicolons.  See bin/snarf-docs for more info.

    - Makefile changes

	Bug fixes: Symlink now made in ttn/testing/benchmarks and bin.

	On "make dist", touch file ttn/testing/ttn/.dont-check-gpgutils
	in dist dir to prevent "make check" from trying to check (ttn
	gpgutils).  See README for more info.

	Documentation is now made and installed.  See news item above.

	Doing "make clean" now removes autoload and docs files.

  - New program: snarf-docs

	This is described above (see "Administrative news").

  - New module: (ttn gpgutils)

	This module exports eight procedures:

	 read-line-no-echo                      siginfo:signer
	 clearsign-message-interactively        siginfo:time
	 verify-signed-message                  siginfo:method
						siginfo:sig
						siginfo:body

	See the documentation for each procedure for more info.

  - Changes to (ttn sanity)

	New proc `know-that-ttn-procedures-docs-is-available!' finds
	ttn-procedures-docs.txt using `%search-load-path' and adds the
	file's absolute path to (ice-9 documentation) variable
	`documentation-files'.  This allows "(help 'grep)" to work, for
	example.  This proc is exported.

	Proc `give-me-a-sane-environment-please!' now calls
	`know-that-ttn-procedures-docs-is-available!'.  Also, it no
	longer tries to activate the readline library.

  - Changes to (ttn edit)

	A bug was fixed in `e:re-search-forward' that caused the "^"
	regular expression operator to fail to match beginning of line.
	This was due to omission of the `regexp/notbol' flag.

  - Changes to (ttn pgtable)

	Column definitions are now extended to have the form:
	  (NAME TYPE [OPTION1 OPTION2...])
	Each OPTION is a string.  Options are concatenated with spaces
	separating them and passed directly to the back end.  The old
	form: (NAME . TYPE) is still recognized but is deprecated and
	will no longer be supported in a future release.

	Three new element extraction procedures are now available:
	`def:col-name', `def:type-name' and `def:type-options'.  These
	DTRT w/ both the old form and the new.


- 0.18 | 2001-04-09

  - Benchmarking news

	Some benchmarks for (ttn ftw) have been added.  You can run them
	specifically with: "make benchmarks=ftw tea".

	The "make tea" command now also works in ttn/testing/benchmarks.

	To reduce maintenance, benchmark common header code now no
	longer uses modules directly with `use-modules', with the
	exception of a few absolutely required modules.  Instead, the
	autoload file created by top-level "make" is copied to the
	benchmark directory and modified so that the module defined to
	receive all the autoloaded bindings is `(guile)' instead of
	`(guile-user)'.  This modified copy is then loaded.

	The release process now specifies capturing logs of two runs of
	"make tea", the second with iteration multiplier of 1/5.  These
	logs are in the benchmark directory as tea-1.0 and tea-0.2,
	respectively.

  - Changes to (ttn ftw)

	Both `ftw' and `nftw' procs now handle optional arg sequence
	`hash-size N', where N specifies the size of the hash table to
	used for recording visited files.  The default size is 211.
	Thanks to Brad Knotwell for this performance enhancement patch.

  - New module: (ttn pgtype)

	PostgreSQL type conversion has been moved into this module.
	Additionally, PostgreSQL array conversion has been generalized.

	Docstrings reproduced here (note that `define-db-col-type'
	docstring has changed from VERSION 0.16 release):

    - `dbcoltype-lookup' TYPE-NAME
    - `dbcoltype:name' LOOKUP-VALUE
    - `dbcoltype:default' LOOKUP-VALUE
    - `dbcoltype:stringifier' LOOKUP-VALUE
    - `dbcoltype:objectifier' LOOKUP-VALUE

	Return lookup value of TYPE-NAME, a symbol, from
	`*db-col-types*'.  Use procs `dbcoltype:name',
	`dbcoltype:default', `dbcoltype:stringifier' and
	`dbcoltype:objectifier' on this value to access those
	components, respectively.

    - `define-db-col-type' NAME DEFAULT STRINGIFIER OBJECTIFIER

	Register type NAME with DEFAULT, STRINGIFIER and OBJECTIFIER
	procs.  NAME is a symbol.  DEFAULT is a string to use if the
	Scheme object is #f.  STRINGIFIER is a proc that takes a Scheme
	object and returns a string suitable for use in an "INSERT
	VALUES" SQL command.  OBJECTIFIER is a proc that takes a string
	and returns the Scheme object parsed out of it.

	Both STRINGIFIER and OBJECTIFIER need not worry about SQL-style
	quoting (using single quotes) and related quote escaping.

	If NAME already exists, it is redefined.

    - `define-db-col-type-array-variant' COMPOSED SIMPLE [PROCS]

	Register type COMPOSED, an array variant of SIMPLE, with
	optional PROCS.  SIMPLE should be a type name already registered
	using `define-db-col-type'.  COMPOSED is conventionally formed
	by appending SIMPLE with one or more pairs of square braces
	"[]", with the number of pairs indicating the array
	dimensionality.  For example, if SIMPLE is `text', a
	two-dimensional text array would be named `text[][]'.

	Optional arg PROCS is a list specifying alternative stringifier
	and objectifier procedures (in that order).  If unspecified,
	SIMPLE is looked up and its stringifier and objectifier are
	used.

	The default value of all array types is "{}" and cannot be
	changed.

  - Changes to (ttn pgtable)

	Typing machinery has been moved to (ttn pgtype).  This includes
	var `*db-col-types*' and convenience proc `double-quote'.  Also,
	some modules (ttn pgtable) depends on are now autoloaded.

  - New program: display-rcs-state-summary

	This is a simple wrapper around the procedure of the same name
	in (ttn rcsutils).  This type of wrapping may be automated in
	the future.

  - Program generate-autoload no longer sets env var GUILE_LOAD_PATH

	The makefile that calls generate-autoload now sets the path
	appropriately.  This change is so that generate-autoload can be
	conveniently called in different contexts.  Generally, programs
	in this directory should not set GUILE_LOAD_PATH.

  - Administrative news

	Doing "make rcs-state-summary" now also checks subdir bin.

	Distribution also includes benchmark logs.  These are to satisfy
	curiosity only; it's unlikely that the numbers mean anything.


- 0.17 | 2001-03-29

  - Administrative changes

    - Autoload file no longer distributed

	Instead it is created when you do "make".  The script
	bin/generate-autoload also serves as a sanity check for the
	package since it uses some of the modules right from the
	distribution.

	The no longer necessary elisp program ttn/update-autoload.el has
	been removed.

    - New directory: bin

	This holds maintenance scripts.  At this time, there is only
	generate-autoload.

  - Change to (ttn stringutils)

    - New proc: string-quoted/escaped STRING

	Return `write'-like output of STRING, with some characters
	escaped.  This means that instead of, for example, a literal
	newline appearing in the output, the two-character sequence
	`\n' is used.  The complete list of escaped characters is:
	#\\, #\newline, #\ht, #\cr, #\page.  These result in `\\',
	`\n', `\t', `\r' and `\f', respectively."

  - Changes to (ttn pgtable)

    - New supported type: `serial'

    - `pgtable-manager' enhancements

	The closure returned by this proc now can handle commands
	`delete-rows' and `update-col', and queries `table-name' and
	`defs'.

	Also, when doing a `drop', if the table has any columns that
	have type `serial', any associated magic PostgreSQL sequences
	are also dropped.  As of PostgreSQL 7.x, the sequence name is
	TABLENAME_COLNAME_seq.  For more info, see the PostgreSQL User
	Guide, Chapter 3: Data Types.

    - Bugfix in type `text[]'

	The objectifier proc now handles the empty set correctly.

  - Bugfix in (ttn fileutils)

	The autoload trigger lists now include `gb->string'.

  - Change to (ttn grep)

	Both `grep' and `grep-matches' now take an additional flag
	`literal' which means that STRING is to be searched for
	literally and not to be considered a regular expression.

  - Change to (ttn gap-buffer)

	Proc `make-gap-buffer' now takes an arg INIT, which can be a
	string or an input port.  If INIT is a port, it is read until
	EOF.  In both cases point is left at the end of the buffer.

  - Changes to (ttn edit)

    - New "e:" proc: `buffer-port'

	This returns a soft port on the current buffer.
	See bin/generate-autoload for an example use.

    - Bugfix in `letrec' and `let*' expansion

	Previously, these forms were expanded correctly with the
	exception of the initial symbol, which was hardcoded to `let'.
	This is now fixed to preserve the variant name.

    - `insert' now also takes symbols and sexps

	The string inserted is computed by applying `symbol->string' to
	symbols, and `write' to sexps.

  - Change to (ttn shellutils)

	The exit value of applying the closure returned by
	`make-buffered-caller' to `execute' is now saved and can be
	retrieved later with the `exit-val' query.  An error is
	signalled if `exit-val' is requested before either `execute' or
	`execute/no-init' has been invoked after initial closure
	creation or command redefinition.

  - New module: (ttn listener)

	One proc is exported: `make-listener'.  Docstring follows:

    - `make-listener' LISTENER-PORT #:SETUP-SOCK #:NQUEUE
                                    #:PRE #:HANDLE #:POST

	Return a thunk that listens on TCP/IP port number LISTENER-PORT.
	Key #:setup-sock specifies a procedure that is passed the
	listener socket right before doing `bind'.  This is a good place
	for `setsockopt' calls.  Key #:nqueue specifies the number of
	queued requests, default 10.  Keys #:pre, #:handle and #:post
	specify procedures that take two args, PORT and CONN, the output
	of `accept', q.v.  These three procedures are thunkified to form
	the branches of a `dynamic-wind' operating in its own thread.
	After calling the post procedure, PORT is closed.  If pre,
	handle, and post are unspecified, the default action is to do
	nothing.


- 0.16 | 2001-03-19

  - Benchmark framework added

	You can now do "make tea" which runs some benchmarks, displaying
	user and system CPU usage for each case.  You can increase or
	decrease the iteration counts by specifying an env var like so:

	make BENCHMARK_ITERATION_MULTIPLIER=N tea

	N can be an integer, a floating point number, or even a rational
	number.  The actual number of iterations computed is displayed.
	See ttn/testing/benchmarks/Makefile var `benchmarks' for the
	full list of benchmarks.  You can specify a subset like so:

	make benchmarks='baseline gap-buffer' tea

	Furthermore, for benchmark FOO, see FOO.scm for various options
	to skip parts of the benchmarks.  For example:

	make benchmarks='gap-buffer' opts='--skip-make-gap-buffer' tea

	At this time, there are only a few benchmarks, but if your
	computer is as slow as the author's, perhaps you can escape the
	scourge of "instant tea".

  - Changes to (ttn pgtable)

    - New proc: `define-db-col-type' NAME DEFAULT STRINGIFIER OBJECTIFIER

	Register type NAME with DEFAULT, STRINGIFIER and OBJECTIFIER
	procs.  NAME is a symbol.  DEFAULT is a string to use if the
	Scheme object is #f.  STRINGIFIER is a proc that takes a Scheme
	object and returns a string suitable for use in an "INSERT
	VALUES" SQL command (including SQL quotes).  OBJECTIFIER is a
	proc that takes a string (without SQL quotes) and returns the
	Scheme object parsed out of it."

	See ttn/pgtable.scm for examples; (ttn pgtable) uses this proc
	for its own initialization.

    - New supported type: `bool'

    - Bugfixes

	The return value of the `table->object-alist' `pgtable-manager'
	managed proc is now ordered properly.

	The `timestamp' stringifier only attempts to convert numbers,
	and leaves strings alone.

  - Changes to (ttn moduleutils)

    - Retired proc: `ls'

	This was such a thin wrapper around `apropos' that we can forego
	it w/o much sadness.

    - New macro: `import-from-module' OTHER-MODULE-NAME . NAMES

	Import into current module from OTHER-MODULE-NAME variables
	NAMES.  Each element of NAMES is taken to be either a variable
	name, or a pair (OLD-NAME . NEW-NAME) describing the mapping to
	be used.  OLD-NAME needs to be defined in OTHER-MODULE-NAME,
	though it need not be exported.

    - New macro: `reexport-from-module' OTHER-MODULE-NAME . SPECIFICALLY

	Re-export variables from module named OTHER-MODULE-NAME.  If
	optional args SPECIFICALLY is the empty list, all public
	variables from OTHER-MODULE-NAME are exported, otherwise, each
	element in SPECIFICALLY is taken to be either a symbol naming a
	variable to be exported, or a pair of symbols of the form
	(OLD-NAME . NEW-NAME) describing the mapping to be used.
	OLD-NAME should name an exported variable in OTHER-MODULE-NAME.

	This macro was snarfed from in THUD.

  - Changes to (ttn fileutils)

    - Performance enhanced procs: `find-file', `save-buffer'

    - New proc: `write-buffer' BUFFER FILENAME

	Write BUFFER into file FILENAME.
	FILENAME is passed through `expand-file-name', q.v.
	This makes the buffer visit that file.
	FILENAME cannot be a directory.

  - Changes to (ttn gap-buffer)

    - Performance enhanced proc: `gb->lines'

    - Bugfix

	`gb-goto-char' now returns `new-point', even if that is an
	invalid point (in which case, point is set to `point-min' or
	`point-max' anyway).  This mimics the behavior of GNU Emacs.

  - New module: (ttn edit)

	This module exports a macro `editing-buffer'.  Docstring:

	Consider BUFFER the "current buffer" and execute BODY.
	In BODY, applications of one of `current-buffer-procs' (see
	source) are handled specially.

	There is too much other stuff to document in this NEWS file, so
	here's a short program that gives a taste of what can be done:

	#!/bin/sh
	exec guile -s $0 $0       # -*- scheme -*-
	!#
	(use-modules (ttn fileutils) (ttn edit) (ttn echo))
	(editing-buffer (find-file (cadr (command-line)))
	  (while (re-search-forward "^.+<([^>]*[^^])>" (point-max) #t)
	    (echo (match-string 1))))
	;; <find this tag>
	;; <but don't find this one:^>
	#!
	<don't find this one either>
	!#

	Save this to foo, do "chmod +x foo", install ttn-pers-scheme
	somewhere in the `%load-path' and run foo.  The program should
	display "find this tag".

	Procs supported: point, point-min, point-max, buffer-string,
	buffer-substring, goto-char, forward-char, backward-char,
	match-data, set-match-data, match-string, looking-at,
	match-beginning, match-end, search-forward, search-backward,
	re-search-forward, replace-match, insert, delete-char,
	erase-buffer, delete-region.

  - Testing news

	Testing (ttn pgtable) is now more flexible; the database name
	you set up need not be the same as your user name.  See README
	for revised instructions.


- 0.15 | 2001-03-07

  - New module: (ttn stringutils)

	This module was actually introduced in version 0.14 but not
	documented then.

	Four procedures are provided: `string-ws-trim',
	`string/sep->substrings', `string-append/separator'
	and `mapconcat'.

	See docstrings for more info.

  - New module: (ttn pgtable)

	Four procedures are exported: `tuples-result->table',
	`display-table', `where-clausifier' and `pgtable-manager'.
	Docstrings are reproduced here.

    - `tuples-result->table' RES

	Extract data from the tuples result RES, and return an annotated
	array.  The array's values correspond to the data from RES, and
	has dimensions `pg-ntuples' by `pg-nfields'.  Annotations are
	object properties:

        names   -- vector of field names
        widths  -- vector of maximum field widths

    - `display-table' TABLE

	Display table TABLE, including header and ASCII field-separators.

    - `pgtable-manager' DB-NAME TABLE-NAME DEFS

	Return a proc that manages a table specified by DB-NAME
	TABLE-NAME DEFS.  DB-NAME and TABLE-NAME are strings.  DEFS is a
	list of symbol pairs, the car being the field name and the cdr
	the PostgreSQL type name.  For example:

	  ((time  . timestamp)
	   (error . text)
	   (files . text[]))

	The returned proc accepts a single arg CHOICE (a symbol) and
	returns either the variable or procedure associated with CHOICE.
	When CHOICE is `help' or `menu', return a list of accepted
	choices.  When CHOICE is `types-list', return a list of
	supported PostgreSQL types.

    - `where-clausifier' STRING

	Return an SQL "where clause" from STRING.

  - Changes to (ttn grep)

	Both `grep' and `grep-matches' now understand a `count' flag
	directing these procs to return the length of the result list
	instead of the list itself.  This flag can be used (in a list)
	with the other flags.

  - Testing news

	There is a new file ttn/testing/ttn/check-pgtable.scm that tests
	(ttn pgtable).  By default, this is not enabled because there
	are dependencies that are not easy to check for or guarantee (we
	might get to it later, however).

	See file README for instructions on how to get things to work.


- 0.14 | 2001-03-05

  - Retired module: (ttn original-sin)

	This was an experiment that may be revived as (ttn reflection).
	A better approach (for strings, at least) is Keisuke Nishida's
	embedded format module -- see (ttn eformat) immediately below.

  - New module: (ttn eformat) --- Repackage Keisuke Nishida's (eformat)

	This code was posted originally w/o copyright by Keisuke
	Nishida:
	     From: kxn30@po.cwru.edu
	     Subject: eformat
	     Date: Tue, 19 Dec 2000 10:22:49 -0500
	     To: guile-sources@gnu.org, guile-user@gnu.org

	The following changes have been made:
	(1) Module name now `(ttn eformat)'.
	(2) Added `activate-eformat' and `deactivate-eformat'.
	(3) Removed surrounding `let'.
	(4) Procedure `embedded-format' now does not expect the first
	    entry to be a port, and no longer applies `format' to that
	    port at read-time.  Instead, we return an application of a
	    closure that does the format, hardcoding the first arg of
	    `format' to #f.  Thus, the result is always a string.  (See
	    example below.)

	To use, place `(activate-eformat)' and `(deactivate-eformat)'
	around the code that represents a string as `#[ stuff ]'.  Note
	that these forms modify `read-hash-procedures' and work by
	read-time side-effects, so they must be at top-level.

	*WARNING WARNING WARNING*: Do not forget `deactivate-eformat'!

	Example:
	(activate-eformat)
	(define (zz x)
	  #[$x is a $(lambda () (if (number? x) "" "non-"))number])
	(deactivate-eformat)

	(zz 42) => "42 is a number"
	(zz zz) => "#<procedure zz (n)> is a non-number"

	See ttn/testing/ttn/check.scm for more examples.


- 0.13 | 2001-02-27

  - New module: (ttn grep) --- Pick mappings according to regexp line match

	Two procs are exported: `grep' and `grep-matches'.
	Docstrings are reproduced here.

    - `grep' RE STRINGS . FLAGS

	Return list of matches to regexp RE from STRINGS list.  Optional
	args FLAGS are zero or more symbols which modify regexp
	matching:
	`invert'    -- the result list is those strings that do NOT match
	`fold-case' -- consider upper and lower case to be identical
	Return list order is the same as in the input list.

    - `grep-matches' RE STRINGS . FLAGS

	Return non-#f match results of `regexp-exec' of regexp RE on
	STRINGS list.  Optional args FLAGS are zero or more symbols which
	modify regexp matching:
	`invert' -- the result list is composed entirely of #t values,
                    with length equal to the number of strings that do
                    NOT match (not very useful; this option listed for
                    completeness)
	`fold-case' -- consider upper and lower case to be identical
	Return list order is the same as in the input list.

  - New module: (ttn fileutils) --- Munge files in core

	Two procs are exported: `expand-file-name' and `find-file'.
	Docstrings are reproduced here.

    - `expand-file-name' NAME . DEFAULT-DIRECTORY

	Convert filename NAME to absolute, and canonicalize it.  Second
	arg DEFAULT-DIRECTORY is directory to start with if NAME is
	relative (does not start with slash); if DEFAULT-DIRECTORY is #f
	or missing, `(getcwd)' is used.  File name components that are
	`.' are removed, and so are file name components followed by
	`..', along with the `..' itself; note that these
	simplifications are done without checking the resulting file
	names in the file system.
	An initial `~/' expands to your home directory.
	An initial `~USER/' expands to USER's home directory.

    - `find-file' FILENAME

	Insert file named FILENAME into a buffer, and return that
	buffer.  FILENAME is passed through `expand-file-name', q.v.
	If the file is not readable, return #f.

    - `save-buffer' BUFFER

	Save BUFFER to disk.  BUFFER should be from `find-file', q.v.
	An error is thrown if BUFFER's `filename' object-property is not
	set.  An error is thrown if the file cannot be written.

  - Changes to (ttn gap-buffer)

    - new proc: `gb-erase!' GB

    - new proc: `gb-filter!' GB PROC-TAKING/RETURNING-STRING

    - new proc: `gb->lines' BG

    - new proc: `gb-filter-lines!' GB PROC-TAKING/RETURNING-LIST

    - `make-gap-buffer-port' now throws error if arg not a gap-buffer

  - Changes to (ttn shellutils)

    - new proc: `args->normalized-list' ARGS

	For ARGS, trim surrounding whitespace, join together separated
	by space, then split apart on space boundaries.  For example:

	(args->normalized-list "rsync a b c" " d " "e" "f g h")
	=>
	("rsync" "a" "b" "c" "d" "e" "f" "g" "h")"

	This is used by `call-process' and `call-process->buffers'.

    - new proc: `call-process' PROGRAM . ARGS

	Call PROGRAM synchronously in separate process w/ optional ARGS.
	Key `#:inp' specifies the program's input port; #f means
	`/dev/null'.  Key `#:outp' specifies the program's output port;
	#f means discard output; 0 means discard and don't wait.  Key
	`#:errp' specifies the program's error output port; #f means
	discard output.  If omitted, ports default to standard input,
	output and error ports, respectively.

	If the output port is 0, return immediately with value #f.
	Otherwise wait for PROGRAM to terminate and return a numeric
	exit status.

	Key `#:norm' specifies whether or not PROGRAM and ARGS should be
	passed through `args->normalized-list' (default no processing).

    - new proc: `call-process->buffers' PROGRAM . ARGS

	Call PROGRAM synchronously in separate process w/ optional ARGS.
	Key `#:inb' specifies the program's input buffer.  Key `#:outb'
	specifies the program's output buffer Key `#:errb' specifies the
	program's error output buffer.  Default is that all buffers are
	specified as `#f.  Key `#:norm' specifies whether or not PROGRAM
	and ARGS should be passed through `args->normalized-list'
	(default no processing).  Return a numeric exit status.

    - new proc: `make-buffered-caller' PROGRAM . ARGS

	Return a procedure that is capable of calling PROGRAM w/
	optional ARGS.  When called, the program stdout and stderr are
	captured to buffers.  The key `#:inb' (default #f) specifies an
	input buffer to use for the call.

	The returned procedure takes one of the following commands (a
	symbol):

	redefine NEW-DEF -- Redefine the called program and its args.
	                    NEW-DEF is one or more strings.
	execute          -- Call program, clearing buffers first.
	                    Return exit status of the program.
	execute/no-init  -- Same as `execute', w/o clearing buffers first.
	outbuf           -- Return output buffer object.
	outbuf-string    -- Return output buffer as a string.
	outbuf-lines     -- Return output buffer as a list of strings.
	errbuf           -- Return error output buffer.
	errbuf-string    -- Return error output buffer as a string.
	errbuf-lines     -- Return error output buffer as a list of strings.


- 0.12 | 2001-02-18

  - New module: (ttn gap-buffer) --- String buffer that supports point

	A gap buffer is a structure that models a string but allows
	relatively efficient insertion of text somewhere in the middle.
	The insertion location is called `point' with minimum value 1,
	and a maximum value of the length of the string (which is not
	fixed).

	A gap buffer port is a soft port (see Guile manual) that wraps a
	gap buffer.  Character and string writes, as well as character
	reads, are supported.  Flushing and closing are not supported.

	These procedures are exported:
	make-gap-buffer
        gb-point GB
        gb-point-min GB
        gb-point-max GB
        gb-insert-string! GB STRING
        gb-insert-char! GB CHAR
        gb-delete-char! GB COUNT
        gb-goto-char GB LOCATION
        gb->string GB
        make-gap-buffer-port GB

	The term and concept of "gap buffer" are borrowed from Emacs.
	We will gladly return them when libemacs.so is available. ;->
	This implementation could be made (much) more efficient; the
	goal of this release is to establish the exported interface.

	See ttn/testing/ttn/check.scm for example usage.


- 0.11 | 2001-01-31

	No user visible changes, if you are using guile-1.4.

	If you are using guile from cvs post 2001-01-21, this release
	allows (ttn shell-command-to-string) to work properly.


- 0.10 | 2001-01-26

  - Change to (ttn cron)

	In the case where COMMAND matches no key, `cron' now silently
	returns #t instead of #f.  This is to facilitate wrapping by
	`exit', as demonstrated here:

	#!/bin/sh
	exec guile -s $0 "$@"
	!#
	(exit
	  (cron (command-line)
	        ((hourly) (write-line "yawn, stretch"))
	        ((daily)  (write-line "zonk"))))

  - Changes to (ttn rcsutils)

    - new proc: `rcs-files-under'

	In directory DIR, return a list of DIR/RCS/*,v files.

    - proc no longer exported: `group-files-by-rcs-state'

	This was used by `display-rcs-state-summary' but nowhere else,
	so we thought it better to hide it until actually needed outside
	the module.

    - `display-rcs-state-summary' output now line-oriented

	Lines are formatted: `STATE YYYY/MM/DD HH:MM:SS FILENAME'
	and are ordered most recently changed first, grouped by state.
	FILENAME is constructed using the DIR argument and will always
	have at least one directory-element separator ("/" for all
	systems ttn cares about).


- 0.09 | 2001-01-12

  - Changes to (ttn html-world)

	The initial image used as a basis for the output now has size
	200 by 200 pixels (it used to be 150 by 150).

  - New module: (ttn shellutils) --- work easily with the Bourne shell

	Two procedures are exported: `system*' and `email-log-if-fail'.
	See docstrings for more info.


- 0.08 | 2001-01-05

  - New module: (ttn cron) --- Provide cron-compatible framework

	This file exports `cron', a macro that has signature:
	(cron CMDLINE-LIST . HANDLERS)

	CMDLINE-LIST is a two-element list, of the form:
	(PROG-NAME COMMAND)

	HANDLERS is zero or more forms:
	((KEY) (BODY1) (BODY2) ...)

	BODY1, BODY2 and so on are optional.  KEY is a symbol or string.
	The body forms for a KEY are executed if COMMAND matches KEY.
	If CMDLINE-LIST does not have exactly two elements, a usage message
	is displayed and a "bad usage" error is signalled.  If COMMAND
	matches no key, `cron' silently returns #f, otherwise, the return
	value is the value of the last body form executed.

	Example:
	#!/bin/sh
	exec guile -s $0 "$@"
	!#
	(cron (command-line)
	      ((hourly) (write-line "yawn, stretch"))
	      ((daily)  (write-line "zonk")))

	See also link at end of http://www.glug.org/projects/wurlds.html
	for example usage.


- 0.07 | 2000-12-26

  - Bug fix for (ttn html-world)

	Line chunking was dropping match prefixes.  This is now fixed.
	Although we don't yet have a test for "make check", here's some
	code you can use to do visual inspection:

	(use-modules (ttn echo) (ttn html-world))
	(echo "<html><body><pre><tt>")
	(render-html-world display (html-world "just" "some" "urls"))
	(echo "</tt></pre></body></html>")

	If you put the above into test.scm, you can do:
	guile -s test.scm > test.html

  - `shell-command->list' now takes optional 2nd arg

	This arg (or the symbol `trim', if omitted) is passed to
	`read-line'.  Valid args are symbols `trim', `peek', `concat'
	and `split'.


- 0.06 | 2000-12-24

  - New module: (ttn mixp) --- simplified abstraction for mixp

	You must install mixp to use this module.
	See http://www.thbz.cx/mixp for more info.

  - Bug fix for "make check"

	A bug was fixed in the Makefile that caused "make check" to fail
	for the distribution but not in the developer build directory,
	which goes to show that the development process is not quite
	mature yet.  (Although "make check" failed, ttn-pers-scheme 0.05
	installed correctly and was usable, anyway.)

  - Path to xearth no longer hardcoded in (ttn html-world)

	To use (ttn html-world) you need to have `xearth', `mogrify',
	`pbmtoascii' and `tr' in your PATH somewhere.


- 0.05 | 2000-12-22

  - Changes in installation procedure

	We now use the standard configure.in-based install procedure,
	which uses "guile-config info pkgdatadir" to determine where to
	install files.  See file README for more info.

	Also, the file INSTALL has been removed from the distribution.

  - New modules

    - (ttn slackful-concat) --- Low-hassle string/symbol concatenation
    - (ttn html-world) --- Render xearth in ASCII, inserting HTML links

	See the Commentary and procedure docstrings in each module
	for more info.

  - Changes to (ttn sanity) and (ttn original-sin)

	The development version of Guile no longer allows certain procs
	(e.g., `getenv') to accept symbols.  We use `symbol->string'
	where things break; however, there has not been a systematic
	code review.

  - Changes to (ttn shell-command-to-string)

    - new procs: `shell-command-to-list', `shell-command->list'

	The second is an alias for the first.  Each line of the shell
	command (sans newline) is an element of the return value, a
	list.


- 0.04 | 2000-10-16

  - New modules

	See the Commentary in each module for more info.

    - (ttn filesystem-tree-to-list) --- grok fs tree as nested list
    - (ttn original-sin) --- frictionless value transfer
    - (ttn moduleutils) --- module system hackery

  - Retired modules

    - (ttn @) --- Macro to peek at variables inside modules
    - (ttn ls) --- Commands for interactive module browsing

	These were merged into (ttn moduleutils).

  - Changes to (ttn echo)

    - new procs: `echow', `echo-n' and `echow-n'

	`echow' is similar to `echo' but uses `write' instead of `display'.
	The "-n"-named variants are similar to those w/o "-n" but omit the
	terminating newline.


- 0.03 | 2000-10-02

  - New modules

	See the Commentary in each module for more info.

    - (ttn scat) --- serial concatenation for building packets
    - (ttn abbrev-tree) --- hierarchical abbreviations
    - (ttn slice) --- sub-sequence selection using indexes

  - Changes to (ttn dirutils)

    - new proc `filename-sans-end-sep-proc'

	This procedure takes one arg, a character used to separate
	components in a filename, and returns a procedure that, given a
	filename, returns the filename w/o a terminating separating
	character.  In the case where the filename consists only of the
	separating character, it is returned unchanged.

  - Changes to (ttn ftw)

    - bug fix in procedure `ftw'

	The procedure `ftw' now returns #t only if all callbacks return
	#t.  Any value other than #t is immediately taken as the return
	value for `ftw'.  Previously, only #f would terminate the walk.

    - procedure `nftw' implemented

	This is similar to `ftw' but w/ more options ("nftw" stands for
	"new filesystem tree walk").  See Commentary in ftw.scm for more
	info.

  - Testing framework added

	The Makefile target "check" (i.e., "make check") now runs
	several tests to ensure functionality of various modules.  If
	you see an error when doing this, please mail the maintainer the
	result of running the command "make DEBUG=1 check".

  - More helpful Makefile

	The default target (i.e., "make") now displays command lines
	grepped from the Makefile directly.  Here is the current output:

	$ make
	No default action makes sense; read the Makefile.
	Or be adventurous, and try one of these:
	 make check
	 make TTN_CHECK_DONT_MUNGE_LOAD_PATH=1 check
	 make instdir=DIR install
	 make newtag=SOME-TAG tag
	 make disttag=FULL-TAG dist
	 make rcs-state-summary
	 make snapshot


- 0.02 | 2000-09-12

  - New modules

	See the source file for each module for more info.

    - (ttn dirutils) --- handle directories
    - (ttn ftw) --- filesystem tree walk
    - (ttn rcsutils) --- handle RCS files
    - (ttn shell-command-to-string) --- capture shell command stdout

  - Makefile now supports "make instdir=DIR install".

	See file INSTALL for installation instructions.

  - You can now use update-autoload.el in the distribution.

	If there is no RCS subdir in the distribution directory, one is
	created and autoload.scm is registered there.  From then, when
	you load update-autoload.el into Emacs, it will do the right
	thing and keep a history of its changes to autoload.scm.


- 0.01 | 2000-07-09

	This initial release is mainly to get the release process going
	so that we may debug it.



	Local Variables:
	mode: outline
	outline-regexp: "\\([ ][ ]\\)*- "
	fill-column: 72
	fill-prefix: "\t"
	End: