2004-01-03  Jens Gecius  <jens@gecius.de>

	* vgrabbj.conf.5/vgrabbj.1: 
			Added all missing options to the
			manuals, corrected misspelled
			options. Certain clarifications.
	* vgrabbj.c:	Added option -X to let daemon
			mode run in foreground (e.g. not
			detaching from current terminal)

2003-12-31  Jens Gecius  <jens@gecius.de>

	* v_config.c: 	Changed pointer-update to be
			correct in any case. Fixed error
			in archiving functionality.

2003-10-06  Robert van der Meulen <rvdm@wiretrip.org>
	* General:	Added support for passive ftp.
 
2003-09-10  Jens Gecius  <jens@gecius.de>
	* various	Included a patch from Koos van den
			Hout and Arthur van Leeuwen to
			enable vgrabbj to flip top to
			bottom. This enables upside-down
			mounting, e.g. on a ceiling.
	* various	Included a patch from David Austin
			to enable vgrabbj to take detailed
			control of the cams hardware 
			settings (e.g. brightness-adjustment,
			hue, etc.).
			This one needs further testing and
			adjustment to the options setup of
			previous versions. As of now this
			version might break your setup.

2003-07-28  Jens Gecius  <jens@gecius.de>
	* v_conf.c:	Fixed typo in help screen
			(thanks to Tobias Wolter who filed
			this bug to DBTS)
			Fixed bug causing segfault if not
			all optional libraries are present
			(thanks to Pascal Brisset for the
			patch! Sorry it took so long to
			apply it)

2002-10-15  Jens Gecius  <jens@gecius.de>

	* vgrabbj.h: 	Added option nousemmap needed for
			certain cams not switchable to
			palette RGB24 but unable to use
			mmap'ed memory.

	* v_options.h:	Fixed a small bug for ForcePalette.
			Improved img_size calculation.

2002-10-14  Jens Gecius  <jens@gecius.de>

	* v_config.c: 	Fixed nasty bug causing boolean
			commandline options not being parsed
			correctly.
			Added switch SwapRL (-R) to mirror
			the resulting image left-right.

2002-04-12  Jens Gecius  <jens@gecius.de>

	* v_config.c: 	Fixed bug for commandline switches
			on boolean values (segfault)
			Fixed possible wrong treatment of
			empty lines in conf

2002-04-02  Jens Gecius  <jens@gecius.de>

	* 		Final touch up, release 0.9.0
			with archive functionallity.

2002-03-29  Jens Gecius  <jens@gecius.de>

	* v_config.c:	rewrote/enhanced archive handling
			archivemax/archiveeach to manage
			archived pictures added.

2002-03-27  Jens Gecius  <jens@gecius.de>

	* v_config.c: 	(after release) Fixed bug in ttf
			init. Changed internals slightly.

	* General: 	Update manuals, uploaded man's to
			cvs, increased version no. to 0.8.9,
			release.

2002-03-26  Jens Gecius  <jens@gecius.de>

	* vgrabbj.c:	Added Archive option to create an
			archive of images.
			Warning: no upper limit for archive
			implemented, yet. This option might
			fill your harddrive!!
			Currently works only if archive is
			on the same filesystem as the temp-
			output (same restriction for output
			and tmpout - they have to be on the
			same filesystem)
	
	* v_config.c: 	Fixed severe bugs in ftp-code:
			1. remoteDir was not read from conf
			2. if remoteDir not set screwy things
			might have happened...

2002-02-24  Jens Gecius  <jens@gecius.de>

	* v_config.c: 	Fixed missing explanation of parameter
			-c on the commandline for the -h
			commanline parameter.

			Enhanced manual.

2002-02-18  Jens Gecius  <jens@gecius.de>

	* vgrabbj.c: 	Finished rewrite of conf-parser. 
			(Almost) identical routines are now 
			used for command-line and conf-file.
			Further information is given in case
			of wrong options.
			Cleanup of code (partially), blow-up
			on some other areas (especially the
			conf-var has to be cleaned up now...)
			Well, ongoing progress, short of
			release 0.8.8 (testing).
			OpenOnce defaults ot On now (to avoid
			memory fragmentation and usb-messages
			in syslog). Use Off only if you need
			to access the device inbetween vgrabbj
			taking frames.

2002-02-16  Jens Gecius  <jens@gecius.de>

	* v_config.c: 	Further rewrite of conf-parser (not
			finished - sigh)
			Definitions of options and certain
			detailed definitones moved to structs
			(former: in code)
			Timestamp string does not need to be in
			paranthesis (conf-file) any more (but
			can - it will be filtered!)

2002-02-12  Jens Gecius  <jens@gecius.de>

	* vgrabbj.c: 	Changed handling of open-device in case
			of openonce (it is really opened only once
			now), avoiding lot's of kernel-uhci-msgs
			on my box. Hope you guys benefit, too.
			Again, I strongly recommend to use OpenOnce

2002-02-11  Jens Gecius  <jens@gecius.de>

	* vgrabbj.c: 	More split-up of source to make it more
			readable. Re-organization of mmap (now
			close to device open to prevent memory
			fragmentation. If the device is not
			otherwise needed, I strongly advise to 
			use OpenOnce yes (now default).
			Changed handling of Timestamp, Font-eninge
			is loaded/initialized once only at startup
			and if config is re-read.

2002-02-10  Jens Gecius  <jens@gecius.de>

	* v_config.c: 	Added DaemonSeconds as alternative to Daemon
			in conf - value given is taken as seconds
			instead of milliseconds (see conf).

	* vgrabbj.c: 	Added some exit routine to mmap'ed capture
			to avoid endless loops (2).

2002-02-09  Jens Gecius  <jens@gecius.de>

	* v_config.c:	Added a parameter to discard frames before
			the actual picture is taken. Might be
			necessary, if your cam needs a little time
			for adjustment of the current brightness.

2002-02-08  Jens Gecius  <jens@gecius.de>

	* v_config.c:	Fixed a #define bug in checkfiles (if you
			do not have TTF on your system it's
			relevant, otherwise not)
			
	* configure.in:	Certain systems seem to have the freetype
			header (v1) in a different directory than
			debian does. Added appropriate check.

2002-02-06  Jens Gecius  <jens@gecius.de>

	* v_config.c: 	Closed a memory hole generated by the
			config-re-read code.
			Changed the default of SetImageSize in
			the .conf to On (Off should be the special
			case for certain cams, not vice versa).

2002-02-05  Jens Gecius  <jens@gecius.de>

	* vgrabbj.c: 	Rearranged the initializing of vgrabbj to
			accomodate a new function: the conf-file
			will be re-read if you send a SIGHUP signal
			to the daemon!
			Moved more stuff to v_config, which has 
			nothing to do with the actual grabbing.
			Adjusted the changed commandline parameters
			in the help output (thanks to Michael)			
			Reorganized memory management (hopefully
			without memory-leak - let me know if it
			causes problems - I'm currently trying to
			run it as long as possible here - 
			unfortunately my usb-drivers are a little
			irrational...

2002-02-03  Jens Gecius  <jens@gecius.de>

			(released 0.8.3)
	
	* v_config.c: 	The usage of a temporary file to be copied to
			the output file is now switchable. If you
			don't want it, use the appropriate command-
			line switch or conf-option. This may impair
			the use of vgrabbj if used on a webserver
			(webserver may access the file while vgrabbj
			writes data).
			
			IMPORTANT!! SWITCHED COMMANDLINE PARAMETER!!

			MAY BREAK YOUR CONFIG!! READ BEFORE USE!!

			If you use vgrabbj with commandline options
			instead of the configuration file, beware that
			the following switches have changed:
			-w (dont/set windowsize) is no -g
			-W (image width) is now -w
			This is due to compliance with POSIX.2 (-W).

	* vgrabbj.c: 	Merged several patches (c) Michael Janssen,
			Debian Package maintainer to streamline the
			check for files and the video-device.
			Temporary output file will now be properly
			unlinked after each use (fixes debian bug
			#127722). Shutdown of program streamlined.
			Configuration adjusted - now all file-related
			arguments are read first and checked once
			only at the end. This avoids problems with
			multiple definitions of e.g. the input-
			device. If defined on the command-line as
			last argument, only that device is checked.
			(fixes debian bug #127659).

2001-12-22  Jens Gecius  <jens@gecius.de>

	* vgrabbj.c: 	Various changes for release 0.8.2
			(no new features, make everything work).

2001-12-05  Jens Gecius  <jens@gecius.de>

	* v_writer.c: 	Changed back to system() for the call of
			/bin/cp, but I'm not satisfied with the
			situation due to the problems associated with
			system() (see man-page). Have to find a better
			solution.
			Added a new value for vconf to have the cmd-
			line for the copy action handy when needed
			(performance).

2001-12-04  Jens Gecius  <jens@gecius.de>

	* v_writer.c: 	In case output is set to something different
			than /dev/stdout (default), the output is
			written to a temp-file (name: <outputfile>.tmp)
			and then copied to the final destination
			<outputfile>. This should fix concurrent
			access problems (includes associated changes
			in other source files).

	* v_ftp.c: 	Changed function to close ftp-connection and
			added a sanity-check in an attempt to fix a
			reported error.
			Added #defines to enable successful compile if
			ftp is disabled.

	* configure.in: Changed error-message in ./configure to a
			warning message for the libraries ftplib and
			freetype - fixes longtime problem of not 
			compiling if one of these libs is not present.
			Added options to configure to intentionally
			disable timestamp and/or ftp code.
			
	Again: NO extensive tests have been performed!

2001-11-12  Jens Gecius  <jens@gecius.de> = after release

	* vgrabbj.c	Nasty little typos for the new individual
			imagesize setting and swapped height/width
			for size svga.
	
	* v_writer.c	New to the package, this file contains all
			functions related to the output of the image
			data
			Added an output option for ppm (plain linear
			RGB). You may need to switch R/G (via -S)
			because vgrabbj can not determine what mode
			your videodevice uses.
			(appropriate changes to all other files)
	* all		Removed ftplib.* from package and constructed
			dependency on the ftplib on your system. If
			don't have it, ftp is disabled (like ttf, see
			below)
			More stringent check for available libraries
			AND headers. If one of both is not available,
			timestamp and/or ftp will be disabled.

2001-11-12  Jens Gecius  <jens@gecius.de>

			Release of 0.8.0 with ftp-code, config-file,
			enhanced imagesize (even individual settings),
			bugfixes.
			Be aware: no extensive tests (it runs here -
			without ftp)!

2001-11-11  Jens Gecius  <jens@gecius.de>

	* vgrabbj.c:    Fixed a bug in png write, leading to corrupt
			data in vconf and possible segfaults
			Added xga, sxga, uxga as valid options for
			image size on command line
			Added ftp-code from Robert Wessels; be careful
			I have NOT tested this new part of vgrabbj!!
			Maybe I switch the ftplib.c someday to ftplib
			provided by that package
			Added individual imagesize (-W and -H on 
			commandline); be careful! These values are not
			checked for valid values! You may damage your
			hardware with wrong values (and a poor driver)
			Added openonce to avoid vgrabbj opens and closes
			the videodevice everytime an image is acquired;
			this renders the videodevice unuseable for other
			applications while vgrabbj sits idling in daemon
			mode. It also does not affect non-daemon mode.
			Changed handling of parameter -h in case a 
			config file is available.
			Handling of the config file is as follows:
			1. check for -h as first (!) parameter -> show usage
			2. parse configfile, if available
			3. parse commandline
			WARNING: the code is not very extensively tested!

	* v_config.c:   Added xga, sxga, uxga as valid options for
			image size in config file
			Added ftp-code from Robert Wessels
			Added individual imagesize settings (see above)
			Added openonce (see above)
	
	* (configure)	Various changes to reflect the addition of a
			config file

	* AUTHORS, README, TODO etc.
			Updated due to changes (see above)

2001-07-23  Jens Gecius  <jens@gecius.de>

			Release of 0.7.4 without any asm code (should
			keep everybody happy for porting), use of mmap,
			etc. See below for details.

2001-07-22  Jens Gecius  <jens@gecius.de>

	* vgrabbj.c:    Minor Changes, added YUV422 to known routines,
			because with my Philips cam this one looks
			identical to YUV422.
			Reminder: brightness adjust by vgrabbj is OFF
			by default now and -b switches it, BUT you 
			have to have RGB24 selected as input palette!
			Because I switched back from the asm-code to
			pure c-code (portability of the conversion
			functions), and the conversion from YUV420 is
			not exactly what it should look	like, I use a
			different approach: do two conversion, which
			work - at least here at my box. Let me know if
			something goes wrong, I'll try to fix it then.

	* ccvt_c.c: 	Added YUYV (=YUV422?) and YUV420p to the
			conversion routines (I'm getting there, soon
			vgrabbj will be portable again).
			But: YUV420i still gives me odd images...

2001-07-21  Jens Gecius  <jens@gecius.de> (no release - CVS only)

	* vgrabbj.c: 	* Added:
			Added mmap image grabbing as default, if mmap
			fails, vgrabbj falls back to read().
			Added parameter -F to force a specific palette.
			But be aware, if it is not one of vgrabbj's
			supported palettes (RGB32, RGB24, YUYV, YUV420,
			YUV420P) it will fallback to one of those (if
			supported by the device). Good to check devices
			capabilities.
			* Bugs fixed:
			Fixed a bug for RGB32-RGB24 conversion (it used
			to strip 1/4th of the image).
			* Misc:
			Brightness CHANGED! Default is now FALSE and
			-b switches the default.
			Code cleanup, better checking for v4l device.
			Buffers are now kept for the time vgrabbj runs.
			Rearranged code to avoid forward function defs.
			Put fork'ing into function (away from main()).
			Switched handling of device setup and image read.
			Setup (and finding out which palette will work)
			is now done only at startup. At every read the
			the palette and imagesize is set (if allowed to),
			in case some other program changed it (because
			vgrabbj keeps the device opened only for during
			read, while vgrabbj sleeps, you can access the
			dev with any other program).

	* vgrabbj.h: 	More code cleanup, put definitions in vgrabbj.h
			where they belong

2001-07-18  Jens Gecius  <jens@gecius.de> (no release - CVS only)

	* vgrabbj.c: 	Fixed handling of parameter -w, which was kind of
			ignored. Now it is properly handled and vgrabbj
			doesn't exit right away if it was set.
			Update hereto can be found at
			http://vgrabbj.gecius.de/vgrabbj/windowsize.diff
			please apply by entering the following command:
			cd <vgrabbj-source-dir>
			patch -p1 < windowsize.diff
			(ignore the missing header warning)
			compile as usual. Now it should work

	* configure.in: Added certain headers to check for compilation.

	* <general>	Missed to point out that 0.7.x works on i386 only.
			This is due to the conversion routines, which I
			only have in asm so far. I hope to find some more
			complete set in C, soon.
	
2001-07-14  Jens Gecius  <jens@gecius.de> (0.7.3)

	* vgrabbj.c:	Finally fixed the memory leak while using timestamp.
			Implemented a SIGTERM handler (very simple).
			vgrabbj will always finish the current image (if any)
			and then exit gracefully.
			Renamed DEBUG to LOGLEVEL and implemented DEBUGGING,
			which has to be set manually to use mtrace(). Off
			by default, of course. Usually, you shouldn't worry
			about this.

2001-07-12  Jens Gecius  <jens@gecius.de> (0.7.2)

	* vgrabbj.c: 	Fixed missing free() for the temp-buffer. This led to
			the box running out of memory and hence kill vgrabbj
			(only daemon mode).
			Fixed interchanged factors in conv_rgb32_rgb24 leading
			to a seg-fault. 

2001-07-09  Jens Gecius  <jens@gecius.de> (0.7.1)

	* vgrabbj.h: 	Changed default debug/log-level to 4.

	* vgrabbj.c: 	Changed error handling. Now there are 6 log-levels
			(silent to debug) with two types of critical errors
			resulting in abortion of execution:
			* critical errors, leading to immediate exit
			  e.g. memory-failure
			* near critical errors, leading to immediate exit
			  in non-daemon mode (user is available to take care
			  of the problem) and
			  leading NOT to exit in daemon mode. vgrabbj still
			  tries to get the action done and fails only if a
			  maximum of 3600 errors is reached. Inbetween
			  vgrabbj sleeps for a second, so this is roughly a
			  one hour (!) try-period. After this, vgrabbj exits,
			  no matter what (e.g. device access error).
			  I did this mainly because I have only one video-
			  device and didn't want to shut it down while
			  testing. This way, it tries with appropriate sleeps
			  and resets the error-counter after the current run.
			Perhaps I'll make this errorcounter configurable one
			day (let me know if you want).
			The critical error-exits are obeyed, even if you made
			vgrabbj totally silent.

			Added video palettes YUYV, YUV420P to the try-palettes.
			I hope this covers almost all devices out there.

2001-07-08  Jens Gecius  <jens@gecius.de> (0.7.0  -  not very useful, though)

	* vgrabbj.c: 	Added convertion routine from YUV 4-2-0-planar and YUYV
			to RGB24 to accomodate (hopefully) all V4L devices, 
			especially afer	recent discussion on linux-kernel list
			that there (will/is/ should) not be any conversion in
			kernel space. Only RGB24, RGB32 and YUV420 are tried,
			if the device is not set to a recognized mode.
		
			Added BGR to RGB routine, in case color of output looks
			odd.

			Made library ttf optional. If it is not available, the
			timestamp features are disabled.

			Major cleanup of code - but still not perfect ;-).

	* Makefile.am:  Added reconf and conversion files to distribution.

			font.h is now (correctly) font.c.

	* configure.in: Added config.h initialization and GPL notice.

	* vgrabbj.h: 	Added .h file to clean .c file.

	* other		Thanks to Damien, who helped a lot with the cleanup.

			vgrabbj on irc -> irc.openprojects.net #vgrabbj :-)

2001-06-16	0.6.4
		* Fixed a bug regarding picture setting. This caused vgrabbj to
		  pretend it'll work even though the device was not set to the
		  correct RGB24 mode. Result was a 'hanging' vgrabbj process.
		  This should help on several devices reported not to work.
		  Even if vgrabbj is unable to set the device to RGB24 it should
		  at least report this. There will be no switch to avoid this
		  setting because RGB24 is necessary for proper operation.		  
	
2001-03-27	0.6.3
		* Added parameter -w to disable vgrabbj's setting of the V4L
		  image size. This is needed for certain cams which do not
		  allow programs to set it. They have to be set at module load
		  time. Reported by St0rm and nate - thanks.

2001-03-26	0.6.2
		* Fixed a bug which may or may not occur due to uninitialized
		  variables for a system call (VIDIOCSWIN). Now, the current
		  values are read from the device first and taken for the
		  SET-call.

2001-02-11	0.6.1
		* Way too many problems with the new autoconf/make dist.
		  I really have to work on quality control here...
		  Sorry folks!
	
2001-02-10	0.6.0
		* More GNU'ification - now using ./ocnfigure to check for
		  required libs and headers. I hope I didn't forget any.
		* timestamp now widely configurable. Chose the following
		  options:
		  Position (one of upper/lower left/right or center
		  FontSize (between 1 and 100, but only 8-25 makes sense :-)
		  Border (how many pixels should be blended besides the text)
		  Blend (0-100) how much of the original image shall be
		   viewable under the timestamp
		* also rewrote the code (more to come) to modularize it more.
		  right now there's just to much code in main{}...

2001-01-29	0.5.2
		* Updated Address of Free Software Foundation, Inc., in
		  several files. No technical changes.

2001-01-06	0.5.1
		* Unfortunately, the timestamper-code was enable by default
		  and there was no chance to stop it without changing the 
		  source. Fixed.
		  Drawback:
		  If you want the timestamper enabled, you have to tell
		  vgrabbj on the command-line!

2000-12-16	0.5.0
		* And another one: freetype was initialized multiple times,
		  posing the risk of segmentation fault. fixed. Code should
		  be pretty clean now (?)
		* Changed sort of CHANGES (newest now on top)

2000-12-12	0.4.4
		* Thanks to Zsolt I fount a little bug - Freetype font was
		  closed even though it even might not have been open.

2000-09-18	0.4.3a
		* 0.4.2 only a sleep was inserted to accomodate unloaded
		  modules at that time. This didn't really work, so ensure
		  all modules are loaded BEFORE you start vgrabbj.
		* 0.4.3a is a REAL alpha. Be prepared that it may mess up
		  your machine - up to total frozen state!!
		  Even though I run this version here on my machine, it
		  may mess up yours!
		* changed the handling of startup for timestamp. Now the
		  Font-Engine is started during commandline evaluation.
		* Font-Engine is never shut down via TT_Done_Engine - I
		  only get segmentation faults with that function :-(.
		  But it doesn't really matter:
		  If run in daemon mode, you need the engine anyway,
		  If run standalone, it should be freed when the program
		  terminates...
		  This should solve a bug reported with a totally frozen
		  machine, I hope.
		* Same applies for "Face_Open" from the included font.c...
		* Added a lot of simple debug-code. Enable this via Makefile
		  DEBUG=1 (and recompile)

2000-07-05	0.4.1
		* added five seconds delay if started as daemon to accomodate
		  time to load up the driver during boot of the machine.
		  This is not really needed if vgrabbj is started after the
		  appropriate driver modules have been loaded or the drivers
		  are compiled into the kernel. Maybe this fixes another bug
		  reported (where no picture-output took place at all).

2000-05-22	0.4.1
		* added ability to size image to sqcif, qsif, qcif, sif, cif,
		  vga. Still only rgb24
		* removed wrong syslog-reporting in rare condition when not
		  running as daemon

		  
2000-05-02	0.4.0 (finally released 2000-05-11 after 9 days continous run)
		* Argument check realized, I hope it works but I also KNOW
		  that not every error in the command line is found.
		* filenames, numeric options are definetly checked for
		  existence and sense, respectively; if a value makes no sense
		  at all (e.g. quality=-1, sleeptime=-1) or a filename/device
		  is not accessible vgrabbj ends with a rc!=0
		* realized real daemon mode (fork, child process)
		* removed option -n (not necessary in daemon mode)
		  
2000-04-30	0.3.4
		* Found a missing fclose()... Hope this solves the problem.
		  I returned back to the 0.3.1 source.

2000-04-29	0.3.0
		* Implemented PNG output format.
		* Added timestamp (try it by just saying "-e" on the
		  commandline). This is fully configurable to your system
		  needs (except the timestamp is always in the upper right
		  of the image). Thanks to rasca for the font.c code and
		  the example in w3cam
		* Fixed a bug in errorhandling while opening input device.
		  It didn't even try to open the input-device again if an
		  error was encountered...
		* In loop-mode all err/messages are now directed to syslog.
		  You'll get a whole bunch of msg if an error is encountered,
		  but so what...
		* Increased the usleep time to a quarter second. Gives much
		  less msg to syslog serves my needs better :-)
		Still missing: argument check

		0.3.1
		* Added -b to disable vgrabbj's own attempt to adjust the
		  brightness of the picture. Thanks to Ari Pollak

		0.3.2
		* If you encounter a problem regarding freetype and have a
		  recent lib installed, please change the following:
		  #include <freetype.h>
		  to
		  #include <freetype/freetype.h>
		  (Thanks to Ari Pollak)
		* Serious Bug regarding the loop-mode again. Because I changed
		  the logic vgrabbj had serious problems opening the device
		  after about 1:30h. I changed the logic back and hope it'll
		  work. (currently test-run at my machine)
		  Therefore, not yet released.
		  
	    	0.3.3 (released)
		* Still the same problem - after 1.5h access to /dev/video is
		  no longer possible.
		  Now I change the code so /dev/video will be kept open ALL
		  the time vgrabbj is running. I hope this helps...
		  I guess it is/was a big logical bug somewhere inside.
		  /dev/video is open all the time now, this speeds operation
		  a little bit, but on the other hand no other application
		  can use /dev/video while vgrabb is running.
		  The code is a mess right now and I will further investigate
		  the problem.
2000-04-26	0.2.0 (first development release)
		* OK, redesigned the structure of the source to better
		  accomodate possible failures during loop-running (it
		  shouldn't just shut down while experiencing some problems).
		  It's not optimal because it just keeps trying to do the
		  things - but included a couple of small usleeps()
		* Added option -s <device>. Get all information about the
		  video-device
		* Still missing: errorhandler, syslogger, argumentcheck

2000-04-23	0.0.1
		First version. Just tested with Philips Vesta Pro Input
		and output to a predefined jpg file. Never got released...
		
		0.1.0
		Finally, I made some enhancements, based on the vidcat
		utility. Now you can
		- change the quality
		- run a forever loop with optional (sleep) between reads
		- define an output filename (still only jpg)
		- define an input video stream
		- get the help screen
		Be aware, this is still kind of alpha code because, again,
		I'm not a programmer...
		First release.
		
		0.1.1
		Wow, what a memory leak. If used in loop-mode, you loose about
		(width*height*3) bytes of memory with each image drawn...
		To clean this, I had to find another memory related bug...