

   This file describes [1mPicBoot[0m, version 2.8, a program that shows an
IFF ILBM or GIF picture during boot.

Introduction
************

   Have you removed all output in your 2.0+ startup, and only see a
black screen during boot? Wouldn't it be nice to have a picture
instead? A picture that disappeared when the [1mWorkbench[0m screen opened?

   If so, [1mPicBoot[0m is certainly a program for you. What it will do is
to read any IFF ILBM or GIF file containing a picture, and show it. As
soon as the [1mWorkbench[0m screen appears (or you press any mouse-button),
the picture will go away.

   Features:

   *  Fast picture unpacking, using highly optimized assembler. The
     entire picture is read into memory and then unpacked. This applies
     both to the IFF and the GIF unpacker.

   *  Optional auto-detaching; the picture is loaded as fast as
     possible, with minimal memory fragmentation.

   *  A picture can be shown a user-specified time after the
     [1mWorkbench[0m screen opens (see DELAY and PATCH).

   *  Extremely flexible argument parser.

   *  Random select among any number of pictures, in several different
     ways.

   *  Force a certain display mode for a picture, even if saved with
     another (can be selected on a picture by picture basis).

   *  IFF ILBM pictures may be packed with [1mXpk[0m.

   *  The comment field of a file may be used to specify options.

   *  Optional screen centering (horisontally) and border blanking.

   *  Optional screen fading (in various forms). Uses the increased
     palette range in AA (24 bits).

   *  Should work with most "Intuition emulators" for gfx-cards.

   *  Builtin [1mFixJump[0m, by courtesy of Stefan Sommerfeld.

   *  BORDERBLANK option.

System requirements
*******************

   Apart from OS 2.04+, [1mPicBoot[0m doesn't require any special
libraries. The only non-ROM library required is [1miffparse.library[0m
(which normally resides in [1mLibs:[0m).

       [1mPicBoot[0m have full support for OS 3.0+ and AA graphics. It
should even work with gfx-cards that have an "Intuition emulator".

Legal information
*****************

   This program is freeware. You may copy and use this program freely,
as long as the following conditions are met:

   *  All files are copied in an unmodified state. If additional
     information is needed, place it in a separate file. Preferably
     redistribute in the original archive form ([1m*.Lha[0m).

     Exception: So called BBS ads may [3mnot[0m be added!

   *  The copying is done on a non-commercial and non-profit basis
     only. A copy fee to cover media costs, postage etc. may be
     charged. This fee may not exceed the fee to obtain an AmigaLibDisk
     from Fred Fish.

   *  The copier/spreader is not claiming the Copyright © of this
     program.

   Any exceptions from these restrictions requires written permission
from the author, Magnus Holmgren (see Author contact).

Disclaimer
==========

   Magnus Holmgren neither assume nor accept any responsibility for the
use or misuse of these programs. He also will not be held liable for
damages or any compensation due to loss of profit or any other damages
arising out of the use, or inability to use these programs.

   Magnus Holmgren will not be liable for any damage arising from the
failure of these programs to perform as described, or any destruction
of other programs or data residing on a system attempting to run the
programs. While he know of no damaging errors, the user of these
programs uses it at his or her own risk.

Usage
*****

   To activate [1mPicBoot[0m, add a line to your [1mS:Startup-Sequence[0m,
looking something like this:

     PicBoot Pics:Hi-res/Calvin01.Pic DETACH

Or, if you have a list of files in "Work:Text/PicList":

     PicBoot Work:Text/PicList LIST DETACH

   Or, if your drawer [1mPics:BootPics[0m contains some pictures:

     PicBoot Pics:BootPics/#?.(Pic|Gif) DETACH

   This line should be located near the beginning in the
[1mS:Startup-Sequence[0m (no point in placing it near the [1mLoadWB[0m
command, is it? :), but keep it after [1mSetPatch[0m. [1mPicBoot[0m will only
output any text if it fails, so don't re-direct its output. Note however
that if you place [1mPicBoot[0m [3mbefore[0m any additional monitors are
installed, you'll be stuck with the default.monitor for showing your
picture. The DEFAULT switch may be of help here.

   Make sure no program makes any output in the CLI window, since then
the [1mWorkbench[0m screen will open with a boring CLI-window instead...

   Options:

FILES
=====

   This is the only required argument. Here you specify the name of the
picture you want to view. You may enter several files here, in which
case [1mPicBoot[0m will select one of them randomly, and show that one.

   The name(s) can also be the name of an ASCII file containing a
filename list if you specified the LIST option.

   The name(s) can also be the name of a drawer, in which case
[1mPicBoot[0m will randomly select one of the files in this drawer. To
use a pattern during this scanning, simply enter the pattern like it
had been the name of a file in the drawer. Example:

     Work:Pics/#?.gif

which would make [1mPicBoot[0m select a file ending in [1m.gif[0m in the drawer
[1mWork:Pics[0m

   If the comment field of the selected listfile or picture starts with
"*PicBoot*: " (case sensitive), then the rest of the comment is taken
to be arguments, like those in a LIST file.

   You may freely mix picture and drawer names. Listfiles can only be
mixed with the other two ones if the comment contains the LIST switch.
In that case, the LIST argument should not be used on the command line
(or in a list file). Ofcourse you can random select among list files
with the "drawer scanner" if you like.. :)

MODEID
======

   Short form: M

   NOTE: This argument is mainly for the more "advanced" user.

   This argument should be a decimal number specifying which screen
mode to use.  It basically replaces the so called CAMG hunk in an ILBM
file (since it contains which screen mode to use). Thus, you must
select mode with care, or else the picture will look like trash
(nothing more serious can happen. I hope! :). When showing GIF files,
it will override the internal "best mode" routines (which aren't good
at all. But I haven't bothered to add code to make them better.. :).

   To make it easier for you to find out which display mode id to use,
there is a small program called [1mGetModeID[0m included, which uses the
[1mReqTools[0m or [1mAsl[0m screenmode requester. Simply select the display
mode you want, and it will print out the number you should use here.
See GetModeID.

   The mode id will be passed through the same validity checking as a
normal so called CAMG chunk, so [1mPicBoot[0m should handle bad values
properly (although I haven't tested this much.. :).

PASSWORD
========

   Short form: PW

   Here you can specify the password for an [1mXpk[0m-encrypted IFF ILBM
picture.

LIST
====

   Short form: L

   If this switch is specified, [1mPicBoot[0m will interpret the files in
the FILES argument as names of files containing a list of pictures (or
rather, argument lines). [1mPicBoot[0m will then randomly select one of the
lines in the selected file, and process it almost like a normal argument
line. The only difference is that you can't use the DETACH, DELAY,
PATCH or ACTIVATEWB arguments. These arguments may not be specified in
a listfile (no point in doing it anyway).

   The listfile is an ASCII (text) file with a simple layout. On the
first line you specify the number of argument lines in the file. This is
usually <number of lines in file>-2 (one line is occupied by the count,
and the other is the last linefeed). If this value is zero, then
[1mPicBoot[0m will exit silently. The rest of the file is simply the
argument lines to choose from. An example:

     4
     Work:Pics/Comics/Calvin02.Pic MODEID 137220
     Work:Pics/Comics/Calvin03.Gif
     Work:Text/MoreCalvins.txt LIST CENTER ON
     Work:Pics/Misc/#?.Gif

   Note that any arguments specified from the CLI, or in any previous
listfile, will be taken as the new default value. In the listfile you
may alter this default. This does not include the LIST argument
(ofcourse). It is always turned off before parsing a line.

       [3mWarning: [0mSince you may enter a new listfile within a listfile,
you can be caught in an endless loop, constantly changing (maybe to the
same) listfile. No checking for this is done. Also, since there is no
CLI-window around, you have no chance to send [1mPicBoot[0m any CTRL-C, if
[1mPicBoot[0m should happen to listen to this. You have been warned! :)

       [3mNote: [0mA line in a listfile may not be more than 512 chars, or
it will be truncated when read. This shouldn't cause any problems I
think..

CENTER
======

   Short form: C

   Possible arguments: YES, ON, NO, OFF. Default is NO.

   If this switch is on (argument is YES or ON), [1mPicBoot[0m will center
the picture. This centering should work fine for most screen modes, but
one can never know.. :) If a screen promotor is active, then
[1mPicBoot[0m can get it wrong (if the screen is opened in another mode
than [1mPicBoot[0m had asked for).

DEFAULT
=======

   Short form: DEF

   Possible arguments: YES, ON, NO, OFF. Default is NO.

   If this switch is on (argument is YES or ON), [1mPicBoot[0m will force
the picture to use the default.monitor, regardless of what was actually
stored in the picture (in the CAMG chunk). This is needed since very
early in the startup, default.monitor is the only monitor available
(e.g.  multiscan.monitor is normally not available). In the future, I
might add more types of "forcing" (e.g. force a picture to PAL, NTSC or
whatever that might be useful).

   This switch also acts on the MODEID parameter, if specified.

AUTOSCROLL
==========

   Short form: AS

   Possible arguments: YES, ON, NO, OFF. Default is NO.

   If this switch is on (argument is YES or ON), the OS 2.0+
autoscrolling of screens will be enabled.

       [3mNote:[0m During boot, this switch may make the actual display a
bit smaller than normally possible. There is nothing I can do about
that... :) You can, however, by ensuring that ENV:/IPREFS is properly
set up before [1mPicBoot[0m is started. Or you could try the VIDEOOVERSCAN
switch.

VIDEOOVERSCAN
=============

   Short form: VO

   Possible arguments: YES, ON, NO, OFF. Default is NO.

   If this switch is on (argument is YES or ON), the visible size of the
opened screen will be as large as the system can handle (assuming the
picture is large enough). Forces AUTOSCROLL to YES.

RTG
===

   Possible arguments: YES, ON, NO, OFF. Default is NO.

   If this switch is on (argument in YES or ON), then [1mPicBoot[0m will do
things a little differently, in an attempt to make it work with
gfx-cards (it have been tested with [1mPicasso II[0m). The main difference
is that the screen is opened first, and the picture is decoded into
this screen (usually the picture is decoded first, and then the screen
is opened).  Thus, you should only use this switch if the picture
should be shown with the gfx-card rather than the native [1mAmiga[0m
graphics. If the gfx-card isn't used, the picture decoding may be a
little slower.

   Oh, btw, RTG stands for ReTargetable Graphics.

       [3mNote: [0mThis feature requires OS 3.0 to work. If you don't have
OS 3.0, then this argument is ignored.

       [3mNote: [0mYou might need to specify a new MODEID in order for
[1mPicBoot[0m to use gfx-card screen.

WRITEPIXELLINE
==============

   Short form: WPL

   Possible arguments: YES, ON, NO, OFF. Default is NO.

   If this switch is on (argument is YES or ON), then [1mPicBoot[0m will use
a ROM function to convert/write the pixel data of a GIF picture.
Please note that this only have any effect if RTG have been used, and
the opened screen isn't a native Amiga screen.

   In some cases, this can make [1mPicBoot[0m faster, but in others, it can
make [1mPicBoot[0m slower. There is an explanation for this, but it is a
bit technical. If you don't understand it (or you don't know enough
about your gfx-card), then I suggest you test it a little, finding out
which is fastest for different pictures.

   Most gfx-cards have a "chunky" screen. That is, each pixel is stored
in one byte, which specifies which color (in a palette) that the pixel
have. This is different from the native [1mAmiga[0m screen that have
bitplanes (where the color number is spread over several bytes). Since
a GIF picture stores the data in a chunky format, it would be a waste
of time to first convert the chunky data to bitplane form, and then
back again, if it should be displayed in a chunky screen.

   In the case of a chunky screen, this switch can improve the speed
quite a lot, if a certain ROM function is patched by the Intuition
emulator.  This patch should simply write the chunky data directly to
the right area of the screen.

   However, not all gfx-card screens are chunky (at least the [1mPicasso
II[0m stores 2-16 color screens as bitplanes), and perhaps the "Intuition
emulator" haven't patched the above mentioned function. In those cases,
then this switch will make things slower (since the original function
(which converts the chunky data to bitplane form) is quite a bit slower
than the routines in [1mPicBoot[0m).

   To test which is fastest, use something similar to the following
commads:

     PicBoot Pics:Gifs/Island.gif RTG ON DETACH
     PicBoot Pics:Gifs/Island.gif RTG ON DETACH WPL

and measure how long time both commands took to run (note that
[1mPicBoot[0m detaches after the picture have been completely decoded),
and only use the WPL switch if it made [1mPicBoot[0m faster.

FADEIN
======

   Short form: FI

   Argument range: 1 to 4. Default is no value.

   This value specifies the speed with which the picture should fade in
when the screen is opened. HAM pictures can't be faded.

FADEOUT
=======

   Short form: FO

   Argument range: 1 to 4. Default is no value.

   This value specifies the speed with which the picture should fade
out when the screen is closed. HAM pictures can't be faded. Only useful
in combination with DELAY (otherwise the picture will be in the back,
were the fade isn't visible! :).

FADEWB
======

   Short form: FWB

   Argument range: 1 to 4. Default is no value.

   This value specifies the speed with which the [1mWorkbench[0m screen
should fade in when the picture screen have been closed. Intended to be
used in combination with the FADEOUT (and DELAY) argument(s).

       [3mNote: [0mIf you have a [1mWorkbench[0m screen with lots of colors,
this fading can cause problems, if you have any background pattern. When
the system loads these patterns, it might try to allocate some of these
colors for use by the pattern. If this happens while [1mPicBoot[0m is
actually doing the fade, it is easy to imagine that the result won't
quite be what you expect (i.e. wrong colors in the pattern). If this
happens for you, simply don't use this feature.

DELAY
=====

   Short form: DL

   A "problem" with [1mPicBoot[0m is that the [1mWorkbench[0m screen first
opens, and then processes the [1mSys:WBStartup[0m drawer, which takes a
little time. This means that the picture [1mPicBoot[0m shows disappear
before the boot is complete. To avoid this problem, the DELAY switch
can be use to specify the number of ticks (there are 50 ticks each
second) [1mPicBoot[0m will wait after the [1mWorkbench[0m screen have opened.

   However, this isn't perfect. When the [1mWorkbench[0m screen opens, the
[1mPicBoot[0m screen must be brought back to the front again. This causes a
little "flicker". To avoid this, use the PATCH parameter as well (this
feature requires OS 3.0+ to work).

   There is a special delay value, 0, which causes [1mPicBoot[0m to wait
until you either press any mouse button, or another program sends
[1mPicBoot[0m a break signal (CTRL-C). The program StopPicBoot was
written to do this.

BORDERBLANK
===========

   Short form: BB

   If this switch is specified, then the special border blanking mode
will be enabled, which can make the picture look a little better. This
only works on OS 3.0 or better.

FIXJUMP
=======

   Short form: FJ

   If this switch is specified, then [1mPicBoot[0m will try to set the
correct overscan settings before opening the picture. This should fix
the "jumping" that otherwise occures when [1mIPrefs[0m is started.

   What it does is simply to read the file
[1mEnvArc:Sys/overscan.prefs[0m, and sets the proper values.

       [3mImportant note: [0mThis is a hack, since it calls an undocumented
system function. The code seems to work well on my system (A4000/040 OS
3.1), but I don't make any guarantees.

PATCH
=====

   Short form: P

   If this switch is specified, [1mPicBoot[0m will install a patch in
[1mIntuition[0m, so that the [1mWorkbench[0m screen (or rather, any screen
opened, that explicitly doesn't say that the screen shouldn't open
behind the others) doesn't open in front of the [1mPicBoot[0m screen. This
removes the "flicker" that normally occurs when using the DELAY option.
For this option to be useful, the DELAY parameter must be used as well.

       [3mNote 1: [0mThis option only works on OS 3.0 or higher. This is
due to the OS (as far as I know), and there is nothing I can do about
it (tech note: [1mWorkbench[0m in OS 2.0x doesn't seem to call the open
screen function via the external library vector).

       [3mNote 2: [0mThis kind of patching is not a recommended thing to
do. Programs should not do temporary patches like this. However, to
avoid the flickering, there is no alternative.. :)

       [3mNote 3: [0mIn case some other program patches the same function
after [1mPicBoot[0m have installed its patch - and you don't have a program
like e.g. [1mSetMan[0m installed - then [1mPicBoot[0m will leave a small memory
allocation behind (6 bytes), to avoid any problems.

DETACH
======

   Short form: D

   If this switch is specified, [1mPicBoot[0m will detach from its calling
CLI when the picture is fully loaded and displayed. If you specify this
option, you shouldn't "Run" [1mPicBoot[0m. This option will reduce memory
fragmentation, and will ensure that the picture gets loaded quickly. I
don't think this feature will cause any problems, but I added the
switch just in case.

ACTIVATEWB
==========

   Short form: AWB

   If this argument is specified, then [1mPicBoot[0m will try to activate a
[1mWorkbench[0m window after closing the picture. This might be useful if
you are using the DELAY argument.

Known problems
**************

   I do not know of any real bugs in [1mPicBoot[0m. However, certain parts
of the program may still contain bugs. E.g.,  pictures that have a mask
bitplane (mskHasMask) are supported, but since I only have one
(compressed) picture that have a mask, there might be a bug in that
code (can't test it properly).  Please report any problems!

   Currently there is no support for SHAM, PCHG and similar pictures.
These pictures aren't that common, and I have an Amiga with
AA-graphics, so... :) Color cycling is currently ignored (I have no
need for it).

   Interlaced GIF pictures aren't supported at the moment. The
decompression of such pictures would be slower, and there are tools that
can de-interlace a picture.

       [1mPicBoot[0m doesn't remap GIF files in any way. Even if you have
ECS, GIF files can still be useful. This is because a GIF file can have
from 2 to 256 colors (inclusive). Thus, if you have a program that can
save a 16-color picture as a 16-color GIF file, there will be no
problem to view it with [1mPicBoot[0m.

       [1mPicBoot[0m doesn't make use of any "chunky to planar" hardware,
if it should happen to be installed (e.g. Aikiko). Anyone who have it,
so I can test it if I should decide implement it? :) It would be fairly
simple to do, since my own (rather fast, I might add :) chunky to
planar routines have very similar restrictions.

   The "best mode" routine used in the GIF reader isn't good at all
(this includes the ROM function in OS 3.0+! :). I suggest you use the
MODEID parameter instead (Correction: The ROM function isn't good when
there are several different monitors to choose from. If only one or two
(similar) monitors are available, then the result is usually rather
good).

   The "monitor promotion" (i.e. the DEFAULT parameter) is not very
intelligent.

   Pictures with more than 8 bitplanes are currently not supported by
[1mPicBoot[0m.

   The centering for (some?) Super72 screens doesn't work. I suspect
this is an OS-"bug" (I know that [1mPicBoot[0m calculates a reasonable
offset, which [1mIntuition[0m seems to ignore).

StopPicBoot
***********

       [1mStopPicBoot[0m is a small program that simply tells [1mPicBoot[0m to
quit, if it should happen to be in memory. This is useful in
combination with the DELAY option. If this is set to 0, [1mPicBoot[0m
expects someone to tell it when it is time to exit, and this is what
[1mStopPicBoot[0m does.

   By having [1mStopPicBoot[0m in [1mSys:WBStartup[0m, then [1mPicBoot[0m will
close its screen when the boot process almost complete (the tooltype
[1mSTARTPRI[0m should be very low (-120 or so), so that [1mStopPicBoot[0m is
started as the last program)

GetModeID
*********

       [1mGetModeID[0m is a simle program that shows a [1mAsl[0m or
[1mReqTools[0m screenmode requester, whichever is available. The program
will then print out the decimal identifier for the selected screenmode,
suitable for use together with the MODEID parameter. This program can
only be used from a [1mShell[0m. Example usage:

       PicBoot Island.Gif MODEID `GetModeID`

   This will first show a screenmode requester (if you have one, that
is), and then show the GIF-picture in the selected screenmode.

UnpackILBM
**********

       [1mUnpackILBM[0m is another simple program (at least in theory...
:). It will take any IFF ILBM picture and unpack the so called BODY
chunk in it (this is the actual image data). This means that e.g.
[1mPicBoot[0m will be able to display that image a little faster, at
least if loading it from some fast media. Or you could repack the
picture with [1mPowerPacker[0m/[1mXpk[0m, to maximize the compression (as the
compression used in IFF ILBM isn't a very efficient one. But on the
other hand, it is rather fast and simple). Example usage:

       UnpackILBM Island.Pic Island.Pic.NoComp
       UnpackILBM Island.Pic

   The first example till unpack the picture to a new one, while the
other will - via a temprary file - overwrite the original picture with
the uncompressed version.

       [3mNote:[0m I haven't tested this program [3mthat[0m much. I've converted
a few pictures, so it seems to work fine (at least when there aren't any
errors), but one can never now.. Please report any problems!

       [3mNote:[0m This program doesn't strip any information. All chunks
will remain. The picture data is only decompressed.

Author contact
**************

       [1mPicBoot[0m was written by Magnus Holmgren. If you have any
comments etc, feel free to send me a note. You can reach me via
internet on this address:

     cmh@augs.se

   Fido-net messages should go to "Magnus Holmgren",
2:204/404.6@fidonet.org.  Snail mail should reach me if you write the
following address on the envelope:

     Magnus Holmgren
     Kvarnbergsvägen 4
     S-444 47  Stenungsund
     SWEDEN

Version history
***************

Version 1.00-1.03
=================

   Ancient versions.

Version 2.0
===========

   Release date: 29 Mar 94

   *  BLACK argument removed. Not needed any more, since [1mPicBoot[0m now
     will first allocate the needed memory, decode the picture into
     this memory, and then open the screen. This makes the screen
     opening/closing a little faster too (practically instaneous on my
     A4000/040).

   *  Pictures (or rather, brushes) that were less than 16 pixels wide
     wouldn't decompress properly... :)

   *  Added support for the CMAPOK flag in the BitMapHeader.bmh_Flags
     (previously called bmh_Pad) field (if this flag is set it
     indicates that the color map contains 8 bits/color rather than 4
     bits/color).

   *  Major code cleanup. Made the program somewhat larger, but... :)

   *  GIF support added. Should be a little faster than [1mPPShow[0m.. :)

   *  The MODEID argument wasn't properly "passed on" to any following
     listfile(s).

   *  Rewrote [1mrtGetModeID[0m into [1mGetModeID[0m, that first checks for
     [1mAsl[0m, and then tries with [1mReqTools[0m before giving up. This new
     version is in C, compiled with DICE, without any startup code, and
     is fully residentable. :) See GetModeID.

   *  Included [1mUnpackILBM[0m, that takes any IFF ILBM file (with a BODY
     chunk, i.e. a normal picture) and writes it with an uncompressed
     BODY instead. Written upon user request. See UnpackILBM.

Version 2.1
===========

   Release date: 14 May 94

   *  The DEFAULT parameter didn't do anything. Fixed

   *  [1mUpackILBM[0m and [1mGetModeID[0m updated a little. Version string
     added, recompiled with [1mDICE[0m 3.0 and some other minor changes.

Version 2.2
===========

   Release date: 12 Jul 94

   *  If the listfile was too short (i.e. not enough number of lines in
     it), [1mPicBoot[0m would crash.

   *  Made the detaching code more system friendly. I hope this will
     fix the problems a few users have had.

   *  A few minor bugs fixed + some minor optimizations...

   *  If the number on the first line in the listfile is 0, then
     [1mPicBoot[0m will exit silently. Now why did I add this... >;)

   *  Improved the random number algorithm.

   *  Added the DELAY parameter.

   *  Added the PATCH parameter.

   *  Tweeked the GIF-unpacker a little. Found yet another Macro68
     (V3.170) bug while doing that.. :/ (Watch out for bra.l to other
     sections/modules when generating code for the 68020+. The branch
     target is [3mnot [0mcorrect. :)

Version 2.3
===========

   Release date: 30 Aug 94

   *  Rewrote startup code and argument parser in C, for easier
     maintainance (and to simplify the implementation of some of the
     features below).

   *  You can now also specify a directory (with optional pattern
     matching), and [1mPicBoot[0m will randomly select among the files
     found. As usual, you can use this feature whereever [1mPicBoot[0m used
     to expect a file name.

   *  If the comment field of a file that [1mPicBoot[0m will read (i.e. a
     list file or a picture) starts with the string "*PicBoot*: " (case
     sensitive), then the rest of the comment is assumed to be
     arguments, to be parsed like they had been found in a list file.

   *  [1mUnpackILBM[0m will not delete the temp file if it couldn't be
     renamed to the original.

   *  Removed a piece of debug code in the [1mOpenScreen()[0m patch (it
     flashed the screen). Harmless, but annoying.. :)

Version 2.4
===========

   Release date: 11 Oct 94

   *  Added the arguments FADEIN, FADEOUT and FADEWB, to make various
     color fades when the screen is opened/closed.

   *  Modified the random routines again (changed the seed source). I
     hope it works better now.. :)

   *  If DETACH was used and a relative filename was used, then the file
     wasn't found (except in some cases).

   *  Using PATCH in combination with DELAY on pre-OS 3.0 systems
     caused a "Not enough memory" message to be printed, and the picture
     wasn't showed (the argument should silently be ignored).

   *  Some error messages lacked a final linefeed char (an autodoc was
     a bit misleading.. :).

   *  The DELAY was incorrectly interpreted as seconds, and not as
     ticks.

   *  Closing the picture before a DELAY timeout had expired caused a
     crash.

Version 2.5
===========

   Release date: 13 Oct 94

     The [1mWorkbench[0m screen wasn't "unlocked" in some cases (when either
     FADEWB or ACTIVATEWB had been used).

Version 2.6
===========

   Release date: 30 Nov 94

   *  Added RTG switch, which will make [1mPicBoot[0m attempt to be more
     compatible with various "Intuition emulators" for gfx-cards. Don't
     use this switch if you don't need to; it will cause the screen to
     be opened before the rendering (possibly making it slower).

     Also added WRITEPIXELLINE, which may improve the speed of the GIF
     reader sometimes, when using the RTG-mode.

     Thanks go to Roger Westerlund, who made me actually try to make
     [1mPicBoot[0m RTG friendly, and also did all the needed testing (I
     don't have any gfx-card :/).

   *  Changed the screenmode id "validator". Needed for better RTG
     compatibility.

   *  Some fade related code was a bit broken, causing problems with
     e.g.  HAM-pictures.

   *  [1mUnpackILBM[0m is now able to decompress via a temporary file if the
     file and the current directory are on different volumes. Also
     improved some error reporting.

   *  [1mGetModeID[0m now filters out some useless modes (the dual payfield
     (DPF) ones). Also, the [1mReqTools[0m screen mode requester didn't
     show "non standard" modes (such as HAM and EHB).

Version 2.7
===========

   Release date: 17 Jul 95

   Well, I started working on this version a couple of months ago
(making it almost finished), but then I forgot all about it.. :)

   *  [1mPicBoot[0m didn't unlock the drawer after a directory scanning.

   *  [1mPicBoot[0m could hang if a DELAY was requested, and the picture was
     closed before the [1mWorkbench[0m window had opened.

   *  [1mPicBoot[0m didn't handle names with spaces in them very well..

   *  Fixed some bugs in the IFF reader that could cause crashes.

   *  Added support for [1mXpk[0m packed/encrypted IFF files.

   *  Made [1mGetModeID[0m a bit smaller. Didn't think twice before adding
     the mode filter.. :)

   *  [1mUnpackILBM[0m would report an error and delete the file if the same
     file was specified in both FROM and TO. Now only the error is
     reported.

Version 2.8
===========

   Release date: 02 Nov 95

   *  Added the FIXJUMP switch. Removes the "jumping" that usually
     occures when [1mIPrefs[0m gets loaded. Thanks to Stefan Sommerfeld for
     sending me the sources to his program FixJump (which this
     implementation is based on).  Note that this is a hack, and may
     not work on all systems.

   *  Made a few changes, in an attempt to fix the fading problems a
     couple of users have reported.

   *  Added the BORDERBLANK switch.

