==========================
Changelog for JPEGView 3.3
==========================

----
Bugs
----

(fc1) Removed the unnecessary use of the Time Manager to measure
image drawing time, simply counting ticks instead.  The alleviates
a number of system-level crashes that would occur if JPEGView
crashed during decompression.

(b5) The "Startup Screen" option in the Save As file format menu has
been changed to read "PICT Resource" so that users don't get confused
as to why their compressed startup screens won't display properly.

(b5) Small image files (<600 bytes) would be incorrectly reported as
invalid; this has been fixed.

(b5) Choosing to crop the icon only when saving, and using proportional
icons (dog-ear or no) would cause the icon cropping not to kick in.

(b5) The rectangle the icons are drawn into is now inset by 1 pixel, so
that the 1-pixel black border around the icon doesn't overwrite so much
of the actual icon image.

(b4) An unresolved alias found while searching a slide show directory
would dump JPEGView into an infinite loop or cause it to abort the
scan of the current folder.  This situation should be remedied.

(b4) The Inited flag is now cleared when saving images; this seems to
force the Finder to update custom icons better.

(b3) The JPEGView JFIF Preview has been recompiled yet again.  It is now
a true fat component, created according to the Component Manager 3.0 specs.
This should hopefully solve the occasional weird problems I was getting
with JFIF previews.

(b3) Slide shows with virtual memory on no longer allow offscreen bitmaps
to be created in temporary memory.  This results in more paging than with
version 3.2, but it is quite tolerable -- at least on my PM6100/60av!

(b2) Certain color startup screens would not be displayed properly if there
was also a black & white version stored in the data fork.  JPEGView now
checks for a color screen, and ignores the black & white one if it is
found.

(b2) Selecting the Desktop for a slide show scan finally seems to get it
right (I think!)  Also, I put in a check during the slide show scan so
that infinitely-recursive alias resolution can no longer get you into
trouble!

(b2) The dog-ears on large icons are now slightly bigger, and should be
comparable to dog-ears from other apps.

(b1) Revealing a hidden part of a window during two-pass color reduction
of the image in that window would abort the quantization. This was
particularly bad when doing the color reduction during a save.

(b1) Somehow, bundle bits in the Finder were not set on the version
3.2.1 I released; this is now done explicitly during the build.  In
addition to the bundle bit, the shared bit is now set so multiple copies
of JPEGView can be run from a single copy on a server.

(b1) There was a conflict between the "always require bitmaps"
preference option and the "no bitmaps for uncompressed images" option --
that is, JPEGView would complain it didn't have enough memory to make a
bitmap for an uncompressed image, because you told it not to :-)

(b1) If you had selected a slide show folder that was on a removable
medium, JPEGView would prompt you to insert that disk everytime you
started up.  JPEGView now does a less-intrusive check for the existence
of the last slide show folder you chose, and if it doesn't find it, it
resets the path to point to the Desktop.

(b1) Some people were detecting a write to memory location 0 when
quitting JPEGView.

(b1) The QuickTime VM patch would sometimes be incorrectly installed on
machines that didn't need it, resulting in a significant slowdown.

(b1) When scaling really large images (>1000 pixels in both dimensions)
down to icon size, the very high quality drawing (which I use for making
the icons) would sometimes crap out and not draw anything.  This was due
to the fact that when a large number of pixels in the source image
contribute to a single pixel in the destination, each pixel's contribution
could end up rounding down to 0, since I only kept track of things to
1/1024th of a pixel.  I have since modified the very high quality drawing
to retain resolution down to 1/65536th of a pixel, which appears to solve
this problem for the most part.

(b1) The Select Screen Area command would not be available if an image
that would have fit in a full screen window was instead being displayed
scaled in a normal window.

(b1) Printing images to the LaserWriter 8.1.1 driver seemed to go awry
somewhere along the line.  This appears to be gone with the new
implementation of the QuickTime VM patch.

------------
Interim bugs
------------

(fc1, introduced in b1) Dragging a selection so the cursor was outside
the image rectangle would cause an unimplemented trap error, because I
was failing to check for the Drag Manager ahead of time.

(b5, introduced in b3) The JPEGView JFIF Preview wasn't displaying
previews properly for files with type 'JPEG'.  Files with type 'JFIF'
were working all right, though.

(b2, introduced in b1) Fixed high quality drawing on the PowerPC.

--------
Features
--------

(fc1) The comments and statistics text can now be dragged out of
JPEGView and into another application.

(fc1) Progress dialogs and displays are updated more frequently now.

(b5) A variant of the Independent JPEG Group's JPEG decompression code
has been added as an alternative to using QuickTime for displaying
JFIF images.  The IJG code is much more robust at error handling than
the QuickTime code, but is considerably slower on 680x0 machines.  For
this reason, QuickTime is still the default for 680x0 machines, while
the IJG code is used automatically on PowerPC machines.  If QuickTime
is not installed in the system, then the IJG code will automatically be
used on either system.  User control over which code to use can be
found in the Miscellany section of the Preferences, at the bottom.

(b3) A new preference has been added: icon styles, under the "Files"
section of the preferences.  This allows you to choose one of four preview
icon styles: square or proportional (a la Photoshop), with or without
"dog ears".  Hopefully that will be enough to please most everyone.

(b3) The scale AppleScript property of an image now accepts integers
from 1-100 instead of fixed point numbers which AppleScript doesn't know
how to deal with.

(b2) The File Format popup menu is no longer displayed if you are saving
a non-JPEG image.  Hopefully this will stop people asking me for an
"unlocked version" which supports this. :-)

(b2) Added a new mini-popup menu to the bottom of the save dialog which
lets you do something with the currently selected area of the image
you're saving.  The options are: do nothing, use selected area to
specify the icon cropping (it still has to be square for now, though),
or use selected area to specify image cropping.

(b1) After toying with a version of what I considered to be "low" quality
drawing (i.e., really fast but pretty crummy), I discovered that my low
quality was actually equivalent to my high quality in 95% of the cases. 
With that knowledge in hand, I rewrote high quality drawing to do "low"
quality drawing instead, and came up with a good 20-75% speed increase
(you'll a see 5-25% increase overall, including decompression) with very
little additional quality loss.  The high quality dithering is still
included, so it will continue to be labelled "high" quality, despite the
fact that the scaling quality isn't quite on par with Color QuickDraw's.

(b1) Due to the above change, the default drawing quality is now set to
"very high".

(b1) Display times in the statistics floating window are now given to
the 1/100th of a second.  I was mainly using this for PowerPC
benchmarking, and may decide not to keep it in for good.

(b1) On a drag-aware system (that is, with the Macintosh Drag and Drop
extension *and* a drag-aware Finder -- e.g., a Mac with system 7.1.2 and
the extension), you can now drag any folder or disk icon from the Finder
into the slide show options window to set the slide show folder (very
cool :-).

(b1) Dragging a selection outside of its window will, on Drag-aware
systems, allow you to drag and drop portions of an image into other
apps, or into clippings files in the Finder.

(b1) Updated the operation and use of the internal QuickTime VM patch, so
that it now allows the use of up to 512k of temporary memory for image
decompression.  Its effects are also now disabled before giving time
to other applications so that extensions that need to use temporary
memory can still do their thing.

(b1) Opening PICTs at other than 72dpi now expands them to the full
resolution.

(b1) The "marching ants" used to indicate a selection have been redone
completely, and are no longer rendered invisible in medium-gray areas of
images.

(b1) Once you've cropped an image, you are now allowed to save it; when
you do this, however, what you are actually saving is not the cropped
image alone, but rather the entire original image plus a 'RECT'
resource which specifies the initial cropping.  Thus, images may be
cropped for better viewing, but may always be recovered.


=================================
Remaining "bugs" for JPEGView 3.3
=================================

----
Bugs
----

(?) Statistics for cropped images saved with previous versions are
messed up.

(?) The JPEGView Convert To PICT script saves palettes for grayscale
JFIFs.

(?) JPEGView's progress window can poke through the Finder's window when
running in the background.

(?) Dragging windows from 1-bit screens to deeper screens doesn't always
force a complete redraw?

(?) I've had two reports of bugs when running a second slide show during
the same JPEGView session.  As far as I can tell, it's linked to selecting
a different folder for the second show, but I can't reproduce the problem
at all.

--------
Features
--------

(?) Modify the way the slide show works internally to read any number of
aliases from the specified file.  This is in anticipation of the future,
where we will store slide show information in a separate file, allowing
multiple files and/or folders.  Although we will put in the code to
handle multiple entries and external files, 3.2.2 will always read from
the Preferences file, and will always find only one folder specified. 
Consider it a step forward.

(future) More detailed preferences for specifying dithering, display
quality, and color reduction for each image depth.  Something like
this:

When using...   Reduce
                colors?   Dither?    Quality?
   16 colors      x         x       normal    v
  256 colors      x         x       very high v
   Thousands                x       very high v
    Millions                        very high v