A hello to everyone who's requested beta's of SPHINX2.6, I never
knew it was so popular.

Since most of you are using 2.5y I'll run down the major changes:

(a note to WB1.3 users: SPHINX may have the 2.0 embossed look, but
it still works under 1.3. However I got 2.0 a few weeks ago, so
this isn't going to last forever.)


1)First, the config window is finished! Everything can be changed on
the fly, and configs can be saved and loaded at any time.

Also there is an internal "default" config, so there is no
SPHINXConfig program any more. Instead when you start SPHINX the
first time it will complain that it found no config file (NCfg).
Just press [OK]. Then once the SPHINX window appears, press
RightAmiga-C, edit the config to be how you like it, and then press
[Save Config]. If you have the arp.library around you'll get a file
requester, if not the filename s:SPHINX.config is used.

The 2.5y config files cannot be loaded by 2.6x_April19th, so if you
have a special config you love very much you'll have to copy it by
hand into the config window.


2)Second some 2.0 parameters are supported: CLOSE (and it always
closes the window---if opencount is non-zero it closes the window,
frees the saved text, and waits around for the requisite number of
close packets.---the close should be handled.), INACTIVE, WAIT,
AUTO, BACKDROP, NODRAG, NOBORDER, NOSIZE, and NODEPTH. Only NOCLOSE
(which doesn't need support right now, since the default is no-close
gg), SMART & SIMPLE, because simple_refresh windows are still slower
than smart_refresh, WINDOW and SCREEN are not supported yet.

In addition there are two sphinx specific parameters:
SPHINXNOGADGETS, which removes the sphinx gadgets (iconify, scroll
and undo) from the window, and SPHINXCONFIG, which lets you specify
the config file to use. The way this works is the next "parameter";
i.e. the next thing between "/"s, is considered a filename, and that
file is searched for in the s: directory. It doesn't allow you to
specify the whole path because I don't see a simple way of encoding
a filepath+name into something which is already a filepath+name
without dos choking on it. So you'd use this this way:

  SPHINX:////WdTitle/SPHINXCONFIG/SPHINX.dissassembler.config

and the window would use the config found in the file

  s:SPHINX.dissassembler.config

(you can put the SPHINXCONFIG parameter anywhere in the list of
parameters after the window title---just make sure that it is
followed immeadiately by the config filename.

Because you can specify the config at sphinx invocation time there
is no point in having multiple mounts of SPHINX: devices any more,
and so all the (somewhat illegal) code which played with the
seglist in order to unload the unneeded code is gone.


3)Thirdly the memory interaction has been changed. SPHINX used to
use AllocMem every time it needed to store something, which was fine
except that it lead to fragmented free memory rather quickly because
the shell would load in a command, the the command would output
something, so sphinx would save that, then the command would end,
and now there was a hole the size of the command in the free mem
list.

Now sphinx allocates memory in large chunks and uses those up, using
an internal memory handler to "allocate" small pieces out of the
chunks, before it goes back to AllocMem for more. The size of the
chunks sphinx takes from AllocMem can be set by the Memory Chunk
Size field in the [Buffer Size] config window pannel.



Bugs/problems in this beta:

(for 2.0 users) If the default font is not 8x8 the config window
will not realize it, and so the text will be drawn all over the
place. Workaround: set the Screen Font in the Font prefs to be an
8x8 font.

The Ctrl-C,D,E,F doesn't seem to work all the time with the 2.0
Shell. I don't know why yet, so I can't help you here.

The top line of the iconify gadget's imagery dissapears under the
depth-gadget under 2.0.

The cursor-left-n-chars, and cr<>lf routines are not ready yet, so
dearchivers, format, diskcopy etc... will still spew out many lines
of uninterresting numbers. I've set my ENV:LhAOPTS to include "-n"
so I get no byte progress indicator, which is a temporary
workaround. I know how I'm going to do it, I just have to
finish the code.

Sphinx does not have a RAW_MODE. This means that if you are using it
with a "smart" shell which tries to do its own window handling stuff
you have to tell the shell to dummy-up and treat the window as a
file (the C= shell does not need any changes--it's just dumb enough
to work, however WShell needs to be told to dummy-up; I'm told by
people who have it that this can be done, but I don't know the
option myself. Csh also is a "smart" shell and tries to do fancy
things. If it has a dummy-up option, then try it, else sphinx
doesn't work nicely with csh yet.)

The "cycle" gadgets at the bottom of the config window do not cycle
backwards when you shift-click on them. (It's not in the 2.0 user
manual, but it is in the 2.0 programmer's style guide: cycle gadgets
are supposed to cycle backwards when you shift-click on them.)

The bug in the scrolling using the scroll gg is still there. Sigh.



And now, the bug fixes and minor changes (taken from the forever
expanding Changes.doc):

36)Fixed bug error recovery code for not finding configfile upon
   startup (was freeing wrong mem).

37)Fixed typo in read_from_clipboard routine which lead the
   buffer length to be wrong, and so was eating 8 bytes every time
   the clipboard was read.

38)Changed SRA-L: now if you are in unlocked-literal state you go to
   locked- literal state, instead of to no-literal.

42)Changed FF line across the window so that it has an embossed
   look, and really doesn't look like a line of `ญญญญญญญญ'.

44)Fixed bug! If many SPHINXs were all having the spot-10 image
   visible, then some could have the parsing image visible too, even
   though they weren't parsing. I fixed this, and then removed the
   spot-10 image altogether, anyhow.

46)Fixed bug in the recovery code used when the iconized window
   cannot be reopened (after an attempt at opening the main window
   has failed)

47)Changed the look and the positioning of the main wd's gadgets.
   Now they come out better positioned under 2.0, and they have the
   2.0 embossed look.

48)Dropped the ClipOneTime clip modes, making then all "locked", to
   use the old vocabulary. The imagery is of the old one-time clip
   modes.

50)WdAg gurucode was not handled correctly (in fact, you didn't even
   see a WdAg guru, but rather some other garbage).

51)DiskInfo packets were not being sent back with res1 set to TRUE
   (-1). Fixed

52)Fixed two bugs in the low level drawing code: one happened when
   non-blocking typeahead was in use, and the upperleft character
   was inside the inputsave, and not the first character of the
   inputsave; the other happened when the half window scroll was
   enabled. Both did not change all internal variables correctly.

53)Fixed bug in Lie About Window toggle wherein the lying did not
   actually happen until the next time the window was resized. Now
   lying starts and starts with the toggle.

56)RA-V inserts from clipboard, unit 0, same as RA-I

57)Made the Ctrl-C/D/E/F code less paranoid. It will actually signal
   someone more often now. (apparently not enough for 2.0, though)



Now the obvious question is "Why is this an April 19th version when
it's May 13th? Well I'm graduating May 17th (physics: anyone want a
bilingual (french-english) physics major who's also not too bad at
math and computer-science? good with TeX, Amigas, cats, children and
small mesons), and so I've spent this last month finishing up
things. Also I started rewriting the config window so that the
pannels were replaced by having all the gadgets float around in a
"gadget-space", where they use whatever font you want, and move
around, in logical groupings, as you resize the window, in order to
best use the space available, and that will take a week or more of
work to finish up & debug.

I don't know what's going to happen to my signon
(nsd20463@uxa.cso.uiuc.edu) after May 16th since the site itself is
going to be changed into a NeXT lab, so the uxa machine might
dissappear, or they might leave it running in a corner. I'll know in
4 days, I guess.

Also you may distribute this version anywhere you wish within the
limits imposed by the opening message, but remember that it is a
beta, so don't go wild. (but putting it on a ftp site would be ok).

=Nicolas Dade