home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
AmigActive 6
/
AACD06.ISO
/
AACD
/
System
/
FBlit
/
ReadMe.doc
< prev
next >
Wrap
Text File
|
1999-10-02
|
6KB
|
180 lines
FBlit 3.51a
-----------
'FBlit' and it's associated files are © Stephen Brookes 1997-99
Blah
----
This software is experimental, incomplete and fundamentaly dangerous, so
don't use it! I will take no responsibility for any undesirable effects
resulting from the use of any software or information in this archive.
Distribution
------------
This archive is freely distributable. The latest version is available from
<http://www.tpec.u-net.com>. Please ensure that you have the latest release
from that address before reporting bugs.
Requirements
------------
020+ CPU
v39+ (OS3.0)
Fast RAM
The GUI also requires MUI.
Installation
------------
Copy 'FBlit' and 'FBlitGUI' to 'C:'.
Copy 'fblit.library' to 'Libs:'.
Or, copy the whole drawer anywhere you like...
Add the following line to 'S:startup-sequence' before anything else that
may patch gfx functions, and after anything that may patch memory
functions. Just before 'BindDrivers' is usually ok.
C:FBlit
(or <path to FBlit drawer>FBlit)
If updating a previous FBlit installation, you should also 'delete
ENVARC:fblit.cfg' to restore the default configuration before rebooting.
Notes:
FBlit should *not* be launched with 'run' (or 'asyncrun', 'runback' or
anything like that)! Doing so may cause patch order problems with CGX,
NewIcons etc.
If FBlit's OSTLPatch is installed (it is by default), you may have to
launch FBlit before any bootpic software, or use PatchControl (or equiv'),
but FBlit should not be installed before ENV: & ENVARC: have been assigned.
Do not try to rename the FBlit files ;) If you change the name of
'fblit.library', 'FBlit' or 'FBlitGUI', things may not function as
expected!
A related point. FBlit currently generates *no* error output. If it fails,
it will do so silently (or with a guru).
Usage
-----
The GUI is invoked by launching FBlit a second time.
What's this?
------------
This is a public test version of FBlit. There is no new documentation yet
so if you don't know what this is about, please download FBlit v2.45a and
have a look at the doc's in that archive, but be aware that much of the
detailed information in that is not applicable to current versions.
These versions are intended to run with FAllocBitMap set to 'exclude' mode
(in the GUI), so you may want to try that ('include' mode is still the
default), but it is a little risky so be prepared for trouble! Exclude mode
is not really recommended unless you have a fair grasp of what FBlit
actually does and the possible side-effects (also, a recent backup of any
valuable data is a good idea) ...re-read the 'Blah' section above.
For those I haven't put off yet... There are a couple of known issues with
exclude mode. Line related corruption problems, caused by the incomplete
FDraw() patch are more common, but these are harmless and, for the most
part, bearable. Voyager's task 'V' (up to, at this time, V³pre2) will go
down in flames if promoted, when displaying scaled images, so it is
advisable to add 'V' to the 'exclude task' list if you use Voyager. I have
also had reports of corruption (safe) with DOpus' internal animation
viewer.
More blah
---------
The current temporary FDraw() patch is rather limited in that it won't do
patterned, or complement mode lines. It is also inaccurate and slow. None
of these things cause much trouble in day to day use, but they can create
(safe) graphics corruption under certain circumstances! (especially if it
is set to process all chip data)
3.36...
FText() has gone, hopefully without any fallout. You can still accelerate
text functions to some extent by setting FBltTemplate() to process chip
data.
3.40...
Hmmm. The rules for FAreaEnd()'s discard operation have changed
(VisualPrefs etc. should now work).
FDraw() and FBltPattern() are still incomplete :/
OSTLPatch is a new patch. It checks any custom bitmap supplied to
OpenScreenTagList(), and if this bitmap is not in chip RAM, it attempts to
reallocate it with #BMF_DISPLAYABLE. This should solve the problem of
multiview opening custom screens in fast RAM with certain datatypes.
3.47...
ditto FDraw() and FBltPattern().
This version contains a new system to deal with RTG NewIcons. Any BOB image
which is found to be in fast mem will be copied back to chip mem. The rest
is left up to the OS to deal with (ie. BOBs are again rendered on the
blitter). This also means that BOBs will use chip mem while they are
dragged, and they may dissapear (or become corrupt under extremes) when
there is very little chip mem left. The only likely downside of this, is
that it takes time (when the 'drag' is started) to copy the BOB images back
to chip, and that may be significant when large numbers of BOBs are dragged
on slower processors.
3.51...
FBltBitMap() has a new option to check the stack size of the calling task.
This is really just a debug feature for my own use. If enabled, output will
be generated when the calling tasks free stack falls below a threshold,
this is set at $200 (512 bytes) and is NOT user definable (unless you hack
the library). No other action will be taken. Output is in the form of
'enforcer' hits, so you will need to run 'enforcer', 'cyberguard',
'muforce' or whatever to see it. The hits are identified by several
features. They are caused by a long read from $0, and originate in
'fblit.library'. Also, d0 will contain $c0ded00d. d1 will contain the tasks
current free stack space, negative values may indicate stack overflow. The
fact that no output is generated does not prove that you will not have
stack problems, it simply shows (probably!) that there is sufficient stack
at a certain point (ie. when calling BltBitMap()). Equally, if you do get
hits (and you almost certainly will), it does not necessarily mean you have
a problem, though 512bytes is sailing pretty close to the wind
(FBltBitMap() itself requires arround 256bytes ($100)). Also, some things
(eg. NewIcons, and many other 'E' programmes) have messed up stack
parameters which may generate hits.
FDraw() is still incomplete but everything else is basically finished
(other than documentation).
There are no known problems with this version, other than line related
issues on promoted gfx.
Contact
-------
Stephen Brookes <sbrookes@tpec.u-net.com>