home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Atari Mega Archive 1
/
Atari Mega Archive - Volume 1.iso
/
lists
/
gem
/
l_1199
/
1151
< prev
next >
Wrap
Internet Message Format
|
1994-08-27
|
7KB
Date: Mon, 1 Aug 94 02:03 CDT
From: ekl@sdf.lonestar.org (Evan K. Langlois)
To: gem-list@world.std.com
Subject: Re: GEM List
Precedence: bulk
vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv
Toolboxes are not the only place background windows should be enforced. Make
it so that if a user creates a window with your library, they have the
option to use WF_BEVENT. If they are not using the right TOS version, then
handle the WF_BEVENT bit yourself!
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Why not set the WF_BEVENT bit for all windows. Handle topping yourself,
and handle changing tops to clicks yourself. The only restriction here
is that a drag cannot be done from a backrgund window unless the OS
has WF_BEVENT.
vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv
assed coder. TRY something for a change, rather than coming up with excuses
to get out of everything before even trying it.
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Just because you try it, and it works, does NOT mean that it is portable, nor
does it mean it is good programming style (and I don't mean the position of
your brackets, I mean inter-application cooperation).
vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv
gadgets. XAES doesn't *force* you to do anything. XAES is all about
*choices*, not being forced to do anything you don't want it to.
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
It should look and feel like standard GEM, and actually USE standard GEM
(no faking any of it!). Only if the user specifically sets a certain
option should the new gadgets be used. Maybe something like :
SomeApp.Motif.Gadgets: On
Otherwise, don't!
vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv
Have you used WinX? They have different window gadgets, and yet, I don't
hear anyone complaining.
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
No, the gadgets are NOT different! You can have the arrows positioned
differently, but this is system-wide! Not per-application (although
anyone that really wants to, could make it per application). A system
wide, and minor change at that, is OK. But switching from Gem gadgets to
Motif Widgets for just one app is gonna irritate some people. I would
suggest only using them for the above app-defs line, otherwise, pass
some GEM Gadgets to your wind_create call eh?
vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv
I would have to see this to believe it. MiNT itself is slower than snot.
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
What tests have you done? MiNT .95b was about 2-3% slower than TOS 1.62
when running over TOS 1.62. I used numerous benchmarks. MiNT has been
getting faster and more compatible ever since. When 1.11 comes out, it
will be freely available. You can see for yourself when its released just
how fast it is. How fast can your system print text? On my MiNT system
I get about 8000 cps! With scrolling, using Cconws or Fwrite, in 80
character chunks. It would go faster if I just used one long 32K write
and timed it! If you need help using MiNT, email me. You must be doing
something weird with it (not surprising considering your abuse of the AES)
vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv
Put the new AES on top of it and things really go downhill. If you'd like to
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Any AES will slow the system since the AES does a constant polling for
events which eats up CPU time. Something your own programs do! A non-polling
AES will be released at a later date.
vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv
Better yet, do some benchmarks and give some *hard* facts, numbers that
can be *duplicated* by anyone. Try something like GemBench and Speedometer
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
I don't currently have either of those programs right now, but GemBench
used to give me 96-99% with MiNT .95, and with 1.10+ its even more efficient,
especially in BIOS IO. Since I now have an AdSpeed, posting tests would
not be an accurate comparison anymore (which is why I posted results for
MiNT .95, which is also freely available in binary form- try it yourself).
vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv
Atari got their head out of their .... with this new MultiTOS (at least in
terms of speed, not usability -- that is a completely different issue.)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Flame on:
The new MultiTOS is faster because it uses device drivers and Fselect(), or
something similar instead of polling the hardware. Fselect() is like
evnt_multi for file handles. You yourself have turned evnt_multi into
a polling call by using 0ms timers. Your programs will still be VERY slow
under any operating system because you STILL haven't gotten your head out
of YOUR ASS and realized that polling is a bad programming practice. You
talk alot about others and ATARI, but you are doing the exact same thing.
In fact, what do is worse, because there is a way around it. Until MiNT
came along, ATARI didn't have a better way, or even a reason. Maybe instead
of insulting a lot of people that are really doing wonders for the OS,
you should shut the hell up because you don't know what you are talking
about, and that goes for MiNT -AND- how you think the AES works and your
demand for benchmarks (which can't show half of what you ask anyway).
Flame off:
vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv
I'm not sure where you got that figure. You must have lots of OTHER
programs contributing to that figure. And don't use Warp9 with MiNT
either - that steals MORE system time than MiNT does. You can take two
identicle systems, one running MiNT and one running Warp9. The MiNT
one will do graphics slow, and CPU operations fast. The Warp9 one will
do graphics fast, and CPU operations will be slower than MiNT! If you
combine MiNT and NVDI, you get a nice fast system all-around that can
handle multiple tasks. See :
PID PPID PRI CURPRI STATUS SIZE TIME COMMAND
000 000 0 0 Wait 46800 01:13.43 MiNT
003 000 0 0 Ready 136864 09:25.54 vconsd
005 000 0 0 Ready 37328 18:21.51 ttyd0d
007 000 0 0 Sleep 98992 00:00.14 sh
009 000 0 -3 Ready 76576 01:51:12 top
010 000 0 0 Sleep 98992 00:00.98 sh
013 000 0 0 Wait 100336 00:08.63 GEM
014 013 0 -1 Ready 237488 13:09:43 TOSWIN_W
016 000 0 0 Sleep 237040 00:28.45 bash
019 000 0 0 Sleep 35984 00:06.82 wterm
020 019 0 0 Sleep 40688 09:16:56 wterm
110 000 0 0 Sleep 77344 01:55.52 me
123 016 0 -5 Ready 29376 00:00.63 ps
And I'm not doing that much multitasking right now! I'm not very busy.
Normally I've got 2 or 3 editors going and a couple file viewers while
a zmodem is going. And there is no slowdown. I have to run about 2
or 3 archivers and the AES before getting a "slow as snot" speed. Or
GNU C - the constant disk access makes things jump. 16Mhz system.