Re: Twilight II updates (was Re: Screen Saver)

[nathan@cco.caltech.edu (Nathan Mates)]

In article <3311c2$r9r@news.iastate.edu>,
Ian Schmidt <irsman@iastate.edu> wrote:
>In article <330us3$h5q@gap.cco.caltech.edu> nathan@cco.caltech.edu (Nathan Mates) writes:

>Yes, and delaying is Bad, especially when the delay occurs inside an
>interrupt handler.  A GNO screen saver done right would use a daemon
>process, so as not to unnecessarily take up interrupt time.  This will
>be critical when Derek finishes GS/TCP, as SLIP will need huge amounts
>of packet processing in order to work, as well as speedy servicing of
>serial interrupts (indeed, it has it's own serial handler in order to
>chop every cycle possible).

   Interrupt handlers can do o instruction, and they won't be delayed
in the middle. But, if you want a GNO screen saver as opposed to a GS
one, there should be a few system IPCs that could be agreed upon
("GNOScreenSavers" is what you'd call it, but I digress) and used to
tell any installed screen saver to do its blanking. Also, GNO was an
obscure unstable environment (only 1.0) when T2 1.1 came out; now that
it has gained a niche market, maybe someone with the specs on a daemon
and all could make one.


>Heh.  My "longstanding opposition" to T2 has more to do with it's effect
>on my system's performance than anything.  On decent-sized builds
>there's a significant slowdown in compile speed with T2 installed.
>Beyond that, I like the program.  I wrote a couple of modules for it
>that I use when I have it enabled.  It very much needs a design reality
>check however; 3 VBL tasks to do what UltraBlank does in 1 is
>unrealistic on a machine as relatively slow as the GS.

   Are you ever going to learn what's going on, rather than freaking
out at the sight of the number 3? [Hint: The Minbari _are_ out to get
you ;)] Of the 3, one is always inactive, as the GS can be in only
text or graphical mode at any one time. The second is a controller HB
task, to switch between the others, so it has a long period.

   As to slowing down the system, I decided to conduct my own tests
with the somewhat optimized 2.0b1... On my Zip 9/32 system, a set
of builds in Orca took roughly 11 minutes to complete. Deactivating
T2 and rebooting shaved 15 seconds off it. Noticeable? 1.5 seconds
per minute? Maybe. Or maybe it's an anti-placbo effect for you. You
think things go slower, so in your mind, they are magnified. Not
much I can do about that...

   (I tested out this with GNO's time command... But, after working
for a while from the GNO shell, I decided to check this time again.
With T2 2.0b1 installed (just like the first check), it was up to 7.5
minutes. It seems as if there's a system time leak somewhere.)

>Uh huh.  And would I have beta-tested GNO until and including the
>present and written 6 utilites for it if I were as critical of it as you say?

   "If I had written a couple modules for T2 would I be as critical
of it as you say"? Look at your words above for the answer.


Nathan Mates


-- 
* Nathan Mates   http://www.ugcs.caltech.edu/~nathan/               *
* Ftp humor archiver: ftp to cco.caltech.edu, look in pub/humor     *
* Largest collection of Clinton Jokes, other canonical lists        *
* Support Twilight II, the best screen saver for the Apple IIGS!    *