Gribnif@umass (Dan Wilga at UMASS Amherst) (02/21/88)
Greetings! Is there some way to actually "tell" desk accessories that are currectly opened within an application to close? I am writing a program that uses Pexec() to call other programs, but the problem I am running into is that any DA's that occupy windows and are left open when my program executes another do not receive the AC_CLOSE message and so do not know they should be closed. Of course, I cannot simply close their windows for them; that only messes them up. Pterm() somehow does this for itself since it automagically sends the correct message when the application terminates. shel_write() also seems to suffer from this same problem. Is there any good solution, short of hacking-into the OS? ---------------------------------------------------------------------------- "In those days, men were real men, Dan Wilga women were real women, and small furry creatures from Alpha Centauri Gribnif@UMASS.Bitnet were REAL small furry creatures from Alpha Centauri" -- Hitchhikers
apratt@atari.UUCP (Allan Pratt) (02/24/88)
in article <8802210020416D5.BLBS@Mars.UCC.UMass.EDU>, Gribnif@umass (Dan Wilga at UMASS Amherst) says: > > Greetings! > > Is there some way to actually "tell" desk accessories that are currectly > opened within an application to close? I am writing a program that uses > Pexec() to call other programs, but the problem I am running into is that > any DA's that occupy windows and are left open when my program executes > another do not receive the AC_CLOSE message and so do not know they should > be closed. Of course, I cannot simply close their windows for them; that > only messes them up. Pterm() somehow does this for itself since it > automagically sends the correct message when the application terminates. > shel_write() also seems to suffer from this same problem. Is there any > good solution, short of hacking-into the OS? I think you hit on the solution without knowing it: closing their windows doesn't pay, but you can SEND THEM the AC_CLOSE message (I think). Check the message library for "send a process a message" and check the application library for "get ap_id for an accessory" (I don't really know that much AES). Then send the accessory the close message. Sending it wind_close is usually enough, since (like the Control Panel) clicking "close" in the window usually shuts down the accessory as well. Failing that, you could try the opposite of app_init -- I forget its name -- because I think it's that call which closes accessories, not Pterm(). Pterm() doesn't know a thing about accessories, or AES, or anything at that level. So try app_deinit before the Pexec and app_init after it. (Remember that your window handles might be different after the new init... They probably won't be, but write your code as though they might be.) Like I said, I don't know much about AES -- don't take this as Gospel. ============================================ Opinions expressed above do not necessarily -- Allan Pratt, Atari Corp. reflect those of Atari Corp. or anyone else. ...ames!atari!apratt
Gribnif@UMass.BITNET (Dan Wilga at UMASS Amherst) (02/27/88)
Greetings!
In my previous letter I was looking for a way to close desk accessories before
using Pexec() to call a second GEM application from within a program. Well,
I have found something that seems to work (after finally discovering some
docs that actually explain the way appl_write() REALLY works). The following
code segment seems to do the trick.
int buff[8], w_hand, i, AES_handle, dum;
AES_handle = appl_init(); /* get the handle of my application */
.
. /* the guts */
. /* close any windows still open */
wind_get( 0, WF_TOP, &w_hand, &dum, &dum, &dum ); /* topmost window */
while( w_hand > 0 ) { /* until no more windows */
wind_close( w_hand );
wind_delete( w_hand ); /* now the next one down is top */
wind_get( 0, WF_TOP, &w_hand, &dum, &dum, &dum ); /* get the next one */
}
buff[0] = AC_CLOSE; /* emulate the message normally received */
buff[1] = AES_handle; /* handle of app. that caused the close */
for( i=0; i<AES_handle; i++ ) { /* assume the DAs have lower handles */
buff[2] = 2+i; /* an educated guess */
appl_write( i, 16, buff ); /* the message is read the next time the */
} /* DA does evnt_mesag() or evnt_multi() */
I am just wondering two things: first, is it safe to assume that any DAs will
have AES handles less than the current application? Second, assuming a DA
actually does care what menu id (as stored in buff[2]) the DA is assigned to
would my little guess be correct? Granted, there really is no reason I can
see why a DA should care if the menu id passed matches what it sees as being
correct, but considering how few correctly written DAs there are out there,
who knows.
Now for my opinion on the Copyright issue....naw, just kidding.
Dan Wilga
-----------------------------------------------------------------------------
"You may think it's a long way down Gribnif@UMASS.Bitnet
the road to the chemist, but that's Gribnif%UMASS.forsythe.stanford.EDU
just peanuts to space." (methinks)
-- Hitchhikers