rosenkra@hall.cray.com (Bill Rosenkranz) (04/23/89)
--- question: i notice that on startup of a GEM .prg file, the command line in the basepage is all 0's. this is true before appl_init and after. does AES/VDI or anything else ever use this and (more importantly) can i stick things in it? question: does anybody have a WORKING demo showing message passing between an application and a desk accessory? the communications should be able to initiate on both sides. the docs on the various gl_apids from atari are totally vague and atari GEM gurus themselves don't know what is the diff between gl_apid (set by AES during appl_init) and the return from the appl_init itself (often not the same value, but that could be gulam screwing up). on a related issue, i have been trying to install some temporary drivers (for midi stuff, just calls Midiws, Bconstat, etc) using Pexec(3,...) and invoking with Pexec(4,...) without much luck. actually the installs and invocations work fine but i try to Malloc (not malloc) AFTER installing 2 or 3 drivers and find only 50 bytes left (pretty suprising on a MEGA 4 8^). the code hangs on the 11/85 ROMs to boot. what gives? is Pexec just a waste of time? why can't atari just fix the damn memory management already? sheesh. heres what i want/tried: memory synosis: ______________________ appl (GEM or ttp) ______________________ AES (rsc, workstat,etc) ______________________ Malloc (screen) ______________________ driver 1 (Pexec(3...)) ______________________ . . ______________________ driver n ______________________ Malloc (most of mem) ______________________ the last Malloc shows little or no mem left. the drivers are only 25k or so, and i DO do Mshrink (or rather atari's gemstart.s does it) in each driver. The Pexec(3,...) gives weird addr for each driver (i tried 2). by weird i mean driver 2 comes before driver 1. i tried this both with a ram disk and in gulam as well as from the desktop with no other memory resident stuff and get similar results. i used Ptermres to exit the driver (until the main appl tells it to stop via shared memory communication block in which case driver does a Pterm). i have no problems invoking the drivers any number of times but i need to do the Malloc last since the appl needs all the mem it can get and i don't really want to have to figure out how much space the drivers will need before hand. even if i do this, i fear the Pexec(3,...) will probably fail anyway. BIG :-(... i use alcyon 4.14 and friends (aln, libs, gemstart) on both Mega4 and 1040 (11/85 ROMs). any ideas? pls reply with email as our site misses about half of what we get via postings. thanx... -bill rosenkra@hall.cray.com or rosenkra@boston.cray.com ...!umn-sc!hall!boston!rosenkra