[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference noted::hackers_v1

Title:	-={ H A C K E R S }=-
Notice:	Write locked - see NOTED::HACKERS
Moderator:	DIEHRD::MORRIS

Created:	Thu Feb 20 1986
Last Modified:	Mon Aug 03 1992
Last Successful Update:	Fri Jun 06 1997
Number of topics:	680
Total number of notes:	5456

623.0. "Subprocess MBX input state?" by JENEVR::MEAGHER (W. C. Fields Forever) Mon Dec 07 1987 07:51

If this question has been asked and/or answered before, please
forgive me -- and point me to the answer!

Anyway...

I have a program that creates 2 mailboxes, then creates a subprocess
with SYS$INPUT and SYS$OUTPUT equated to the mailboxes. Fine, now
I can send commands to the subprocess and receive the output. But...
how can my program sense whether a read hung on SYS$INPUT was done from
DCL or from an image running in the subprocess? In other words, how
can my program tell whether the subprocess is in DCL "$" state?

Obviously I can do a $GETJPI to see if there's an image running, but
if at DCL, my subprocess might be executing a command procedure which
might have issued a "$ READ SYS$INPUT FOO" or an "$ INQUIRE". Is there
a magic bit I can check?

Thanks,

bob

T.R	Title	User	Personal Name	Date	Lines
623.1	May be worth a try	SQM::HALLYB	Khan or bust!	`Mon Dec 07 1987 17:09`	2
	Check the subprocess' PSL bits 24:25. If ^B10, then DCL. (PSL is in PHD$L_PSL).
623.2	Is there a less-priv'ed way?	32288::MEAGHER	Too good to be true	`Sun Dec 20 1987 16:32`	4
	Thanks, but is there any way to do this without CMKRNL (or any other god-like) privilege? b
623.3		PSW::WINALSKI	Paul S. Winalski	`Sun Dec 20 1987 16:51`	9
	RE: .2 No, there is not. Note also that the mode field in the PSL won't let you distinguish DCL trying to read a command ("$" state) from DCL reading on behalf of a user command procedure ("$ READ SYS$INPUT FOO", "$ INQUIRE"). --PSW
623.4	Oh well...	32288::MEAGHER	Too good to be true	`Mon Dec 21 1987 07:09`	12
	Well, thanks for trying... I wonder how ALL-IN-1 is able to know when its subprocess is in a "$" state, particularly when the user presses RETURN. At that point, ALL-IN-1 assumes control of the terminal again. ALL-IN-1 does not, I believe, need to be run with CMKRNL privilege. Does anyone know how that mechanism works, offhand? If not, I guess I'll post this question in the ALL-IN-1 notes file. Or read the fiche, if I can get 'em. b
623.5	How its done...	CHOVAX::YOUNG	Back from the Shadows Again,	`Mon Dec 21 1987 09:01`	23
	Well as I recall from having written some ALL-IN-1 applications during its V1 days, it does it pretty much the same way that the infamous TELL.COM does it. Both TELL and ALL-IN-1 take no chances when they do this, they use a command procedure in the slave process to explicitly prompt for input. Thus, ALL-IN-1 creates a subprocess with output, say, A1_MBX_OUT, and input A1_MBX_IN, and error A1_MBX_ERR (I forget how it actually goes so I am improvising). The first command it puts in A1_MBX_IN is "@A1_SUB_CTRL.COM". This subprocess would then do something like this: $ LOOP: $ Write sys$input "~WAITING FOR INPUT$" $ Read sys$input COMMAND_STRING $ COMMAND_STRING $ goto LOOP Whenever the Master process saw the string "~WAITING FOR INPUT$" it knew that it could send another command. -- Barry
623.6		PSW::WINALSKI	Paul S. Winalski	`Tue Jan 05 1988 16:32`	9
	RE: .5, .4 Yup, .5 has it more or less right. This is how SCOPE tool cooperation did it, and how MMS does it, as well. You send a command down SYS$INPUT to the subprocess that does a DCL WRITE of a known string to SYS$OUTPUT, then you know that once you see your string back again, DCL is ready for another command. --PSW
623.7	Thanx!	JENEVR::MEAGHER	Comfortably Numb	`Wed Jan 06 1988 08:28`	8
	Thanks, all. I got spoiled early on by RSTS pseudo-keyboards, which could tell the controlling program whether the process was in a system command wait state. I was hoping for a similar or equivalent mechanism to be hidden somewhere in VMS. Oh well... bob
623.8		CHOVAX::YOUNG	Back from the Shadows Again,	`Thu Jan 07 1988 23:32`	6
	Re .7: Yeah, Bob. That's what we get for leaving a sophisticated Operating System like RSTS/E for something primitive like VMS. -- Barry