oliphant@telepro.UUCP (Mike Oliphant) (08/29/90)
Does anyone know of a way to get behavior identical to Manx's exec family of routines in Lattice? I need to chain one program to another, loading the second program over top of the first. Manx does this beautifully, but Lattice doens't seem to be able to do it at all. system() and the fork routines all start up new processes, and that's no good for what I want to do. Any help would be greatly appreciated. This is the only problem holding me back from switching to Lattice. -- Mike Oliphant UUCP: alberta!herald!telepro!oliphant Internet: oliphant@telepro.uucp FidoNet: (1:140/91) - ZMH only * * Call TelePro, the development system for DIALOG Professional * * Phone: +1 306 249 2352 300/1200/2400/9600/14400 baud HST * FidoNet: (1:140/90) *
peterk@cbmger.UUCP (Peter Kittel GERMANY) (08/30/90)
In article <oliphant.3972@telepro.UUCP> oliphant@telepro.UUCP (Mike Oliphant) writes: >Does anyone know of a way to get behavior identical to Manx's exec family of >routines in Lattice? I need to chain one program to another, loading the >second program over top of the first. Why not use the function Execute() of the Amiga exec library? You can pass it a complete CLI command line. If you want to let your old program leave the system then precede this command line with a "run" and then close down the first program. If you problems with input channels then use "newcli ... from batchfile". If you are more expert with Amiga topics then use LoadSeg(). > Manx does this beautifully, but Lattice >doens't seem to be able to do it at all. I can't believe that Manx is able to load a new program on top of it, into the same memory area, as I understand it from your posting. Also I think it cannot use the same task structure, so it just MUST open a new process. If Manx does all this, tell me, that would sound interesting. -- Best regards, Dr. Peter Kittel // E-Mail to Commodore Frankfurt, Germany \X/ rutgers!cbmvax!cbmger!peterk
p554mve@mpirbn.mpifr-bonn.mpg.de (Michael van Elst) (09/01/90)
In article <350@cbmger.UUCP> peterk@cbmger.UUCP (Peter Kittel GERMANY) writes: >In article <oliphant.3972@telepro.UUCP> oliphant@telepro.UUCP (Mike Oliphant) writes: >I can't believe that Manx is able to load a new program on top of it, >into the same memory area, as I understand it from your posting. >Also I think it cannot use the same task structure, so it just MUST >open a new process. If Manx does all this, tell me, that would sound >interesting. As far as I know, Manx uses several - say undocumented - techniques to load a program and pass control to it. Now, how does the CLI start a command ? The same is done by the Manx routines, LoadSeg()ing the new module, creating certain system structures, patching others and then roll up the old segment list to free the memory of the old module. Who needs system conformance ? Let's force OS revisions to make Manx executables work :-) Regards, -- Michael van Elst UUCP: universe!local-cluster!milky-way!sol!earth!uunet!unido!mpirbn!p554mve Internet: p554mve@mpirbn.mpifr-bonn.mpg.de "A potential Snark may lurk in every tree."
dillon@overload.Berkeley.CA.US (Matthew Dillon) (09/05/90)
Neither Manx nor Lattice handles exec or fork/exec properly (nor DICE
for that matter). It is not their fault however. The problems are due
to deficiencies in 1.3 . As far as I know, 2.0 solves all the problems
with new capabilities and calls.
Here are the problems:
(1) ^C propogation
(2) Starting a task asynchronously
(3) Starting a task asynchronously plus getting an exit code on completion
(4) Starting a task synchronously
(5) Starting a task synchronously plus getting an exit code on completion
Execute() handles (2) and (4) with some amount of overhead but there is
no way to propogate ^C (that I know of) so that it is able to interrupt
the Execute()d program in case (4). The latest SetPatch for 1.3 allows
Execute() to use a resident RUN instead of loading it from disk. Without
the patch Execute() takes a huge amount of overhead.
Aztec's fork/exec handles (1) and (5) but only through major hacks (which
I have never and will never trust).
Lattice's fork/exec stuff handles (3) and (5) but only works with non-BCPL
programs (not a problem under 2.0). However, I've other problems with
Lattice's calls and do not trust them either.
WShell has a patch to get the exit code from an Execute()d program, I
believe, but this is non-standard and you cannot count on it in any
third-party developement you do, and you still have the ^C propogation
problem.
2.0 has solutions to all of these problems but you must be running 2.0
to take advantage of them.
-Matt
In article <350@cbmger.UUCP> peterk@cbmger.UUCP (Peter Kittel GERMANY) writes:
>In article <oliphant.3972@telepro.UUCP> oliphant@telepro.UUCP (Mike Oliphant) writes:
>>Does anyone know of a way to get behavior identical to Manx's exec family of
>>routines in Lattice? I need to chain one program to another, loading the
>>second program over top of the first.
>
>Why not use the function Execute() of the Amiga exec library? You can
>pass it a complete CLI command line. If you want to let your old program
>leave the system then precede this command line with a "run" and then
>close down the first program. If you problems with input channels
>then use "newcli ... from batchfile". If you are more expert with
>Amiga topics then use LoadSeg().
>
>> Manx does this beautifully, but Lattice
>>doens't seem to be able to do it at all.
>
>I can't believe that Manx is able to load a new program on top of it,
>into the same memory area, as I understand it from your posting.
>Also I think it cannot use the same task structure, so it just MUST
>open a new process. If Manx does all this, tell me, that would sound
>interesting.
>--
>Best regards, Dr. Peter Kittel // E-Mail to
>Commodore Frankfurt, Germany \X/ rutgers!cbmvax!cbmger!peterk
--
Matthew Dillon dillon@Overload.Berkeley.CA.US
891 Regal Rd. uunet.uu.net!overload!dillon
Berkeley, Ca. 94708
USA