[comp.sys.amiga] ARP v1.1 developer files, and Manx v3.6a

kim@amdahl.uts.amdahl.com (Kim DeVaughn) (03/23/88)

[ "Nicaragua" is Spanish for "Vietnam."   --Bob Page ]

I received the following email today.  I'm posting it to .tech since that's
where it belongs [ and is for "future development" :-) ], and to the regular
group because many people who should see it don't get .tech (yet).


> Date:     Mon, 21 Mar 88 16:06:26 PST
> From: hplabs!RCKG01M%CALSTATE.BITNET@CUNYVM.CUNY.EDU  (Stephen Walton)
> Subject:  ARP 1.1 developers' stuff
>
> [My motto:  when in doubt, mail instead of post.]

Much to be said for that line of reasoning, given the current S/N ratio.  Think
about it.


> I just FTP'd the ARP V1.04 developers' goodies from purdue.  When I tried
> to use the Manx version of the libraries, I got "Not an object module"
> errors, presumably because the .o format changed from 3.4 to 3.6 (I use
> the latter).  Ok, no sweat, says I, I'll just rebuild the libraries
> from the included sources.  Fine, except that the assembler files
> INCLUDE exec/exec_lib.i, which is full of FUNCDEF references, which
> is not defined anywhere that I could find.
>
> Will 1.1 be built with 3.6?  Can you help me with this FUNCDEF problem
> if it isn't?  Thanks in advance.
>
> Steve Walton, Cal State Northridge
> USENET is acphssrw@csuna.UUCP, but machine is defunct for a week
> This is RCKG01M@CALSTATE.BITNET

I hope Steve will see this reply, as my email to him will probably bounce.
Never can seem to get thru to a BITNET site ...

Unfortunately, I could find nothing in any of the v1.1 ARP docs, Makefile for
Manx, includes, etc. that indicated what version of Manx was used to create
the libs, etc.  I'm *guessing* it was v3.4x, though.

Anyway, I tried making the arp libs for Manx (v3.6a) so I could compare them
with the ones supplied in the v1.1 ARP distribution, and ran into the same
problem as Steve did.  I.e., the  "exec/exec_lib.i"  file supplied with the
v3.6a Manx compiler is full of nothing but FUNCDEF's.

I have a sneaking suspicion this has something to do with the ill-understood
(at least by me)  .fd  files supplied by CBM, but I'm not sure.  Anyone know
what the next step is?

I'm also sending a copy of this posting to "cheath" hisself, and to Jim at
Manx.

Thanks for any help!

/kim



-- 
UUCP:  kim@amdahl.amdahl.com
  or:  {sun,decwrl,hplabs,pyramid,ihnp4,uunet,oliveb,cbosgd,ames}!amdahl!kim
DDD:   408-746-8462
USPS:  Amdahl Corp.  M/S 249,  1250 E. Arques Av,  Sunnyvale, CA 94086
CIS:   76535,25

cmcmanis%pepper@Sun.COM (Chuck McManis) (03/24/88)

The fact that Arp 1.1 will not compile with Manx 3.6 is well known. As
soon as Scott Ballintine(sp?) get's his update it will be made compatible
with 3.6. This is being worked on and everyone who can fix it knows about
it so don't panic.


--Chuck McManis
uucp: {anywhere}!sun!cmcmanis   BIX: cmcmanis  ARPAnet: cmcmanis@sun.com
These opinions are my own and no one elses, but you knew that didn't you.

kim@amdahl.uts.amdahl.com (Kim DeVaughn) (03/24/88)

[ "Nicaragua" is Spanish for "Vietnam."   --Bob Page ]


In article <46655@sun.uucp>, cmcmanis%pepper@Sun.COM (Chuck McManis) writes:
> The fact that Arp 1.1 will not compile with Manx 3.6 is well known. As
> soon as Scott Ballintine(sp?) get's his update it will be made compatible
> with 3.6. This is being worked on and everyone who can fix it knows about
> it so don't panic.

Gee Chuck, I guess I forgot to look into my crystal ball, or I would've known
that too ... :-).   As may be ...

No "panic", just trying to get ... Information.  And it seems to have worked,
as Steve found a workaround, which is attached.  Now people will be able to
develop with ARP functions, *and* use SDB (which is a real win).  Looks like
it might be a little while until "official" v3.6 compatible libs will get
released (see Charlie's note, also attached).


But first, here's the solution from Steve  (seems I *can* talk to at least
one BITNET site afterall):

: Date:     Wed, 23 Mar 88 14:43:20 PST
: From: hplabs!RCKG01M%CALSTATE.BITNET@CUNYVM.CUNY.EDU  (Stephen Walton)
: Subject:  RE: Arp V1.1 Developers' Stuff
: To: amdahl!kim
:
: I did indeed get your mail, Kim.  And thanks for the cross-post;  our
: site bounces .tech as a bogus newsgroup.
:
[ hopefully, not for *too* much longer ]
:
: I solved the problem.  A quick pass with lb V3.4 shows that ARP was
: built with it.  As for FUNCDEF, how does
:
: FUNCDEF MACRO   *
:         public  _LVO\1
:         ENDM
:
: strike you?  It works, though it pulls EVERY _LVO into your executable
: when it is used, so  [ instead of that ]  I just put statements to declare
: _LVOOpenLibrary  and  _LVOAlert  as public in the  arpcrt0.s  file.
:
: Incidentally, two of the routines listed in  exec_lib.i  don't have
: offsets in the Manx libraries (came up undefined with the above
: FUNCDEF macro):  ReleaseSemaphore  and  SumKickData.
:
: Since my USENET access goes away for a week this Sunday, I probably
: won't see your posting;  we're in the boonies, USENET speaking.  So
: you're free to post the above.
:
: Steve

I found that I didn't seem to need to add  _LVOAlert  to  arpcrt0.s, but that
I did need to add  _LVOOpenLibrary  to it, and to  arprescrt0.s  as well.
Also, I needed to add  _LVOFindTask  to the  resgeta4.s  file.

Of course you need to remove the  INCLUDE  "exec/exec_lib.i"  statement, after
adding in the  public _LVO...  statements above.  Just put 'em under the other
Public  statements you'll find.

Then type "make all".  Simple, no?

BTW, I have NOT tried to link in and *use* these new libs in anything, but
"make" and friends seem happy enough.  Caveat emptor!



As to what Charlie Heath had to say:

: Date:  23-Mar-88 08:57 PST
: From:  c heath [74216,2117]
: Subj:  ARP v1.1 problem
:
: We didn't (still don't) have V3.6, so we didn't know Jim had changed the
: library format again.  As soon as MANX sends Scott a copy of V3.6, we'll
: rebuild the library, though likely won't post until we have all the
: revisions we want to make for V1.1 of ARP, in a couple months - we want
: to minimize the revisions to avoid confusion and potential viral clones.
:
: Willy Langeveld said he got the MANX stuff to rebuild with V3.6 just by
: reassembling and building the library, you can probably get that to
: work if you want to try it.
:
:        Charlie


Now, at least y'all know what to expect when the distribution is posted from
purdue to one of the moderated newsgroups.

/kim



-- 
UUCP:  kim@amdahl.amdahl.com
  or:  {sun,decwrl,hplabs,pyramid,ihnp4,uunet,oliveb,cbosgd,ames}!amdahl!kim
DDD:   408-746-8462
USPS:  Amdahl Corp.  M/S 249,  1250 E. Arques Av,  Sunnyvale, CA 94086
CIS:   76535,25

barrett@ektools.UUCP (Chris Barrett) (03/25/88)

> : I solved the problem.  A quick pass with lb V3.4 shows that ARP was
> : built with it.  As for FUNCDEF, how does
> :
> : FUNCDEF MACRO   *
> :         public  _LVO\1
> :         ENDM
> :
> : strike you?  It works, though it pulls EVERY _LVO into your executable
> : when it is used, so  [ instead of that ]  I just put statements to declare
> : _LVOOpenLibrary  and  _LVOAlert  as public in the  arpcrt0.s  file.

I called Manx and got the official definition for FUNCDEF, edit the
exec/exec_lib.i file and put the following lines at the top:
 
count	SET -30
FUNCDEF	MACRO
_LVO\1	EQU count
count	SET count-6
	ENDM
 
I hope this helps! 
 
Chris
> :