| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 500.1 | dave's looking into it | SMURF::BAT | Segui la tua beatitudine | Thu May 08 1997 21:53 | 11 |
| From: SMURF::BAT "Barbara A. Thomson ZKO3-2/X46 1-2955" 8-MAY-1997 17:52:03.11
To: NM%US2RMC::"CWOLF@us.oracle.com"
CC: NM%US2RMC::"nliang@us.oracle.com",BAT
Subj: re: Header file problems with DEC MLS+ 4.0
We just compared with the latest DU pool, and it looks the
same. While we do a cc -E, could you check to determine
if any of the flags that might turn on X_NOT_STDC_ENV
are getting turned on? Those flags are in /usr/include/X11/Xosdefs.h
|
| 500.2 | chris' command line | SMURF::BAT | Segui la tua beatitudine | Tue May 13 1997 00:48 | 55 |
| From: US2RMC::"CWOLF@us.oracle.com" "CWOLF.US.ORACLE.COM" 8-MAY-1997 19:42:25.99
To: smurf::bat
CC:
Subj: Re: RE: Header file problems with DEC MLS+ 4.0
Barbara,
Here is the command line we're using:
cc -c -O0 -oldc -Wf,-XNd15000,-XNp15000,-T100 -std1 -non_shared -DOSF1
-DA_OSF -G8 -D_NO_PROTO -DuiXT -DX11R5 -DuiMOTIF -DuiMOTIF_1_2 -DuiXPM
-DNO_XPM -DMM_EXT
I couldn't reproduce the problem in a small test program.
Thanks,
-Chris
+------------------------------------------------------------------------+
| ORACLE Corporation |
| cwolf@us.oracle.com |
| Chris Wolf, Senior Software Engineer Voice +1.415.506.2529 |
| World Wide Government & Education Fax +1.415.506.7800 |
| Products Division |
+------------------------------------------------------------------------+
--=_ORCL_19551620_0_11919705082014540
Content-Type:message/rfc822
Date: 08 May 97 18:01:22
From:"Barbara A. Thomson ZKO3-2/X46 1-2955 <bat@smurf.ENET.dec.com>" <bat@smurf.ENET.dec.com>
To:nliang@us.oracle.com,cwolf@us.oracle.com
Subject:RE: Header file problems with DEC MLS+ 4.0
Cc:nliang@us.oracle.com,bat@smurf.ENET.dec.com
MIME-Version: 1.0
Content-Transfer-Encoding:7bit
Content-Type:text/plain; charset="US-ASCII"
BTW, X_NOT_STDC_ENV is not defined in the normal MLS+ on an
Alpha environment. In V3.1A there was no distinction,
evidently there are platform distinctions in V4.0A (this
is a base issue, not MLS+ issue).
We haven't found anything in our header files just now
that would have turned it on... can you check yours?
thanks
|
| 500.3 | input from dcj was | SMURF::BAT | Segui la tua beatitudine | Tue May 13 1997 19:27 | 20 |
|
Hi Chris,
Some of those switches turn on things I don't have in
my database.
I got some suggestions from Dave about how to find
the definition that is turning on the other definition:
1. check all your files to see if you have
a Mips or vax (watch capitalization) turned on
2. or send us the include files from the offending files
and we'll search.
Does this happen on a vanilla DU box?
Thanks
BAT
|
| 500.4 | dave hasn't heard of it -- it's only in X stuff | SMURF::BAT | Segui la tua beatitudine | Mon May 19 1997 23:18 | 25 |
| From: US2RMC::"CWOLF@us.oracle.com" "CWOLF.US.ORACLE.COM" 16-MAY-1997 21:08:13.56
To: smurf::bat
CC: NLIANG.US.oramail@us.oracle.com
Subj: Re: RE: network operations can't be killed on MLS+ 4.0
[ excerpted ]
>P.S. Whatever happened re: the ifdef that is not supposed
>to be there?
Norman found the problem:
At the bottom of the new (4.0) version of <X11/Xosdefs.h> is:
#if defined(nec_ews_svr2) || defined(SX) || defined(PC_UX)
#define X_NOT_POSIX
#define X_NOT_STDC_ENV
#endif
The probelm is that Oracle also defines something called "SX". What is
"SX" for DEC? My guess it's some funky OS type that doesn't mean anything
to DEC, but rather is a newly supported OS for X11.
|
| 500.5 | luck has nothing to do with it | SMURF::BAT | Segui la tua beatitudine | Tue May 20 1997 16:31 | 17 |
| From: ALPHA::wiele "John Wiele USG" 20-MAY-1997 10:32:56.85
To: 19.807::bat, alpha::wiele
CC: wiele@DEC:.zko.alpha
Subj: Re: Notefile DEC_MLS_PLUS Note 500.4
Good deduction. If you look in src/xc/config/cf/nec.cf, you will find
that SX is the OS definition for the NEC SX3 Architecture, whatever
that is. Apparently it doesn't have a POSIX and ANSI C environment.
I guess that Oracle can change their variable, or they can edit
Xosdefs.h, or they can manipulate the #defines around an explicit
include of Xosdefs.h.
Good luck.
-JW
|
| 500.6 | from chris | SMURF::BAT | Segui la tua beatitudine | Tue May 20 1997 20:11 | 21 |
| From: US2RMC::"CWOLF@us.oracle.com" "CWOLF.US.ORACLE.COM" 20-MAY-1997 14:38:10.68
To: smurf::bat
CC:
Subj: Re: luck has nothing to do with it IMO
Thanks for checking. We won't change our definition of SX because it is
at the top of a large hierarchy of nested include files. I guess the thing
to do which has the least impact, is to modify Xosdefs.h and add
|| !defined(ORACLE) to qualify that conditional. BTW, luck has nothing
to do with what?
-Chris
+------------------------------------------------------------------------+
| ORACLE Corporation |
| cwolf@us.oracle.com |
| Chris Wolf, Senior Software Engineer Voice +1.415.506.2529 |
| World Wide Government & Education Fax +1.415.506.7800 |
| Products Division |
+------------------------------------------------------------------------+
|
| 500.7 | to chris | SMURF::BAT | Segui la tua beatitudine | Tue May 20 1997 20:12 | 24 |
| From: SMURF::BAT "Barbara A. Thomson ZKO3-2/X46 1-2955" 20-MAY-1997 15:43:13.93
To: US2RMC::"CWOLF@us.oracle.com"
CC: BAT
Subj: Re: luck has nothing to do with it IMO
luck has nothing to do with whether your choice works or not.
two questions:
1) re: modify Xosdefs.h and add || !defined(ORACLE) to qualify
that conditional
i presume you mean something like:
#if defined(FOO) || defined(SX) && !defined(ORACLE) || defined(FUD)
2) at any rate, changing Xosdefs.h shouldn't make any difference
except during installs -- do you guys use the "subset" mechanism
with setld to install Oracle? if yes, you can modify the
.scp file to include a procedure to make sure that the above
change is made to Xosdefs.h -- will this affect any people
installing X apps on top? wouldn't think so, but not sure.
|
| 500.8 | still no "miracle cure" | SMURF::BAT | Segui la tua beatitudine | Tue May 20 1997 21:14 | 27 |
| From: SMURF::BAT "Barbara A. Thomson ZKO3-2/X46 1-2955" 20-MAY-1997 17:09:30.49
To: US2RMC::"cwolf@us.oracle.com"
CC: BAT
Subj: DU documentation
Do you get the regular digital unix documentation,
either on the CD or however?
I have been searching the notes files, and querying people
here, but so far I haven't come up with a light bulb
among them.
I'm sending you the DU V4.0 (not V4.0A) release notes
which has a section on the DEC C compiler.
If things are not working with -oldc then we should
do those things first, and see if we can find the common
thread/error in the header files.
Kris says that the header files in MLS+ are supposedly
identical (except in the security stuff) with the base OS.
Unless of course we made a merge mistake.
Later
BAT
|
| 500.9 | what is the really right way to do this? argue namespace with X? | SMURF::BAT | Segui la tua beatitudine | Wed May 21 1997 18:43 | 26 |
| re: .7
From: US2RMC::"CWOLF@us.oracle.com" "CWOLF.US.ORACLE.COM" 20-MAY-1997 18:27:08.36
To: smurf::bat
CC: ASCHEN.US.oramail@us.oracle.com, NLIANG.US.oramail@us.oracle.com
Subj: Re: luck has nothing to do with it IMO
>i presume you mean something like:
>
> #if defined(FOO) || defined(SX) && !defined(ORACLE) || defined(FUD)
Yes, exactly.
>
> 2) at any rate, changing Xosdefs.h shouldn't make any difference
> except during installs -- do you guys use the "subset" mechanism
> with setld to install Oracle? if yes, you can modify the
> .scp file to include a procedure to make sure that the above
> change is made to Xosdefs.h -- will this affect any people
> installing X apps on top? wouldn't think so, but not sure.
Very good point, although the customer can only relink (not recompile)
Oracle products, they will run into trouble if SX is defined in the
header files we send to the customer. We'll check into it.
-Chris
|