| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 3576.1 | What he said... | GOTIT::harley | Eat at Ralphs and live forever! | Fri Nov 02 1990 16:34 | 7 |
| I've been playing with this all day; the first time that an application
is started, I get the output window, but if I stop the application,
dismiss it's output window, and start it again I _don't_ get an output window...
FWIW, I'm running VMS T5.4-1
/harley_the_FT_junkie
|
| 3576.2 | Do you get error messages in the output window? | LNKUGL::BOWMAN | Bob Bowman, CSC/CS SPACE Team | Fri Nov 02 1990 18:05 | 21 |
| I regularly get unwanted output windows on some applications. In most cases,
these contain a message resulting from a Set Terminal command. Experimentation
leads me to believe there is a timing window in setting up the TWAn: device for
the Fileview subprocess. This initialization only happens for the first applica-
tion run in the subprocess, so stopping the application and restarting it will
not result in the error and output again. The subprocess is initialized via the
file VUE$LIBRARY:VUE$SUBPROCESS_INIT.COM. This file contains the line:
$SET TERMINAL/VT200/NOBROADCAST
Commenting out this line results in no bogus output windows with error messages.
Placing show commands in the file (Show terminal, show logical, etc.) apparently
changes the timing such that the SET TERMINAL command no longer issues the error
message.
I'm working on isolating this a little more before entering a QAR. (Today I
tried playing with default protections on the TWA0: template device, assuming
the issue might be related to not having the privilege to issue the
set terminal command...my testing did not provide any conclusive evidence that
this is the issue.)
|
| 3576.3 | 1st line of VUE$SUBPROCESS_INIT.COM? | LESLIE::LESLIE | Andy Leslie | Mon Nov 05 1990 07:57 | 8 |
| I've tried commenting it out. In actual fact what I see is
$ SE
which may be the $ SET NOVERIFY line!
More investigations ongoing...
|
| 3576.4 | 1st line of VUE$SUBPROCESS_INIT.COM! | LESLIE::LESLIE | Andy Leslie | Mon Nov 05 1990 08:03 | 7 |
| Indeed it is the first chars of that line! I verified this by switching
$ a+f$VERIFY(0) for $ SET NOVERIFY.
Think its the ol' QAR time again....
/andy/
|
| 3576.5 | | LESLIE::LESLIE | Andy Leslie | Mon Nov 05 1990 08:53 | 7 |
| Aha! The "Set Verify" seems to have come from my DECW$LOGIN.COM but
that still leaves a timing problem around the recognition of the $ SET
NOVERIFY as first line and thus the lack of necessity to create a
popup...
/andy/
|
| 3576.6 | Surely a timing issue, as it depends entirely on *when* one first trys to write to FTAn: | LNKUGL::BOWMAN | Bob Bowman, CSC/CS SPACE Team | Mon Nov 05 1990 11:04 | 11 |
| The exact lines I'm getting are:
ET-W-NOTSET, error modifying FTA5:
-SYSTEM-F-ABORT, abort
I note that in the IFT kit the Fileview is using the new pseudo-terminal device
instead of the old flavor...(FTAn: vs TWAn:)...I was hoping when this happened
it would get rid of the bug if it was a timing issue with the TW device...looks
like the timing hole is elsewhere.
I'll also qar this.
|
| 3576.7 | I have located a "fix" for the ET-F-NOTSET errors. | LNKUGL::BOWMAN | Bob Bowman, CSC/CS SPACE Team | Tue Nov 06 1990 13:18 | 23 |
| I found the following in the listing file for the Vue code which handles init of
the Pseudo Terminal device. It appears that there was an issue with the old TWAn
device which was scheduled to be resolved when Vue began to use the newer FTAn
devices. Although V3-IFT uses the FTAn device, this has not yet been changed.
I have entered QAR 469 as a reminder to the developer to look into this.
I have also successfully patched VUE$MASTER.EXE (offset 5c53c from 893b0 to
a93b0) which sets the nobroadcast characteristic for the created FTAn devices.
This eliminates the need for the $SET TERM/nobroadcast/vt200 in the file:
vue$library:vue$subprocess_init.com, and hence the resulting error messages.
(VUETKMPTY.LIS from the V3-IFT results disk)
; 0076 1 ! 13-Aug-1990 (krb) Back out the change. We would need to use IO$_SETCHAR
; 0077 1 ! on the setmode. We would need PHY_IO priv to do this.
; 0078 1 ! This will be fixed when we use the new FT device, so we
; 0079 1 ! will wait until then.
; 0080 1 ! 08-Aug-1990 (krb) Set the dev type and no broadcast when we create the TWA
; 0081 1 ! device. This avoids the set dcl commands in the subprocess
; 0082 1 ! _init.com file. These set commands were cause abort errors
; 0083 1 ! to come from the TW device during heavy load times on the
; 0084 1 ! system.
|