Bugs item #419510, was opened at 2001-04-27 11:47
You can respond by visiting:
https://sourceforge.net/tracker/?func=detail&atid=104664&aid=419510&group_id=4664
Category: Curses interface
Group: None
Status: Open
Resolution: Fixed
Priority: 5
Submitted By: Simon Trimmer (vxsimon)
Assigned to: Hans-Bernhard Broeker (broeker)
Summary: cscope exits before errors can be read
Initial Comment:
in error conditions cscope will often postmsg() and
then exit.
The problem is the exit code clears the screen so the
user experience
will be cscope suddenly exits without warning.
----------------------------------------------------------------------
Comment By: Marty Leisner (leisner)
Date: 2003-02-23 14:35
Message:
Logged In: YES
user_id=17321
I think you can register something with atexit, and
cache fatal errors someplace to come out on the screen
after curses is done...
I've also seen a problem the the tty is in a weird state
if flex fails...atexit() would solve this problem
too...
----------------------------------------------------------------------
Comment By: Hans-Bernhard Broeker (broeker)
Date: 2003-02-23 06:52
Message:
Logged In: YES
user_id=27517
OK, then, I agree. How about a new function 'postfatal'
that replaces sequences of postmsg() / posterr() and
myexit(), and instead does exitcurses(), prints the message
to stderr, and *then* calls myexit().
Any volunteers, or will I have to do this myself?
----------------------------------------------------------------------
Comment By: Marty Leisner (leisner)
Date: 2003-02-22 18:29
Message:
Logged In: YES
user_id=17321
I'm using the cvs version.
I had a similar problem.
On redhat 8.0 --
rxvt works
xterm doesn't work with TERM=xterm
xterm works with TERM=vt100
But I'm not sure I'm thrilled with the error strategy -- its
hidden in the main window -- probably it would be a good
idea to print fatal error messages after curses is exited.
----------------------------------------------------------------------
Comment By: Simon Trimmer (vxsimon)
Date: 2002-07-30 12:22
Message:
Logged In: YES
user_id=120747
True, but you can about broken source navigators :)
I guess this is down to endwin restoring the previous window
state and that if we actually wanted the message to be
visible it'd print it out afterwards.
----------------------------------------------------------------------
Comment By: Hans-Bernhard Broeker (broeker)
Date: 2002-07-29 10:46
Message:
Logged In: YES
user_id=27517
We can't really do much about broken terminal emulators, or
terminfo entries that don't work with the actual terminal
emulator.
The CVS version might still be better for you than 15.3 ---
could you please try it out?
----------------------------------------------------------------------
Comment By: Simon Trimmer (vxsimon)
Date: 2002-07-29 10:27
Message:
Logged In: YES
user_id=120747
redhat 7.3 (with updates applied) cscope 15.3
Your remark prompted some thought and experimentation; it
appears to be a termcap problem, rxvt and gnome-terminal
show the problem whilst xterm and the linux console don't.
The exiting escape codes look like:
xterm: (okay)
cscope: cannot open file
cscope.out^M^[[24d^[[24;1H^[[?1047l^[[?1048l^M^[[?1l^[>
rxvt: (dark)
cscope: cannot open file
cscope.out^M^[[22B^[[24;1H^[[2J^[[?47l^[8^M^[>
I don't have the specs handy, but I'll attach the rxvt
capture (it replays in
an xterm)
----------------------------------------------------------------------
Comment By: Hans-Bernhard Broeker (broeker)
Date: 2002-07-29 09:58
Message:
Logged In: YES
user_id=27517
Hmmm... doesn't seem to happen on this box here (Alpha
running Tru64 Unix and its OSF/1 curses library, remotely in
an xterm). The error message ends up near the top of the
windows, but it's there and readable.
What platform and cscope version did you test it on?
----------------------------------------------------------------------
Comment By: Simon Trimmer (vxsimon)
Date: 2002-07-29 06:37
Message:
Logged In: YES
user_id=120747
No this isn't fixed - as a simple test case run cscope -d in
a directory with no cscope database, the screen blinks but
you can only see the message if you capture the session with
script and view it afterwards.
----------------------------------------------------------------------
Comment By: Hans-Bernhard Broeker (broeker)
Date: 2002-07-28 12:17
Message:
Logged In: YES
user_id=27517
I expect the change from calling exit() to calling myexit()
in all places (except myexit() itself...) to have fixed
this. Or if it didn't, this bug is way too old, and may no
longer apply to the current version anyway. I'm therefore
closing this as "Fixed".
----------------------------------------------------------------------
You can respond by visiting:
https://sourceforge.net/tracker/?func=detail&atid=104664&aid=419510&group_id=4664
|