Bug in gnome-terminal 1.0.53

[Alexander Volovics <awol xs4all nl>]

Gnome-terminal 1.0.53 crashes one of my apps: the statistical program
Stata for Linux.

Stata runs, command driven, in a terminal.
It works without any problems in an xterm, a rxvt and in Eterm.
It also worked without problems in older versions of gnome-terminal.

However when I startup Stata in a gnome-terminal it opens,
shows the welcome screen and appears to be waiting for commands
at the Stata comand prompt.
However when you give a command the bash shell answers that the
command does not exit. Stata has already silently crashed.

I can find no evidence of problems in log files. I tried gdb
but couldn't get it to give me any instructive output
(I am not a programmer).

I then tried strace. Strace does seem to indicate something strange
happening when using gnome-terminal to run Stata.

I first give the result of strace running Stata in an Eterm where
I have no problems. I only present the part of a strace dump where
something relevant appears to happen. The output from strace stops
at the point that I have marked with a ^ in an otherwise empty line.
Stata is now waiting at it's command prompt.
Then, as can be seen, I exit Stata by giving the `exit' command
and strace continues to the end.

write(1, "\377\3000\n\376\nNote:  Your site can add m"..., 148) = 148
fstat(0, {st_mode=S_IFCHR|0666, st_rdev=makedev(3, 0), ...}) = 0
mmap(0, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) =
0x40013000
ioctl(0, TCGETS, {B38400 opost isig -icanon -echo ...}) = 0
read(0, "e", 4096)                      = 1
       ^
write(1, "e", 1)                        = 1
read(0, "x", 4096)                      = 1
write(1, "x", 1)                        = 1
read(0, "i", 4096)                      = 1
write(1, "i", 1)                        = 1
read(0, "t", 4096)                      = 1
write(1, "t", 1)                        = 1
read(0, "\n", 4096)                     = 1
write(1, "\n", 1)                       = 1
write(1, "\377\3005\n\376", 5)          = 5
lseek(4, 48, SEEK_SET)                  = 48
read(4, "^\371\0358\1\0\0\0localhost.localdoma\0wolo"..., 48) = 48
lseek(4, 48, SEEK_SET)                  = 48
write(4, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 48)
= 48
close(4)                                = 0
write(1, "\33[?1l\33>", 7)              = 7
rt_sigaction(SIGINT, {SIG_DFL}, {0x804fbf0, [], SA_RESTART|0x4000000}, 8)
= 0
rt_sigaction(SIGFPE, {SIG_DFL}, {0x804fc0c, [], SA_RESTART|0x4000000}, 8)
= 0
ioctl(0, TCSETA, {B38400 opost isig icanon echo ...}) = 0
write(1, "\377\261\376", 3)             = 3
_exit(0)                                = ?


I now give the result of strace when running Stata in gnome-terminal.
I start at the same point in the strace dump. But with gnome-terminal
Stata does not stop to wait on commands at the point marked with ^.
"Something" (Stata/gnome-terminal) seems to be reading an empty 
string/expression and Stata exits automatically.

write(1, "\377\3000\n\376\nNote:  Your site can add m"..., 148) = 148
fstat(0, {st_mode=S_IFCHR|0620, st_rdev=makedev(136, 0), ...}) = 0
mmap(0, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) =
0x40013000
ioctl(0, TCGETS, {B38400 opost isig -icanon -echo ...}) = 0
read(0, "", 1024)                       = 0
       ^
write(1, "\377\3005\n\376", 5)          = 5
lseek(4, 48, SEEK_SET)                  = 48
read(4, "0\370\0358\1\0\0\0localhost.localdoma\0wolo"..., 48) = 48
lseek(4, 48, SEEK_SET)                  = 48
write(4, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"..., 48)
= 48
close(4)                                = 0
write(1, "\33[?1l\33>", 7)              = 7
rt_sigaction(SIGINT, {SIG_DFL}, {0x804fbf0, [], SA_RESTART|0x4000000}, 8)
= 0
rt_sigaction(SIGFPE, {SIG_DFL}, {0x804fc0c, [], SA_RESTART|0x4000000}, 8)
= 0
ioctl(0, TCSETA, {B38400 opost isig icanon echo ...}) = 0
write(1, "\377\261\376", 3)             = 3
_exit(0)                                = ?

I am sorry this mail is a bit long but I do not know how to express
the problem more compactly. I do not have enough `hard evidence' to
make an official bug report. I do not even know if this is a real bug
or if strace shows anything usefull at all.

I hope somebody in the know will take a look and make suggestions
for obtaining more details. I always find it very irritating when
things happen that I do not understand.

I am using RH 6.1 with updates and october gnome.

Alexander


-------------------------------------
Alexander Volovics
Dept of Statistics & Methodology
Maastricht University, Maastricht, NL
email: a.volovics@stat.unimaas.nl
-------------------------------------