000301 Klaus Weide wrote:
> On Tue, 29 Feb 2000, Philip Webb wrote:
>> since Lynx ^t doesn't help, i remembered strace
> Good idea!
>> read(0, "7", 1) = 1
-- snip --
> I assume this is from the 'P'rint Options page. What is your link #7?
/usr/local/bin/edit %s [ in its proper place in the line ]
-- snip --
>> write(1, "Printing file. Please wait...", 30) = 30
> Now lynx is going to invoke a command via its LYSystem() function.
> What command exactly depends on what link #7 is.
> The following 3 lines are under control of lynx code, see LYSystem() :
>> sigaction(SIGPIPE, {SIG_DFL}, {SIG_IGN}) = 0
>> sigaction(SIGTSTP, NULL, {0xfa6d8f8, [], SA_RESETHAND|SA_NODEFER}) = 0
>> sigaction(SIGTSTP, {SIG_DFL}, {0xfa6d8f8, [], SA_RESETHAND|SA_NODEFER}) = 0
> Then LYSystem() calls system() ,
> all the following happens within the system's system() implementation:
>> sigaction(SIGCHLD, {SIG_DFL}, {SIG_DFL}) = 0
>> fork() = 26308
> A child process is created to execute the external command.
> the parent Lynx process now just waits:
>> waitid(P_PID, 26308, <unfinished ...>
>> --- SIGINT (Interrupt) ---
> ... and gets a SIGINT signal before the child exists.
> Assuming you pressed ^g -- the child changed the tty settings
> so that ^G means intr (this would show up unter 'stty -a'
> as 'intr = ^G' , if you could look at it at it at this stage).
see below
>> <... waitid resumed> 0x7fffa138, WEXITED|WTRAPPED) = -1 EINTR (Interrupted system
>call)
>> --- SIGINT (Interrupt) ---
> ^G causes a SIGINT signal, the same as ^C normally,
> which gets delivered to the parent Lynx as well as the child.
> It causes the Lynx process to terminate besides what it does
> to the child process - probably terminate it, too).
> The problem is the system's system() implementation,
> it doesn't behave as it should. From a Linux system(2) man page:
> system() executes a command specified in string by calling
> /bin/sh -c string, and returns after the command has been
> completed. During execution of the command, SIGCHLD will
> be blocked, and SIGINT and SIGQUIT will be ignored.
-- FreeBSD snipped --
> The important part here is ignoring SIGINT.
> Your system() doesn't do that. What does its man page say?
your analysis is very clear & makes sense;
your other message is taken up further down.
there appear to be 2 versions of system() at CHASS,
neither of which says it ignores SIGINT.
SYSTEM(3F) Silicon Graphics SYSTEM(3F)
NAME system - issue a shell command from Fortran
SYNOPSIS character*N c
call system(c)
DESCRIPTION
system causes its character argument to begiven to sh(1) as input,
as if the string had been typed at a terminal.
The current process waits until the shell has completed.
SEE ALSO exec(2), system(3S). sh(1) in the User's Reference Manual.
ORIGIN AT&T V.3
system(3S) system(3S)
NAME system - issue a shell command
SYNOPSIS #include <stdlib.h>
int system(const char *string);
DESCRIPTION
system causes the string to be given to the shell [see sh(1)] as input,
as if the string had been typed as a command at a terminal.
The current process waits until the shell has completed, then returns
the exit status of the shell in the format specified by waitpid(2).
If string is a NULL pointer, system checks if /sbin/sh exists
and is executable. If /sbin/sh is available, system returns non-zero;
otherwise it returns zero.
system fails if one or more of the following are true:
EAGAIN The system-imposed limit on the total number of processes
under execution by a single user would be exceeded.
EINTR system was interrupted by a signal.
ENOMEM The new process requires more memory
than is allowed by the system-imposed maximum MAXMEM.
SEE ALSO exec(2), waitpid(2), sh(1).
DIAGNOSTICS system forks to create a child process
that in turn execs /sbin/shin order to execute string.
If the fork or exec fails, system returns -1 and sets errno.
>> if i enter ^g after starting Mutt from the Print screen,
>> but before Mutt calls the editor, the print request is cancelled.
> Is that how Mutt, on its own, is supposed to react to ^g?
if you start Mutt to send a message (cp Elm, if that's what you use),
it prompts eg for Cc: if you enter ^g there, it exits (outside Lynx).
this is probably not important, as there's more direct evidence below.
> Since you are using Screen, you sb able to verify
> that while Mutt is in control, intr = ^G is in effect:
> 1. Open two screen sessions -- snip --
strictly, you need only 1 session, surely.
anyway, i did as you suggested, testing my editor & Most in/outside Lynx.
the editor is a very slick little version of Emacs
-- it contrasts with Emacs/Vi as Lynx does with IE/NN -- ,
which is available only here (someone wrote it long ago),
but you use Most, so maybe we should concentrate on that.
there were no differences in output from stty -a
for pseudo-prompt, Print screen & 1st Mutt prompt (asks for filename);
there was a tiny one after 2nd Mutt prompt ( istrip -> -istrip );
the editor had the same output invoked directly from the Print screen
& invoked by Mutt which had been invoked from the Print screen;
otherwise, i got the following diffs:
(1) pseudo-prompt vs Most from Print screen
2,3c2,3
< intr = ^C; quit = ^\; erase = ^H; kill = ^U; eof = ^A; eol = ^@; old-swtch = ^Z;
susp = ^Z
< lnext = ^V; werase = ^W; rprnt = ^R; flush = ^O; stop = ^S; start = ^Q; dsusp = ^@
---
> intr = ^G; quit = ^@; erase = ^H; kill = ^U; eof = ^A; eol = ^@; old-swtch = ^@;
>susp = ^@
> lnext = ^@; werase = ^W; rprnt = ^R; flush = ^O; stop = ^S; start = ^Q; dsusp = ^@
7,8c7,8
< isig -icanon -xcase -echo echoe echok echoke echoctl -echoprt -echonl -noflsh
-flusho -pendin
< opost -olcuc -onlcr -ocrnl -onocr -onlret -ofill -ofdel
---
> isig -icanon -xcase -echo -echoe -echok -echoke -echoctl -echoprt -echonl noflsh
>-flusho -pendin
> opost -olcuc onlcr -ocrnl -onocr -onlret -ofill -ofdel tab3
(2) pseudo-prompt vs Edit from Print screen
2c2
< intr = ^C; quit = ^\; erase = ^H; kill = ^U; eof = ^A; eol = ^@; old-swtch = ^Z;
susp = ^Z
---
> intr = ^G; quit = ^G; erase = ^H; kill = ^U; eof = ^A; eol = ^@; old-swtch <undef>;
>susp <undef>
5,7c5,7
< -ignbrk brkint ignpar -parmrk -inpck istrip -inlcr -igncr -icrnl -iuclc
< ixon ixany -ixoff -imaxbel
< isig -icanon -xcase -echo echoe echok echoke echoctl -echoprt -echonl -noflsh
-flusho -pendin
---
> ignbrk brkint ignpar -parmrk -inpck -istrip -inlcr -igncr -icrnl -iuclc
> -ixon ixany -ixoff -imaxbel
> isig -icanon -xcase -echo echoe echok echoke -echoctl -echoprt -echonl -noflsh
>-flusho -pendin
(3) pseudo-prompt vs Edit outside Lynx
2,3c2,3
< intr = ^C; quit = ^\; erase = ^H; kill = ^U; eof = ^A; eol = ^@; old-swtch = ^Z;
susp = ^Z
< lnext = ^V; werase = ^W; rprnt = ^R; flush = ^O; stop = ^S; start = ^Q; dsusp = ^@
---
> intr = ^C; quit = ^\; erase = ^H; kill = ^U; eof = ^A; eol = ^@; old-swtch = ^@;
>susp = ^@
> lnext = ^@; werase = ^@; rprnt = ^@; flush = ^O; stop = ^S; start = ^Q; dsusp = ^@
5c5
< -ignbrk brkint ignpar -parmrk -inpck istrip -inlcr -igncr -icrnl -iuclc
---
> -ignbrk brkint ignpar -parmrk -inpck istrip inlcr -igncr icrnl -iuclc
7,8c7,8
< isig -icanon -xcase -echo echoe echok echoke echoctl -echoprt -echonl -noflsh
-flusho -pendin
< opost -olcuc -onlcr -ocrnl -onocr -onlret -ofill -ofdel
---
> isig -icanon -xcase -echo echoe -echok echoke echoctl -echoprt -echonl -noflsh
>-flusho -pendin
> opost -olcuc onlcr -ocrnl -onocr -onlret -ofill -ofdel tab3
(4) pseudo-prompt vs Most outside Lynx
2,3c2,3
< intr = ^C; quit = ^\; erase = ^H; kill = ^U; eof = ^A; eol = ^@; old-swtch = ^Z;
susp = ^Z
< lnext = ^V; werase = ^W; rprnt = ^R; flush = ^O; stop = ^S; start = ^Q; dsusp = ^@
---
> intr = ^C; quit = ^\; erase = ^H; kill = ^U; eof = ^A; eol = ^@; old-swtch = ^@;
>susp = ^@
> lnext = ^@; werase = ^@; rprnt = ^@; flush = ^O; stop = ^S; start = ^Q; dsusp = ^@
5c5
< -ignbrk brkint ignpar -parmrk -inpck istrip -inlcr -igncr -icrnl -iuclc
---
> -ignbrk brkint ignpar -parmrk -inpck istrip inlcr -igncr icrnl -iuclc
7,8c7,8
< isig -icanon -xcase -echo echoe echok echoke echoctl -echoprt -echonl -noflsh
-flusho -pendin
< opost -olcuc -onlcr -ocrnl -onocr -onlret -ofill -ofdel
---
> isig -icanon -xcase -echo echoe -echok echoke echoctl -echoprt -echonl -noflsh
>-flusho -pendin
> opost -olcuc onlcr -ocrnl -onocr -onlret -ofill -ofdel tab3
--
========================,,============================================
SUPPORT ___________//___, Philip Webb : [EMAIL PROTECTED]
ELECTRIC /] [] [] [] [] []| Centre for Urban & Community Studies
TRANSIT `-O----------O---' University of Toronto
