Re: lynx-dev strange HTTP/HTCheckForInterrupt() bug

[Klaus Weide]

On Mon, 29 Mar 1999, Klaus Weide wrote:

> I get the following, very similar to yours, by starting
> 
>      no_proxy='*' lynx -dump -tlog http://localhost -trace 2>&1|less

I can also get the effect (lynx apparently unable to resolve "localhost")
just with

    no_proxy='*' lynx -dump http://localhost [RET]
    [RET]
    [RET]
    ....

(Hitting Enter many times or just autorepeat).

It sometimes happens with

   Exiting via interrupt: 15

followed by

   lynx: Can't access startfile http://localhost/

and sometimes only

   lynx: Can't access startfile http://localhost/

I assume the first case happens when the child process doesn't
get scheduled to run at all before the parent kills it (so that
the signal(SIGTERM,quench) never gets executed).  That's the real
problem, since in an interactive session the child process will
try to change the tty modes etc.

The NSL_FORK loop could be changed to somehow not 'count' keypress
event, but that looks like only treating the symptoms.

The only sure thing to prevent the child process from being killed
before it gets a chance to do something: do something with the signal(s)
(change handler, or blocking) *before* the fork().  Ugly because it has
to be undone in the parent after the fork.  Any better ideas?

I still don't understand why something like this would occur in a normal
interactive session, unless the user starts wildly pressing keys instead
of just one 'z'.

    Klaus