Re: lynx-dev COLOR WHITE and Ctl-C

[Gisle Vanem]

Bob Blackledge <address@hidden> said:

> On previous message, forgot to include ^C response:
> Exiting via interrupt: 295
> pcpkt.c (1027): _pkt_inf-NULL
> pcpkt.c (1027): _pkt_inf-NULL
> and then at DOS prompt.

"_pkt_inf == NULL" simply means the packet-driver was deinstalled
before all sockets where properly closed; Lynx/DOS needs to send a
RST if not connection was fully closed. That "someone" is probably
stop_curses(). IMHO stop_curses() should call _eth_release() only
if exiting to it's parent and not to a DOS-shell.

> Separately, ^C (or ^Break) exits LYNX to DOS when long
> DNS lookup (or other lengthy activity) in process...
> "z" does nothing in such circumstances.  Doc-files
> say that these should cancel work-in-process...?

Watt-32 does have a hook for just this. I.e. a hook-routine could be
called while DNS lookups are pending completion (or times out). Look
at '_resolve_hook' in the Watt-32 sources. I'm looking into adding
a patch for this.

Doug Kaufman <address@hidden> said:

> During long DNS lookup, CTRL-C should get you back to lynx, asking
> if you want to quit. Answering "n" should leave you in lynx. If
> you type CTRL-C when not in DNS lookup, you will exit lynx to DOS.
> This behavior is in the WATT-32 code, rather than in the lynx code
> itself.

I cannot see how this could be. I'm compiling with S-Lang, but shouldn't
matter. My version behaves as documented. Watt-32 never calls exit()
when SIGINT is raised. Seems that somehow "signal (SIGINT, SIG_DFL)" is
in effect or you didn't compile Lynx with "IGNORE_CTRL_C". If the latter
case, the exit is intended.

Gisle V.





; To UNSUBSCRIBE: Send "unsubscribe lynx-dev" to address@hidden