On Sun, Nov 03, 2002 at 07:50:42PM -0500, Sergey Okhapkin wrote: > >----- Original Message ----- >From: "Christopher Faylor" <[EMAIL PROTECTED]> >To: <[EMAIL PROTECTED]> >Sent: Sunday, November 03, 2002 7:37 PM >Subject: Re: fhandler_tty patch > >> > * fhandler_tty.cc (fhandler_tty_slave::ioctl): Do nothing if the new >> > window size is equal to the old one. Send SIGWINCH if slave connected >> > to a pseudo tty. >> > (fhandler_pty_master::ioctl): Do nothing if the new window size is >> > equal to the old one. >> >> Is this according to some standard? It seems like we're sending too many >> SIGWINCHes with your patch. >> > >Without the patch we're not sending SIGWINCH at all. ioctl(tty, TIOSWINSZ, >...) supposed to send SIGWINCH if the window size changed. The ioctl() call >should work in the same way for both master and slave ends of pseudo tty, >without the patch the ioctl works for master end only, but many unix >programs (screen for example) change the window size of the slave end.
Ok. I haven't looked closely at this code before. I just am not sure what use setting the size on a pty master would entail. I wonder if it is supposed to be propagated to the slave. Also, isn't the master supposed to be notified of size changes in the slave? I'll apply the patch but I suspect that this is still not exactly right wrt UNIX emulation. cgf
