Samba on VMS

[Luke Kenneth Casson Leighton]

john,

you need to make sure you do a select() on stdin or equivalent, because
you want to check whether smb keep-alives are coming in as _well_ as
keyboard input.  that's what wait_keyboard() is for, and i don't think
readline usage copes with this!

yep, that's a bug :)

On Sun, 2 Jan 2000, John E. Malmberg wrote:

> Luke Kenneth Casson Leighton <lkcl at samba.org> wrote:
> > John Malmberg wrote:
> > >
> > > RPCCLIENT will require a patch.  It does not have LIB_READLINE support,
> and
> >
> > neither does slackware 3.4 or irix.  what gives?  (i'm now using debian
> > whcih TOTALLY rocks:
> 
> I do not know.  I have to go look up what the real specification for it as
> it is not part of The Open Group Official Unix standard.  I did not know
> exactly where to look for it, but now I know to look on a mirror for debian,
> thanks!
> 
> LIB_READLINE is not supplied with OpenVMS, but the native I/O library
> supports Read with prompt and single line editing with one simple I/O call.
> 
> The documentation describes how to implement a history function, but that
> has a slightly more complex setup.
> 
> I implemented one based on the arguments passed to the routine, quickly.  So
> I have a history function that records what lines are typed, but it is not
> linked to the readline routine yet.
> 
> I do not know if I will implement saving the history lines to a file.
> OpenVMS users generally do not expect that behavior.  I am guessing that the
> LINUX implementations do though.
> 
> > > OpenVMS does not implement the extension to the socket library that
> allows
> > > socket calls to be used on non-socket channels.
> >
> > um.... such as select() on stdin?
> 
> Exactly.  That is not portable.  A maintainer of Perl for OpenVMS has told
> me that a new official standard specification for the socket library is in
> the works to address such things.  I do not have a URL for any independent
> reference for the socket library, either new or old.  But the fact is, for
> me it does not work, and while implementing a wrapper routine could be done,
> it is a bit messy.
> 



lok at samba-cvs SAMBA)TNG, i have a n add_chars_to_aray /
free_char_array(0 wrapper set.  really simple, but you any idea _how_ many
times you end up creating char**argv, argc and other character arrays, and
then having to free them?

> That is why the main VMS port patches SMBCLIENT there, but implementing
> LIB_READLINE eliminated that patch for me.
> 
> -John
> wb8tyw at qsl.net
>