🚀 go-pugleaf

In article <7978@baird.cs.strath.ac.uk>, amurphy@cs.strath.ac.uk (Allan Murphy CS4) writes:
> Der Mouse writes:
>> I wrote:
>>> Is there a standard technique / bit of code that X applications are
>>> supposed to carry out when the user specifies a display other than
>>> $DISPLAY ?
>> I haven't seen anything noteworthy - just use the specified string
>> in the XOpenDisplay call instead of the default.
> What I didn't write [...] was that the application is a graphics demo
> for the networked machines in the CS department here; all of these
> machines use the same file server; you can login to any of them with
> the same username/password.....
[I've tried to cut this down to something not too huge without doing
too much violence to Allan's explanation.  I hope I succeeded....]
> Since my app. is unlikely to be ported ( very unlikely ), this might
> ( covers his ass ) be a safe thing to do ??
["this" refers to taking a machine name and appending ":0.0".]
> This a graphics demo for twerps; none of them are going to try run it
> on a display connected to a machine outside this department.... they
> are liable to try something like % demo -display <machine>.

If the argument contains no colons, appending :0.0 would not strike me
as terribly unreasonable, particularly if the given string is a valid
hostname (this may not be worth checking for).

However, I would argue that doing this for a -display argument is a bad
idea because it introduces an inconsistency: very few other programs
that take -display will accept a bare machine name, so getting your
`twerps' used to using a bare machine name after a -display is liable
to confuse them the first time they run into a mainstream X application
with a -display argument.

If you want to use a different flag (-remote or something, maybe), and
document it as the remote machine rather than as the display name, I
would have no problems with what you suggest.

Well, almost none.  While you may now think that your code is unlikely
to be ported, there are at least two reasons for doing it the Right Way
anyway: (a) much code that was never intended to be ported ends up all
over the place - witness UCB talk(1), and (b) it strikes me as a bad
habit to get into.  ((b) is rather ill-defined; it's an aesthetic
judgement....)

>>>   rsh to the remote machine and do an xhost +local machine name
>> ( embarrassingly large list of problems with this idea removed in
>>   order to preserve my ego by not having them posted twice )

Well, the idea wasn't to do violence to your ego!  It just sounded as
though you hadn't really thought about it in much depth; it looked like
the sort of thing that someone with the fabled "little learning" might
suggest - an instance of the "it works on my machine" problem.  Your
explanation makes it clear that (whether or not you did think about it
in depth :-) it's not nearly as inappropriate to your situation as my
list of problems may have made it sound.

					der Mouse

			old: mcgill-vision!mouse
			new: mouse@larry.mcrcim.mcgill.edu