Why aren't these Disney Python jobs for Squeak?

Bert Freudenberg bert at isg.cs.uni-magdeburg.de
Fri Sep 22 21:44:03 UTC 2000


On Fri, 22 Sep 2000, Paul Fernhout wrote:

> But there is a much bigger design issue here. What is a headless Squeak
> doing connecting to an X-server? 

It is just checking if it is connected. You're right, the SetCursor stuff
shouldn't be called (but see below). But I'm sure such checks are in the
VisualWorks VM, too, because some things depend on a display being opened
or not.

> While the patch you suggest may work, the deeper issue is that modular
> code for headless operation should never require such a patch. [...]

What you imply is true, and it would be great if someone would stand up
and implement it. OTOH, the dumb approach works right *now* - it's the
simplest thing that could possibly work. The image doesn't even know it
has no head. It has a Display bitmap as usual where it displays stuff and
just happens to work with a rather idle user. The immediate advantage is
that you use identical images for headed and headless operation.

I'm not saying this is a good solution for anything, but until now, it was
sufficiently good so noone did anything general about it. [Actually, Tim
Rowledge did: he modified an image and his Acorn VM so no window is opened
until the beDisplay primitive is called.]

Besides, for me, Squeak is an inherently visually oriented environment
(even more than VisualWorks). I don't see a compelling reason to write
large-scale *headless* server stuff in it. It surely is no fun to debug. I
even run both my Swikis in a VNC server so I can have a look at them if
needed. Even from my home over an ISDN line. It's so much fun to hack in a
running server image :)

-- Bert





More information about the Squeak-dev mailing list