Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!tut.cis.ohio-state.edu!ucbvax!hplabs!hpl-opus!hpccc!wescheng
From: wescheng@hpccc.HP.COM (Wesley Cheng)
Newsgroups: comp.windows.x
Subject: Re: Switching Widget Sets
Message-ID: <7070005@hpccc.HP.COM>
Date: 19 May 89 17:06:57 GMT
References: <8905171741.AA26738@crdgw1.ge.com>
Organization: HP Application Support Division
Lines: 25

> to perform in the future, depending on how tedious it is.  At DECUS in
> Atlanta, I asked several people who had every right to answer the
> question accurately, how difficult it would be to switch between the DEC
> Widget set and any other widget set.  The answer I got in each case was
> that it was at best non-trivial and at worst slower than re-coding from
> scratch.  It was difficult enough that none of them had previously seriously
> considered doing it.

The Athena, DEC and HP Widget Sets are all based on the Xt Intrinsics.
The underlying Xlib, OO paradigm (creation/destruction of widgets, etc),
translation management, do not change at all. My experience with Athena
and HP widgets tells me that it might not be so bad.  I think the trouble
comes when we grow dependent on the behavior of certain widgets that do
not exist in other widget sets. I have not found it very difficult to 
get the behavior I need by adding translations or writing my own widgets
(once you write one, it's simple). The Motif widgets (DEC) provide a lot
of functionality so it is unlikely one would need to write widgets for
"standard" GUI functions.

I believe that if the design is guided by the Xt Intrinsics, then switching
would not be a major disaster. Many people will be switching to the Motif
Widgets and I don't expect them to recode their applications from scratch
(at least I don't intend to).

wesley