[Freeciv-Dev] Re: jjm: Multi-client configure support added.

[David Pfitzner <dwp@xxxxxxxxxxxxxx>]

Jeff Mallatt wrote:

> At 1999/12/28 06:41 , David Pfitzner wrote:

> >It used to be that the "best guess" would use Xaw3d if available,
> >in preference to plain Xaw (if Gtk+ not available).  The new setup 
> >doesn't seem to do this.  I think it would be good to do something
> >like this, to try to reduce the occurence of Xaw variants problems.
> 
> I figured that the vanilla Xaw would be less inclined to "Xaw variants
> problems" than Xaw3d.

My understanding is that for Xaw, two common system 
configurations are:
  1. have plain Xaw only;
  2. have Xaw and Xaw3d.
In case 1, configure with fail to find Xaw3d (if it looks), so 
will only build as plain Xaw, and will work fine. 

In case 2, configure will "find" Xaw, and compile, but 
typically will actually link to Xaw3d, and will segfault at 
runtime.  In this case configure will also find Xaw3d if it
explicitly looks for Xaw3d, and if compiled and linked 
explicitly for Xaw3d, everything will work fine. 

So its better to use Xaw3d explicitly if it is found.

> >Wouldn't it be more consistent for "--with-xaw3d" to become
> >"--enable-client=xaw3d"?  (Which internally would set the client 
> >to xaw and turn on Xaw3d define etc.)
> 
> I guess it depends upon your point of view.  "xaw" is a client (it has a
> subdirectory in ./client), but "xaw3d" is just a variant (it does not have
> a subdirectory in ./client).  It's more like "--with-efence". 

Yeah, thats what its like when you know the code details,
but for a user it may be more like: Ok, there is a choice
of Gtk+, or Xaw, or Xaw3d (even though clearly Xaw and Xaw3d
are closely related).

-- David