Re: GraphElem methods return values changed?

[Paul Johnson]

Rick Riolo wrote:
> I think there was a change going from Swarm 2.1.1 to 2.2x, 
> namely that this won't work anymore:
>
>   C1GraphElem = [CGraph createElement];
> * [[[C1GraphElem setLabel: "NumInteractions"] setColor: "green"] setWidth: 
> 1];
>
> because those methods to the GraphElem don't return self anymore,
> is that right?
> I didn't see this mentioned in the NEWS that comes with the distribution,
> but maybe I missed it.

Right.  You have to rewrite it like

    C1GraphElem = [CGraph createElement];
    [C1GraphElem setLabel: "NumInteractions"];
    [C1GraphElem setColor: "green"];
    [C1GraphElem setWidth: 1];

This was done to expose more things to the Java interface, but I  do not 
know precisely why.

> By 2.2x i mean one installed from rpm's in:
>     http://lark.cc.ukans.edu/~pauljohn/Swarm/Swarm-2.2x_RPMSforRH8.0/
>
> (Paul---Thanks much for the rpm's! 
>   btw, is there a way to install from the rpm's to a location
>   we choose?  (e.g. a filesystem we have mounted by many machines)
> )
No relocatability in the more recent ones.  Alex Lancaster had worked 
out a way to post process the installed files to allow relocation, but 
somewhere along the line that broke and I got too frustrated with it to 
make it work.

The problem was that libtool broke in the ability to build shared 
libraries that would be installed to a different place than the eventual 
target and I went through hell and back trying to track that down. 
Fixing it took a lot of questions to the libtool and RPM groups, which 
indicate "yes, that's a problem" and "try to hack your source to make 
the DESTDIR option work".

I got it to work in that patched Swarm 2.1.141 I used for the RH8 rpms, 
but I am afraid to check all those changes into the Swarm CVS for fear 
of causing a mistake on some other platform.  The diff showing all the 
changes I made to solve this and a couple of other problems is here:

http://lark.cc.ku.edu/~pauljohn/Swarm/Swarm-2.2x_RPMSforRH8.0/PJswarm141.patch

If you care to review it, perhaps you can tell me if you think something 
is likely to break on other systems.  The DESTDIR changes all appear 
harmless to me. The hack to ltmain.sh is something I got from the 
libtool user email  list and it appears to work, but I don't know why.

One problem we have is that, as of gcc-2002-09-09, gcc does much more 
anal type checking and protocol checking, so I went through and had to 
change a bunch of local variables from unsigned to int and vice versa. 
Those ones appear pretty safe too.

The most suspicious hack I used is on Customize to make sure the 
compiler finds a +createBegin method for objects subclassed from 
Customize. I say "hack" meaning it seemed logical to me and made 
everything work, but I suspect there is a better way to fool the 
compiler.  Here was my approach:

Index: src/defobj/Customize.m
===================================================================
RCS file: /cvsroot/swarm/swarm/src/defobj/Customize.m,v
retrieving revision 1.22
diff -r1.22 Customize.m
96a97,104
>
> + createBegin: aZone
> {
>   [self subclassResponsibility: @selector(createBegin:)];
>   return nil;
> }
>


But relocatability will probably reappear as soon as the RH8 issues are 
all ironed out.


--
Paul E. Johnson                       email: address@hidden
Dept. of Political Science            http://lark.cc.ku.edu/~pauljohn
University of Kansas                  Office: (785) 864-9086
Lawrence, Kansas 66045                FAX: (785) 864-5700


                 ==================================
  Swarm-Support is for discussion of the technical details of the day
  to day usage of Swarm.  For list administration needs (esp.
  [un]subscribing), please send a message to <address@hidden>
  with "help" in the body of the message.