[Clustersim] Re: ClusterSim Mailing Membership List

[John Schneiderman]

On Thursday 29 June 2006 23:56, Brian Moore wrote:
> John,
> I've been looking at the CVS code for our program again.  I was able to
> download it, install it, and run it, so that is good.  From what I can tell
> it appears to be running correctly, I don't see any obvious bugs.  One
> question I have is if you just run ./configure and make (but not make
> install), it appears that the executable shows up in the src directory.  Is
> that right?  
Yeah, that's currently the default directory that the make file compiles the 
binary to. We could change that, and probably should.
> The other thing is that I had to hunt for a config file to 
> start it up (again without doing make install).
In the INSTALL file I gave directions on where the config file was, maybe I 
should be more explicit about it. What do you think?
>
> But, a bigger problem is this.  I've been looking at the code and I still
> think there are big parts of it I don't understand at all.  This combined
> with the fact that we only haltingly understand how to use the CVS (we
> struggled with Cervisia today) means that I think it is unlikely we will
> realistically use the current CVS version for our day-to-day work from now
> until the end of the summer.  Even me understanding my OWN program is
> sometimes difficult!
>
> I like the program from the point of view of the user.  You greatly
> improved the format of the configuration files, etc.  But as I am working
> with people some of whom have no previous training in programming at all, I
> think that the leap may be too big to try to understand all the actual code
> we have in the clustersim program.
>
> We should talk more.  One solution is we create another project, and call
> it something like beta or like that.  This could have the program in its
> current state, as we are working with it, with none of the
> autoconf/automake stuff (at first at least), and just the code with a
> custom Makefile.  The understanding would be that this second project is
> NOT a release version, but it would have some of the new ideas, as we
> create them.
>
> The other possibility is that we perhaps just run a local CVS for the beta
> version, and maybe include a message in the header files on the clustersim
> page to email us for the beta version if they want it.
>
> Another possibility is that you come and visit a few more times and we work
> through the blood and guts of getting me and David to understand everything
> about the program as it is now.  Maybe we can take a few core ideas at
> least.  I really liked the idea of using a map for instance, I have never
> used them before, etc.
>
I like this option the best. I have pleanty of time to come in, and I would be 
more than happy to help. I care about our project. 

> Oh, and a final complication is that I am contemplating using the GNU
> Scientific Library in some of the new revisions.  Obviously, getting the
> GSL is not hard, but it would mean that our code would not compile directly
> our of the cvs without the gsl.
>
This is true, but what advantages do you see from using the GSL?
> Let me know what you think, lets talk more.  I greatly appreciate all the
> work you've done so far.
Not a problem, just let me know what days you have available that you want me 
to come in.

John

pgpp9FWJuaUx3.pgp
Description: PGP signature