Re: Scwm docstrings change

["Greg J. Badros" <gjb at cs dot washington dot edu>]

Clark McGrew <mcgrew@ale.physics.sunysb.edu> writes:

<snip>

>     Greg> I've a lot of experience with the headaches and pains that
>     Greg> macros can cause programmers and tools, yet here I'm
>     Greg> absolutely convinced that the benefits far outweigh those
>     Greg> costs.
> 
> I think we all have experience with cpp headaches.  I've found that
> they usually happen long after the originator has left the project.
> It seems that the brighter the original author, the worse the problems
> tend to be in the long run.
> 
> Obviously you're going to do the work so you get the final vote.  I
> just worry that in 10 or 15 years, somebody is going to look at this
> code and have to figure it out.

Well, the work is now largely done, so you can now judge it for yourself
by updating from anoncvs.

My claim is that without these kinds of changes to Guile source, people
will *not* being looking at it in 10 or 15 years. Period.

Also, people don't look at code and try to figure it out.  People cut
and paste existing code.  My changes make that a less error-prone
ordeal, and when the checks are in place as they are for Scwm, will
automatically and statically report a large class of commonly-made
mistakes.

And the macros are no less understandable than lots of existing Guile
code.  Raising the level of abstraction and factoring out redunancy
helps readability and improves maintainability.  It's unfortunate that
with C a bunch of abstraction must be done w/ cpp, but it's still better 
than leaving the abstraction out.

Greg