FAQS/follow-up on solving a set of symbolic equations

I agree that C1/D1 are annoying and cause problems.

C1/D1 are handy, and making them hard to type would reduce their
usefulness.

In999 and Out999 are easy enough to type, yet not likely to be typed by
mistake.  %C999 is possible; the possibility of confusion with %C[]()
(Tchebyschev polynomials, I think) and %D[]() (some special function, I
don't know which!) is remote.  I don't especially like C_999.

% means the last value calculated, and _ means the last input, so %999
and _999 would be logical.  But %999 are currently used by Optimize
(OPTIMPREFIX) and the tensor package.  And I think they look a bit
obscure.

The cleanest would be In[999] and Out[999], since the linear structure
wouldn't be hidden in the name, but perhaps that is too much typing --
and too much space taken up at the beginning of every line.  (It also
requires more than just changing inchar and outchar).  Since we
currently don't have that, there is %TH(n), the n'th-last result.

%[999] would not be a good idea for D999, since it is often useful to
select the n'th element of the last result.

Well, that's the lay of the land.  What is my preference?  Well, my
preference would be a completely different interaction paradigm, but
let's leave that aside.  I don't think we need to worry about
compatibility with published material, commercial Macsyma, etc., but I
*do* think our own documentation should follow whatever the new
convention is.  It is not completely trivial to do this.

Given the current setup, I agree with CY that %C999/%D999 or
In999/Out999 would be better as a default.  I would also recommend
%Tmp999 for temporaries (OptimPrefix).  Perhaps %Ex999 for intermediate
expressions (LineChar).

       -s