Re: merging gpc into egcs

[Jeffrey A Law <law at upchuck dot cygnus dot com>]

  In message <19990526105517.A26416@anjala.mit.edu>you write:
  > >   > 1. Put the set code into expr.c. Disadvantage is that all languages g
  > et it.
  > > But it may be the right thing to do.  I've stated many times that we need
  >  to
  > > sit down and sort through how Chill and Pascal handle sets and determine 
  > if
  > > they are compatible enough to share or whether we need different
  > > implementations.
  > 
  > Pascal rounds down the lower bound and rounds up the upper bound to a
  > multiple of set_alignment. I don't know what advantage this gives over
  > simply starting from the lower bound and rounding up the size to a
  > multiple of set_alignment.
This doesn't really answer the question though.

In what ways do the pascal implementation of sets differ from the  chill
implementation of sets?

We need to know that before we can decide if we need a new tree node for
pascal sets.


Maybe you just were not clear.  Were you trying to say something like this?

  Pascal rounds down the lower bound and rounds up the upper bound to a
  multiple of set_alignment.   Chill starts from the lower bound and
  rounds up the size of a multiple of set_alignment.


jeff