[climacs-devel] proposal for better error handling

[Robert Strandh]

Hello, 

I would like your reaction to the following proposal, because it
alters the buffer protocol slightly:

  * keep the no-such-offset condition, but make it be signaled only
    when an attempt is made to get to an object before the beginning
    or after the end of the buffer.

  * introduce two new conditions that inherit from no-such-offset:
    offset-before-beginning and offset-after-end with the obvious
    interpretation.

  * introduce a new condition invalid-motion used when an attempt is
    made to move a mark to an offset before the beginning or after the
    end of the buffer.

  * introduce two new conditions that inherit from invalid-motion:
    motion-before-beginning and motion-after-end with the obvious
    interpretation. 

  * leave unspecified what happens if an attempt is made to use a mark
    with an invalid offset (< 0 or > size), other than neither of the
    conditions above will be signaled.  This is a fatal error that
    should never occur, but we can leave it to the underlying buffer
    implementation to detect and signal it, for instance in the form
    of an illegal array reference.  Leaving this unspecified avoids
    having to test for it or establish a handler for some condition in
    the underlying implementation, which will improve performance. 

With this proposal, the top-level loop can handle these conditions and
give some reasonable message.  Climacs commands can choose either not
to bother and let the top-level loop handle them or else to handle
them and do something reasonable in case that turns out to be a better
strategy. 

Let me know what you think. 
-- 
Robert Strandh

---------------------------------------------------------------------
Greenspun's Tenth Rule of Programming: any sufficiently complicated C
or Fortran program contains an ad hoc informally-specified bug-ridden
slow implementation of half of Common Lisp.
---------------------------------------------------------------------