2013-04-01



Today, during a hot bath, I read in Software Tools  by  Kernighan
and  Plauger.  I like the book a lot, although it's hard to read.
(In large parts  it's  only  a  discussion  of  (Ratfor!)  code.)
Currently,  I'm in the editor chapter, which describes the imple-
mentation of ed(1).  While reading that, an idea  occured  to  me
...

In mmh, messages can be specified on the command line using  sym-
bolic sequences. For instance:

     show f          # show the first message
     show n:3        # show the next three messages
     show c-l        # show the messages from current to last

What I've been missing is a way to specify the second  last  mes-
sage  or  the  fifth-next message. `c+5' could be used, but `l-2'
would be treated as a range instead of a  substraction.  But  why
not converting to the notation ed(1) has? With ed, you can speci-
fy these two things.  The difference is, that ed uses  the  comma
to specifiy a range.  Changing this would break compatibility but
would allow to make the notation more powerful. The  minus  would
then be reused for substractions and the plus would be introduced
for additions. An old-style range (`f-l') put into the  new  con-
cept would be detected as ranges needed to be ascending, thus the
substraction result would become negative, which is not  allowed.
In  the new system the substrahend is required to be smaller than
the minuend. Hence, though compatibility will be lost, any compa-
tiblity problems will be detected.

The changes needed are small, I suppose. But then I'll be able to
scan the surrounding messages, finally:

     scan c-5,c+10


btw: It was necessary to use `c' for the current message and  re-
move  the  `.'  alias  as  the  dot is the identifier for relativ
directory paths. This does divert from ed's  notation,  but  only
syntactically.


I'm looking forward to implement this change ... and  restart  to
work on mmh, hopefully.



http://marmaro.de/lue/        markus schnalke <meillo@marmaro.de>