SORT BY:

LIST ORDER
THREAD
AUTHOR
SUBJECT


SEARCH

IPS HOME


    [Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

    RE:



    On Thu, 21 Mar 2002, Eddy Quicksall wrote:
    
    > That would be ok for something like an expression parser but our parser
    > should be very simple minded and it is better if it is not context
    > sensitive. It is the same idea as "don't use two characters".
    
    While I agree simple is good, there has to be some sort of context
    senstivity. I mean you have to figure out if you have one value or a
    range, don't you? How do you do that? Don't you parse one number and then
    look at the next character to decide if you have a range or not? If the
    next character is the range-indicating character, you remove it from the
    stream and then call the number parser again, don't you? Given that, how
    is using a '-' a problem? It's the exact same code (hardware or software)
    as using ';', just a different constant.
    
    Since I really doubt algebraic expressions will ever be valid, the
    sequence " ... number '-' number ... " can never mean anything other than
    a range. So how is that hard for even a simple parser to deal with?
    
    Also, I think you've assumed that: 1) we should use '-' either for
    negative numbers or for ranges not both, and 2) that we would rather use
    '-' for negative numbers than for ranges. While I don't care much about
    the first assumption, if we have to choose between ranges using '-' and
    negative numbers, ranges seem to make more sense to me here. Most programs
    accept them, and it's a rather intuitive way to express them.
    
    Also, are we really ever going to need negative numbers? SCSI doesn't have
    them. IP doesn't. TCP does not.  Since those protocols use unsigned
    numbers, why would we need signed?
    
    > Also, the "looks" of it is not very important.
    
    I think you took my use of "look" at too simplistic a level. I did not say
    aesthetics. :-) My use of "look" reflects parsability of text. If
    something "looks" like something else, I'm saying it's more difficult to
    distinguish the two interpretations in a parser (you need extra context
    which we both agree is usually bad). If things "look" like each other,
    it's harder to tell them apart. We don't want that, in order to keep the
    parser simple as you mentioned above. Thus "looks" are important.
    
    Take care,
    
    Bill
    
    

    • References:
      • RE:
        • From: Eddy Quicksall <Eddy_Quicksall@ivivity.com>


Home

Last updated: Thu Mar 21 14:18:12 2002
9251 messages in chronological order