Re: Seeking advice on casing greek final sigma

From: karl williamson (public@khwilliamson.com)
Date: Thu Nov 12 2009 - 12:04:45 CST

  • Next message: Asmus Freytag: "Re: Seeking advice on casing greek final sigma"

    I realized as I hit send on this that I meant to say lower casing, not
    upper casing. Here it is revised:

    > I'm thinking about upgrading the Perl 5 language to handle Unicode
    > context sensitive language-independent lower casing. The only character
    > in this class that is not also language dependent currently is the Greek
    > capital sigma, which may be lower cased to a small sigma or a final sigma, depending on context
    >
    > I'm writing to this list because it has people on it who hopefully have
    > some insights that I haven't considered.
    >
    > What I'm concerned about is what to do when the last non-caseignorable
    > character of the string is a Greek capital sigma. The string may actually
    > be part of a larger context unavailable to the code at this level.
    >
    > This can happen when the program that does this is assembling a larger
    > string, of which this is a non-terminal component, or more directly,
    > with the Perl regular expression syntax 's/(SOMETHING)/ABC\L$1\Edef/'.
    > This means that 'something' is to be replaced by 'ABCsomethingdef'. The
    > text between the \L and \E is to be lower cased. The casing function
    > only sees 'SOMETHING', and currently has no knowledge that there is a
    > larger context; it may be hard to change that.
    >
    > I can see several possibilities:
    > 1) Just don't do context sensitive casing, meaning no change from
    > current behavior.
    >
    > 2) Do the context sensitive casing only when there is no ambiguity;
    > meaning do nothing at the end of a string.
    >
    > 3) Assume that the end of a string means that there is no context to
    > follow, and go ahead and use the final sigma.
    >
    > So, I'm wondering if anyone here has insights, or knows what other
    > languages have or haven't done with this.
    >
    > Thank you
    >



    This archive was generated by hypermail 2.1.5 : Thu Nov 12 2009 - 12:06:40 CST