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

Re: [StrongED] searching for last word in a field



In message <d7e1386257.jim@xxxxxxxxxxxxxxxx>
          Jim Nagel <stred@xxxxxxxxxxxxxxxx> wrote:

>The premise "preceded by a blank line" doesn't hold, though.  In the 
>example in my original post I inserted the blank lines just to make the 
>example easier to read.  But that introduced a green herring.
>The actual file has no blank lines.  I could easily add them, of course.

There has to be an algorithmic way of determining which substring of
a line is a "surname" to be singled out for sorting. As I have
pointed out previously, this can be quite an involved business,
and should preferably have been thought through when the TSV files
are created. No doubt businesses everywhere have to keep reinventing
this particular wheel. I can remember the problems our university
exams database had with foreign names that do not conform to local
prejudices. The same person might be entered with different names
because there is no standard transliteration protocol from, say
Arabic or Russian names, to ASCII. So you have to check that different
records do not refer to the same person.

Personally I think labelled records make life a lot easier. CSV files
I have often inveighed against. TSV files can be unhelpful because the
omission or accidental duplication of a tab will alter the significance
of the rest of the line, and they are hard to read. Labelled records
are more forgiving, easier to read, and easier to manipulate.
--
Gavin Wraith (gavin@xxxxxxxxxxxxxxx)
Home page: http://www.wra1th.plus.com/

-- 
To unsubscribe send a mail to StrongED+unsubscribe@xxxxxxxxxxx
List archives and instructions at
http://stronged.torrens.org/index.html