## source code mgmt (was: RE: [ENH] String>>browseUrl)

Richard A. O'Keefe ok at atlas.otago.ac.nz
Sun Dec 10 22:49:50 UTC 2000

I wrote:
> I actually demonstrated to about a hundred school pupils last January
> that they couldn't type a simple list of names into Word.  Yet the
> neatly printed list had been entered using a plain ASCII terminal...

Jon Hylands replied:
Okay, I'm curious... How exactly did you do that.

Here are the instructions I gave them:
I have a list of 50 names and addresses.  The given names are all
real given names; the family names are not all real {\em names},
but they are all real {\em words}.  The addresses are all either
real streets in Dunedin, real M\=aori place names, or real words

There were
10 English names
10 other British Isles names
10 Maori names
10 German names
10 Turkish names

The "not real names" part applied to Turkish; I had a book that gave me
enough given names, but was rather short on family names.  I certainly
didn't have a Turkish gazetteer.  But I *have* had Turkish students.

The first 20 should have given them no trouble, and if they figured
out that it was a good idea to type the equivalent of
M<small><sup>c</sup></small> *once* and then copy/paste it whenever
needed (this is a Scottish settlement, after all...) it shouldn't even
have been all that laborious.  (I had to tell them how to find &AElig
which *is* used in some English names.)

The Maori names and addresses should have been easy.  Maori is the
language of the original inhabitants of these islands, all University
buildings are labelled in Maori and English, Government departments have
Maori names as well as English ones, children are taught the Maori words
for numbers and colours at kindergarten.  It's an *important* language
in this country.  Maori only has the consonants H, K, M, N, NG, P, R, T,
W, WH and the vowels A, AA, E, EE, I, II, O, OO, U, and UU.  Except that
for some reason the United Tribes decided to make the "macron" orthography
the official one, so the long vowels are &amacr; &emacr; &imacr; &omacr;
and &umacr.  Which none of the installed fonts Word could use on their

Do I really mean to say that Apple and Microsoft sell systems and software
in this country that cannot support one of the two official languages "out
of the box" AT ALL?  Yes, I do.

Well, there's a hack.  The University makes a fewer (just a few!) fonts
available that have been butchered to have <vowel> WITH MACRON in the
place where to be standards conformant they should have <vowel> WITH
DIARESIS.  There _is_ an ISO 8859 standard that would work beautifully.
Does anyone in this country use it?  Nope!  Great things, standards.
Any E-mail you receive from New Zealand that claims to be in ISO 8859-1
or Windows 1252 is most likely lying.  Great things, standards.

Oh yeah.  I forgot to mention that these butchered fonts are *not*
provided ready for use in the X Window System.

OK.  So if they download, install, and use the butchered fonts, they can
type Maori (really of course M&amacr;ori).  So I had some Germand names
in there as well, didn't I?  Naughty old me.  Because those names require
vowels with umlauts.  And the usable-for-Maori fonts just clobbered them.
Oh dear. None of them actually got that far.  Had they done so, they would
have found themselves having to switch fonts every couple of lines.
Indeed, they'd have to switch fonts to alternate between carefully written
Maori and carefully written English, where the diaresis is still a useful,
even important, sign.

And then there were the Turkish names, which I intended to be the real
killer.  *None* of the fonts available to these students included
LATIN CAPITAL LETTER I WITH DOT ABOVE  or
LATIN SMALL LETTER DOTLESS I....

I was able to write this list easily using either (La)TeX or SGML.
But Word's approach (you may *type* only what you have screen fonts for)
literally made it impossible for them to *type* all the letters.

In fact this wasn't the main point of the lesson.  The point was to tell
them about errors in computer data, and to show them that a different
approach (SGML) could be used to catch many typing errors at the point of
origin.  They had a list of 19 probable errors to check for, nearly half
of which an SGML-based approach rather than a Word-based approach could
catch.  Half of the rest, the SGML approach could prevent.

This has taken long enough, so I don't propose to go into details.