gnc-glossary.txt proposal

[Richard Braakman]

Hi, I've been working on a translation for nl.  (It's going slowly,
so I'm targeting it for the 1.8 release.)  I noticed that I'm not
using the gnc-glossary.txt file because its format is awkward.

You see, its fields are tab-separated, but the entries vary greatly
in width.  This is not so bad yet for the de_DE column, but later
columns are not really columns at all anymore.  And the lines become
much longer than my window is wide, adding to the confusion :)

I propose using a multiline format for the glossary, for example:

Term: account
Explanation: A detailed record of money spent and received
de: Konto
nl: rekening

Term: account name
de: Kontobezeichnung
nl: rekeningnaam

This is the rfc822 style used in many places by Debian.  Each paragraph
contains information for one term, and paragraphs are separated by blank
lines.  Long fields can be split over multiple lines by using continuation
lines that start with whitespace.  Fields with no value can be omitted.

An XML format might be better suited for this project, especially if
some of the translations will use funky character sets.  But I don't
know enough about XML to propose one :-)  The rfc822 style could still
work if the charset for each language is defined to be the same as
is used in its po file.

As a side effect, using a multiline format will make it easier to
merge conflicts in CVS, because it doesn't involve peering at the
diff lines to figure out which columns have changed.

What do you think?

Richard Braakman