Re: Would it save time for translators to fold similar lines?

["Christian Rose" <menthos gnome org>]

On 12/4/07, Leonardo Ferreira Fontenelle <leo fontenelle gmail com> wrote:
> Em Seg, 2007-12-03 às 15:07 -0500, Thomas Thurman escreveu:
> > I quite understand the morphology issue, but in this context it has to
> > do with keywords (which necessarily are immune from morphology). I was
> > not suggesting (and would not suggest) folding strings together which
> > varied only by arbitrary nouns, as in the two examples you give.
>
> But the morphology of the surrounding sentence can be changed if the
> keyword is male, female, singular, plural etc.

In the specific examples Thomas mentioned:

  #: ../src/theme-parser.c:1547
  #, c-format
  msgid "No \"bottom\" attribute on element <%s>"
  msgstr ""

the first noun is the word "attribute", so gender nor plural should
matter -- the word "attribute" will always have the same gender and
plural.

Granted, in some languages the morphology will still matter, because
of rules that make other words change depending on whether the %s
starts with a wovel or not, for example. But in such cases, the
problem can often be avoided by rewriting the sentence in the
translation, and changing the sentence order:

  msgid "No \"bottom\" attribute on element <%s>"
  msgstr "The element <%2$s> lacks the following attribute: \"%1$s\""

The example is fictious, but should demonstrate the general idea. If a
language has morphology issues like this, it is often solveable by
rearranging the translation.


In the case of explaining keywords, like in the fr and vi cases, my
take on this is that it depends on the message. In my opinion, plain
error messages do not need explanations, they're there to be short and
concise and explain the condition that triggered the error, not to
explain everything there is to know.
Also, in the particular examples above, the error message is there to
tell you the following pieces of the information:

1) An error occurred
2) The error was an missing attribute
3) What element lacked that attribute

No more than that. In particular, the answer to the question "what
does this attribute mean" is not part of the original message, and is
out of scope. It is simply irrelevant to the error condition. If you
need to know more about the particular attribute or element, look it
up in the docs.

On the other hand, schema descriptions, like the following fictious
example, I treat differently:

  #: foo.schemas.in:42
  msgid "Vertical alignment. Possible values are \"top\", \"middle\",
and \"bottom\"."
  msgstr "Vertikal justering. Möjliga värden är \"top\" (överkant),
\"middle\" (mitten) och \"bottom\" (nederkant)."

The idea behind this distinction is that long schema descriptions are
there to be descriptive and fully contained -- you should not need to
further look up what the values mean. Also, the length of the message
does not matter much. This is unlike most error messages and other
short messages where keywords may appear.

So, I still fully support and welcome the original suggestion to
change the messages into

  msgid "No \"%s\" attribute on element <%s>"



Christian