Re: Markup in messages

["Christian Rose" <menthos gnome org>]

On 7/25/06, Abel Cheung <abelcheung gmail com> wrote:
> On 7/18/06, Christian Rose <menthos gnome org> wrote:
> > Behdad, this is about http://bugzilla.gnome.org/show_bug.cgi?id=347110
> > -- it was only recently that I discovered that you have included a
> > section about markup to http://live.gnome.org/GnomeI18nDeveloperTips :
> >
> > "Following is a list of examples that need to be marked for
> > translation, but were not in some cases:
> > [...]
> > '<b>%s</b>': That is an innocent way to mark something to make it
> > boldface in the interface, to emphasize importance or make it a
> > header. But not every language has a concept of modern boldface
> > typefaces, or even if it has such fonts, they may not be the preferred
> > font for such kind of emphasis."
> >
> > So are you suggesting that developers include the surrounding markup
> > in the translateable message, in case you may need to change it for
> > Persian for rendering purposes?
>
> My guess is that they are just reluctant to changing anything
> not visually beneficial, not that they have already foreseen
> the problem. But to make an excuse, this is a good
> one, especially when what Behdad said is valid. I have been
> bitten by this kind of markup/font issues myself.
>
> Due to absense of freely available boldface Chinese font, I
> have seen somebody translate things like:
>
> msgid "<b>%s</b>"
> msgstr "<i>%s</i>"
>
> in order to distinguish it from normal text.
>
> Of course in most cases the proper 'fix' is to have a boldtype
> font. But creating a boldface font is not always that easy,
> when one is talking about complex scripts that consists of
> at least hundreds or thousands of glyphs; and it needs quite
> some human horsepower as well, which may not be available for
> newly available languages in F/OSS world.

I understand that problem, but I for sure don't agree that this is the
proper way to fix it, as this hack, that is, gratiously including
surrounding markup inside the translateable message content, causes a
lot of problems for all translators. As a translator you want to
translate the *content*, not the surrounding presentation. I don't
care whether the words will be written on a billboard, a folder, or
on-screen, the words will be the same. As Clytie explained:

> I support you. Not only is it a problem, one more thing that can
> break, but it's one more difference preventing exact matches using
> msgmerge.
>
> I spend way too much time changing fuzzy strings like this:
>
> ,fuzzy
> msgid "<b>Preferences</b>"
> msgstr "Tùy thích"
>
> to
>
> msgid "<b>Preferences</b>"
> msgstr "<b>Tùy thích</b>"
>
> simply adding, modifying or removing formatting to match the original
> string. The actual text is fine.
>
> I really don't see it as our job to adjust formatting. I would indeed
> prefer to concentrate on the actual translation.


Here are some ways to solve the problem of bad rendering:

1) Fix the fonts. Agreed, not an easy task.

2) Solve the problem at the Pango level. Instead of encouraging the
use of <b> and <i> tags, add abstract <strong>, <em> tags/attributes
or some such, allowing for different interpretations for different
scripts. If we know that boldface fonts for Persian always suck,
<strong> can have a different representation for Persian script.
Then the HIG and other such specifications can specify these
attributes instead of a particular font style.


And here's how to solve the (ab)use of using surrounding PangoMarkup
inside translateable messages:

3) Make gtk+/Pango have support for *attributes* instead of forcing
everyone to use PangoMarkup. If I want a label to be bold, italic,
smaller, or larger, I as a developer should be able to simply set
attributes to that effect. Right now, it's too common to see something
like this:

msgid "<span size=\"xx-large\">Bug Buddy</span>"
msgstr "<span size=\"xx-large\">Bug Buddy</span>"

msgid "<span weight=\"bold\">Date & Time</span>"
msgstr "<span weight=\"bold\">Datum och tid</span>"

msgid "<b>Date & Time</b>"
msgstr "<b>Datum och tid</b>"

msgid "<span size=\"medium\"><b>No file</b></span>"
msgstr "<span size=\"medium\"><b>Ingen fil</b></span>"

msgid "No file"
msgstr "Ingen fil"

Often, you will have duplicated messages in the same file, just with
different markup!

If gtk+/Pango and libglade would make it possible for application
writers to set *attributes* for these things instead of having to
resort to PangoMarkup, all problems would be solved.


Christian