Re: I18N spam part II -- "reviewed" po files

[Christian Rose <menthos menthos com>]

On 21 Jul 2001, Fatih Demir wrote:
> What do you all say? I'd like to have it this way, the header comment
> case could cause some weird problems under some circumstances..
>
> Which language teams are already working with QA issues like review?

Hmm, in the Swedish team, we use a very informal po review process.
Albeit it being very informal, I think it is very successful.

Whenever there is a big update to a translation, a new release coming,
or a long time since last time it was reviewed, we send the translation to
the sv@li.org mailing list where all other translators and interested
people can have a look at it.
People then comment on typos, messages they think are wrong or can be
improved, submit their comments to the list, the translator then
responds and makes the changes he likes and comments on the list on every
previous comment. And so on, until there is nothing more to discuss.

It is very simple, but it works. What I don't like about the
"explicitly-mark-messages-as-reviewed" idea is:

* It's bureaucratic and doesn't fit well with the volunteer style of
projects that most free software projects are, and I also think it puts
pressure on the reviewer when in fact the translator should be
responsible for his translation. What I mean is that there is not many
people that would want to be held responsible for other people's
translations, I think.
I think we have an advantage in our informal system that way - anybody can
comment, and anybody does. In many cases there is two or three people
looking through the translation. I doubt that we would have that if people
were required to do an offical, formal review.

* The "mark-every-message-as-reviewed" system adds too much overhead to
files. Po files can be insanely big as they are, we don't have add
more clutter to every message. Also, the tools don't support it - I don't
think the reviewd-by comments will survive many translation memory
updates, merges, etc. Or am I wrong?

* The "mark-whole-file-as-reviewed" system doen't work in a rapidly moving
project like GNOME, I think. Some translations I update daily, and
messages are thus added and changed daily, making the effective lifetime
for a review to one day or so...


In short, I think this has to be solved by more informal reviews and
getting the teams to really use their mailing lists, rather than inventing
some formal review syntax, that is rather unlikely to be supported by
gettext & co anytime soon.


Just my opinion :)

Christian