Re: IRC Meeting 2012-10-06 21:00h UTC+2

[Luc Pionchon <pionchon luc gmail com>]

On 9 October 2012 12:29, Gil Forcada <gforcada gnome org> wrote:

El dl 08 de 10 de 2012 a les 03:05 +0300, en/na Luc Pionchon va
escriure:

> On 7 October 2012 01:51, Gil Forcada <gforcada gnome org> wrote:
>
> > - switching D-L to only show words
> > -- strong arguments on both sides, more discussion, mockups or subteams
> > working on proposals would help changing things (if there's a need for
> > it at all)
>
> My nightly 2cts,
>
> For the untranslated work, I would use both word and string counts.
> The combination of these two values gives more info than either one
> alone. Many strings and a small amount of words is different from a
> few strings with many words.

The problem on showing both, is obviously that you need twice the space
to show them

I don't see the problem. It's a web page, expandable by nature.

 

and if you do not have labels to put them along with both
words and strings counters then users will get more confused,

I don't think so. It all depends how it is done.

so to boil
it down:
- either one or the other: is simple, does not need much space
- both at the same time: you need labels or visual clues and double the
space to show up to 6 numbers (plus the string/words) instead of only 3
numbers.

If both numbers (strings and words) bring value, we must find a convenient way to present this information, rather than seeing it as a problem. It would be very surprising if one extra column becomes a space problem.

 

> For fuzzy strings, I would use strings, obviously. The amount of words
> to be worked in a fuzzy string is by definition unknown.

Well, you could say that as you have to review the original string, the
words are all words from the original, the fuzziness only tells you that
some not-so-random string will appear as a pre-translation.

I am not sure to understand what you mean (???)

> For the accomplished work, I would use either words or strings or
> both. It's not very important. If it is used for coverage evaluation,
> maybe a formula including both parameters might make sense. For
> example to give a stronger importance to short strings (labels, menus
> entries) - just as an example.

Coming up with good formula here can be quite tricky (note that I'm not
against it) but as a first approach maybe we can use strings on [1] and
words on [2].

As we will not please anyone if we (not) change the stats showing, a
compromise would be to have a switch (just like we have a "hide
completed modules") to show either strings or words and defaulting to
what I said above (so strings in [1] and words on [2] for example).

When it is about pleasing people, a compromise has the possibility of pleasing nobody :)


To sum up what I mean, and in a more visual way, I would preserve the string counts the way people are used to, and add the word count for untranslated strings (in red) :

         15%  3545  1884  16945  (22653)

The space question is not a problem, there are several possibilities to present this information in a convenient way.