[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-i18n-doc
Subject:    Re: clarification of a string
From:       Albert Astals Cid <aacid () kde ! org>
Date:       2010-12-22 17:53:58
Message-ID: 201012221753.58694.aacid () kde ! org
[Download RAW message or body]

A Dimecres, 22 de desembre de 2010, Chusslove Illich va escriure:
> >> [: Chusslove Illich :]
> >> [...] it requires the translator to closely follow the *positional
> >> context*, which means always keeping in sight messages above and below
> >> the current one, and their source references in relation to one another.
> > 
> > [: Federico Zenith :]
> > I disagree. Translating in a straight line is not necessarily the best
> > way to do this [...]
> 
> Following the positional context does not require translating in a straight
> line. It requires only having messages around the message currently being
> translated in plain sight, and understaing the meaning of message parts
> (source references, extracted comments, etc.)
> 
> > Case 1: the translating group has decided that e.g. "override" will be
> > translated in a certain way. In Lokalize, I would filter for "override"
> > and translate all messages containing the word (and correct previous
> > translations). This filters the message list and will lump together
> > messages from different parts of the code. I know I do it often and it
> > makes my work easier, and I don't think I am the only one.
> 
> I know for a fact that you are not the only one :) However, I do consider
> it a slightly defective practice because it negates the positional
> context.
> 
> Instead, when I go around modifying and translating scattered messages with
> some common theme, I do it quite similarly but with a twist: while the
> filtering is such that I can jump from one to another message of interest,
> it does not make all other messages invisible. That is, I will go from one
> selected message to the next translating or modifying it, but always having
> the surrounding non-selected messages in direct sight.
> 
> > Case 2: Positional context is no guarantee. [...]
> 
> That is right, but nothing is guarantee, and everything servers to improve
> the odds.
> 
> Even a manual context can be inconclusive. Such a context:
> 
>   msgctxt "verb"
>   msgid "Open File"
>   msgstr ""
> 
> would be near useless to me, because if this were a button text I would use
> a command-like translation, and if it were a dialog title, I would use a
> noun-like formulation. Therefore, more useful to me would be seeing:
> 
>   #: foobar.cpp:12
>   msgid "Open File"
>   msgstr ""
> 
>   #: foobar.cpp:13
>   msgid "This file looks like a binary, are you sure you want to open it?"
>   msgstr ""
> 
> This would give away (also noting source references) that the first message
> is most likely a message box title (it is less likely the confirmation
> button, since normally that would be just "Open"; though I would check the
> source file if at hand).
> 
> > We cannot expect translators to dig in the source code.
> 
> That is fair enough from the point of view of being realistic. But, the
> less digging in the source code, the more the translation glitches (in
> short messages). So, I would slightly modify and lengthen this to: we
> cannot expect *all* translators to dig in the source code, but we should
> always point out that the possibility exists, and encourage translators to
> try it out.
> 
> (On a side note, while a translator may not look in the source code, a
> reviewer, someone with more technical insight, might at a later time. That
> is what I do.)
> 
> > If developers want good translations, they must invest some extra 10-20
> > keystrokes (5 seconds?) to add a msgctxt every time the message is prone
> > to misunderstanding. If in doubt, don't leave it out.
> 
> Unfortunatelly, programmers usually do not know when a message is prone to
> misunderstanding. Worse yet, what I need to know about a message in my
> language, may not be needed in another language (the "Open File" example).
> Even cases that are easily formulated (like "never leave a single adjective
> message without a context") are more often than not forgotten by
> programmers.
> 
> So, we should make use of (or: point out the possibility to make use of)
> whatever means available to improve the odds at understanding the context.

We should improve existing or unexisting contexts as we realize we need them. 
So who's adding that context to this particular message?

Albert
[prev in list] [next in list] [prev in thread] [next in thread]