Re: comments/string changes and issues with dpkg's messages
On 7/27/05, Christian Perrier <email@example.com> wrote:
> Quoting Eddy Petrisor (firstname.lastname@example.org):
> > Hello,
> > As I have said in at a previous time, I have started working on adding
> > comments in the dpkg code, in order to add automatic comments to the
> > translatable strings.
> If you can do it on your own TLA branch, that'd be great. See Scott's
> wonderful explanations about arch/tla/baz at http://www.dpkg.org
Downloaded the info, as I am offline @home; if there is a way to make
my repo (after I set it up) available online, I will; If not I will
work with patches.
I didn't realized that tla takes inodes into consideration when
diff-ing and during other operations that should be local (I made the
checkout on a connected machine then copied all the data on my laptop;
when I tried a diff, I saw this behaviour) - apparently tla does not
have this feature, like, iirc, monotone does.
> > 2) in some languages forms like "3 installed packages", "1 installed package"
> > and "6 installed packages" need three different forms for the word
> ngettext is what you need as others explained.
So I will need to change the dpkg code; that answers my question.
> > 3) in some cases the messages are so cryptical that rephrasing them would be a
> > much better solution than adding automatic comments:
I guess that for the ones that are not cryptical commenting is a
better option, isn't it?
> > Solutions:
> > a) change the messages
> > consequnce:
> > - translations will be fuzzied; many translators will cuss at me
> > + messages will be easier to translate
> > + no more changes will be necessary later
> > + messages will be easy to understand for regular Joes (although they will
> > probably won't be able to do anything with the resulted info
> Do this. We are at a moiment in the devel cycle where breaking
> translaiton is not *that* a problem. After all, working with broken
> strings is our daily life..:-)
Ok, that's what I was inclined to do, too.
> More seriously, even if I sometimes have to slow down upstream by
> requesting not changing strings too often, when a string is broken, it
> has to be fixed, whether or not it breaks the translations.
> > 4) the best packaging practices states that the program should not "speak
> > as a person"; impersonal messages are preffered.
> > this is what I have found, at a glance:
> > "...like what you want, and you can `reject' my suggestions by using the
> > capital..." - there shouldn't be "my sugestions"
> > =>
> > "and you can `reject' the suggestions by using the capital"
> Yes, also. Throw out these "amateur-style" messages. There is NO
> REASON to use first person in a computer program.
Thanks for the "strong" argument :) . Seriousely, there are these two
oppinions between translators (not only Romanians) and I favor the
impersonal speaking, although it leads, usually, to longer phrases.
"Imagination is more important than knowledge" A.Einstein