Re: and it's not a bunny rabbit, either

On Sat, Jan 1, 2011 at 9:53 AM, Guillaume Lelarge
<guillaume(at)lelarge(dot)info> wrote:
> Le 01/01/2011 06:05, Robert Haas a écrit :
>> On Fri, Dec 31, 2010 at 8:48 AM, Peter Eisentraut <peter_e(at)gmx(dot)net> wrote:
>>> On tor, 2010-12-30 at 11:03 -0500, Robert Haas wrote:
>>>> No, quite the opposite.  With the other approach, you needed:
>>>>
>>>> constraints cannot be used on views
>>>> constraints cannot be used on composite types
>>>> constraints cannot be used on TOAST tables
>>>> constraints cannot be used on indexes
>>>> constraints cannot be used on foreign tables
>>>>
>>>> With this, you just need:
>>>>
>>>> constraints can only be used on tables
>>>
>>> At the beginning of this thread you said that the error messages should
>>> focus on what you tried to do, not what you could do instead.
>>
>> Yeah, and I still believe that.  I'm having difficulty coming up with
>> a workable approach, though.  It would be simple enough if we could
>> write:
>>
>> /* translator: first %s is a feature, second %s is a relation type */
>> %s cannot be used on %s
>>
>> ...but I think this is likely to cause some translation headaches.
>
> Actually, this is simply not translatable in some languages. We had the
> same issue on pgAdmin, and we resolved this by having quite a big number
> of new strings to translate. Harder one time for the translator, but
> results in a much better experience for the user.

Is it in any better if we write one string per feature, like this:

constraints cannot be used on %s
triggers cannot be used on %s

...where %s is a plural object type (views, foreign tables, etc.).

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company