On Jul 12, 2010, at 3:50 PM, mdipierro wrote:
> As a test, I implemented this in trunk
>
> T(' hello world ') -> ' hello world
> '
> T(' hello world # token') -> 'hello
> world'
> T('hello # world # token') -> 'hello # world'
>
> I.e. if no # nothing changes. If one '#', everything on its right is
> not translated, everything is stripped. More than one '#' is ignored
> and this would allow to have '#' in the translated string.
That seems about right. I'm a *little* wary of using #, since it's pretty
likely that there are existing T strings with a # in them, and they'll have to
be fixed, no?
>
> What do you think?
>
> On 12 Lug, 17:30, Jonathan Lundell <[email protected]> wrote:
>> On Jul 12, 2010, at 3:26 PM, mdipierro wrote:
>>
>>> I like ((...)). If I understand you suggest stripping the whitespaces
>>> ONLY inside ((..)). Did I understand?
>>
>> Stripping the main string, too--wasn't that the point of this exercise in
>> the first place? I don't have a strong opinion, though.
>>
>>
>>
>>> On 12 Lug, 17:22, Jonathan Lundell <[email protected]> wrote:
>>>> On Jul 12, 2010, at 3:15 PM, mdipierro wrote:
>>
>>>>> Ok but can I propose we use
>>
>>>>> T('canto /* my room */') instead of [[...]] to avoid confusion with
>>>>> markmin syntax? This would b easy to implement.
>>
>>>> Or ((my room)).
>>
>>>> Or <<my room>>.
>>
>>>> Regardless, you'd want to specify what happens to white space. Strip
>>>> everything, I think, so T(' canto /* my room */') is equivalent to
>>>> T('canto/*my room*/').
>>
>>>> (That's why I don't much like /*...*/, though; it's sort of ugly without
>>>> extra spaces.)
>>
>>>>> On 12 Lug, 15:39, Jonathan Lundell <[email protected]> wrote:
>>>>>> On Jul 12, 2010, at 1:32 PM, Álvaro Justen wrote:
>>
>>>>>>> On Mon, Jul 12, 2010 at 17:16, Jonathan Lundell <[email protected]>
>>>>>>> wrote:
>>>>>>>> On Jul 12, 2010, at 12:51 PM, mdipierro wrote:
>>
>>>>>>>>> suggestions? :-)
>>
>>>>>>>> Ideally (from a usability pov), a variant="something" argument to T(),
>>>>>>>> I suppose. But I can't think of a non-messy way to do it.
>>
>>>>>>>> How about something like this:
>>
>>>>>>>> T('string to translate [[VARIANT something]]')
>>>>>>>> T('string to translate [[VARIANT something else]]')
>>
>>>>>>>> If there's no translation in effect, then [[VARIANT .*?]] *$ gets
>>>>>>>> stripped from the string. Otherwise it's part of the lookup. It could
>>>>>>>> just be
>>
>>>>>>>> [[something]]
>>>>>>>> [[something else]]
>>
>>>>>>>> ...keeping in mind that if you really wanted that at the end of a T()
>>>>>>>> string, you could write:
>>
>>>>>>>> T('blah blah [[blah]][[]]')
>>
>>>>>>>> ...and only the trailing [[]] would be stripped.
>>
>>>>>>>> BTW, there's a typo in languages.py:
>>
>>>>>>>> # patter for a valid accept_language
>>
>>>>>>>> (and the pattern could use a comment or three)
>>
>>>>>>> I don't like the idea of changing the string to be translated. We can
>>>>>>> use a 'context' parameter as I said in other email some time ago,
>>>>>>> like:
>>>>>>> T('canto', context='my room') #translating from pt-BR to English
>>>>>>> should be 'corner'
>>>>>>> T('canto', context='music') #translating from pt-BR to English should
>>>>>>> be 'sing'
>>
>>>>>> That is, btw, my 'variant=' suggestion, above. If it's practical, I'd
>>>>>> prefer it. But embedding the variant/context into the string would be (I
>>>>>> think) less disruptive.
>>
>>>>>> T('canto [[my room]]')
>>>>>> T('canto [[music]]')
>>
>>>>>> Not as pretty, but almost identical in effect.
