[xliff-tools] Re: Format-tags placeholders

[David Fraser]

Asgeir Frimannsson wrote:

>Hi,
>
>On Tue, 3 May 2005 01:26, Fredrik Corneliusson wrote:
>  
>
>>This brings up a question about how much help filters can give the
>>translator. For example if a message uses c-formatting and the placeholders
>>changes sequence in the translation the XLIFF to PO filter could
>>automatically add or check the sequence representation (eg. %2$s %1$s)
>>based on the ph tag id. I think could lower the entry bar for
>>translators, make translation more efficient and at the provide
>>functionality not provided by the current PO editors.
>>
>>What's your take on this?
>>    
>>
>
>I'm all for this approach (extracting parameters to <ph> elements), if it's 
>correctly implemented. Gettext supports a wide variety of source formats, and 
>extraction should occur when one of the following flags are present:
>awk-format
>c-format
>csharp-format
>elips-format
>gcc-internal-format
>java-format
>librep-format
>lisp-format
>objc-format
>perl-format
>perl-brace-format
>php-format
>python-format
>sh-format
>tcl-format
>ycp-format
>
>(might have missed some)
>  
>
Is there a reference somewhere for these?

>And all these languages have different rules for how a parameter is 
>represented (E.g. {1,number} for Java, %d for C), and validating/parsing all 
>these can get complex.
>
>..Then you have the side-cases, like when handling parameters in Farsi, you 
>have to change %d to %Id [1].
>
>But yeah, it's fully possible, and I hope to implement this (including 
>re-ordering) for at least c-format in my filters. But I don't think we should 
>have it as a requirement in the guide, -but we should further discuss the 
>options.
>
>Another issue: Are we are using a too low level of abstraction in c-based opne 
>source applications? In Qt/KDE we're using %1, %2 etc, and do further 
>formatting of the parameters in the source code, and I see this as a better 
>approach than having complex c-parameters.
>
>Maybe we should start using something like:
>printf( _("My name is %1s and I have %2s in my bank account"), name 
>format_currency(value, current_locale) )
>
>in favour of this approach:
>printf( _("My name is %1s and I have USD %2d.02 in my bank account",name, 
>value)
>  
>
Yes, that looks Good.
In OpenOffice.org we have to deal with the menace of having at least 9 
different styles of declaring variables:
&xmlstyle;
%percentbeforeandafter%
%percentbefore
$(dollarparentheses)
$dollarbeforeandafter$
$[dollarsquarebrackets]
${dollarcurlybrackets}
#hashbeforeandafter#
($justtoconfuseeveryone)

David