Re: [PATCH] q2c: Eliminate many translations for generated .c files.

[Ben Pfaff]

Ben Pfaff <address@hidden> writes:

> John Darrington <address@hidden> writes:
>
>> On Sat, Mar 19, 2011 at 05:45:50PM -0700, Ben Pfaff wrote:
>>      Most of the %s directives used in msg() format strings that q2c
>>      generates don't have to be substituted at q2c translation time.  By
>>      translating them later, we can eliminate a lot of the work that
>>      translators need to do for q2c-generated fiels.
>>      ---
>>      This seems correct to me but I'd like to have another set of eyes
>>      look it over before I push it.
>>
>> I don't think this is going to work. We would need to add all the generated 
>> .c
>> files to the potfile manifest.  That will cause problems because some of the
>> strings therein have already been translated.  Even if we can solve that 
>> issue,
>> it'll been that the potfile will depend on files generated at build time 
>> which
>> is not a good idea.  It'll also mean that the potfile comments will get 
>> bloated
>> with references to every generated .c
>>
>> Instead, I suggest we solve this problem as follows:
>>
>> 1. Create a new function within q2c.c similar to
>>
>>   static void dump_msg (int severity, const char *msg, ...); 
>>
>> 2. We can then call this function within q2c.c like this:
>>
>>      dump_msg (SE, N_("`(' expected after %s specifier of %s subcommand."), 
>> s->specname, sbc->name);
>>
>>    and it will generate the following code:
>>
>>         msg (SE, gettext ("`(' expected after %s specifier of %s 
>> subcommand."), "the-specname", "the-name");
>>
>> 3.  We will also have to add q2c.c to the list of translatable files.
>>
>> Does this make sense?
>
> It is definitely an improvement on the status quo.  I think it
> will work.  I'll give it a shot when I have time.

I came up with a strategy I like even better.  I'll post a series
to implement it soon.
-- 
Ben Pfaff 
http://benpfaff.org