[PATCH, c++ diagnostics] Fix c++ front end i18n problems
Shujing Zhao
pearly.zhao@oracle.com
Tue Dec 1 09:40:00 GMT 2009
On 12/01/2009 04:39 PM, Paolo Bonzini wrote:
> On 12/01/2009 04:51 AM, Shujing Zhao wrote:
>> + if (candidates->next)
>> + str = "candidates are:";
>> + else
>> + str = "candidate is:";
>> +
>> /* Remove deleted candidates. */
>
> Shouldn't the decision be taken _after_ removing deleted candidates?
>
>> +/* The various readonly error string used by readonly_error. */
>> +typedef enum readonly_error_str
>> +{
>> + /* assignment */
>> + RES_ASSIGNMENT,
>> + /* assignment (via 'asm' output) */
>> + RES_ASSIGNMENT_ASM,
>> + /* increment */
>> + RES_INCREMENT,
>> + /* decrement */
>> + RES_DECREMENT
>> +} readonly_error_str;
>
> Leave this as an enum (no typedef).
>
>
>> + issure an error-message and return. */
>
> "issue"
>
>> + if OVL_NEXT (fns)
>
> Leave parentheses even though they are technically unnecessary.
Ok, thanks.
I
> disagree with extracting the choice between candidates are/candidate is
> to a separate function; however, I'd just compute a bool here and pass
> that to print_overloaded_functions.
>
>> @@ -15747,12 +15756,17 @@ most_specialized_class (tree type, tree
>
> You need to call gettext on str and wrap the strings (candidates
> are/candidate is) with G_("..."). Also, there is a "candidates is:"
> typo here.
>
>> - str = " ";
>> + str = " ";
>
> I'd just make str = "" here, since "candidate is" and "candidates are"
> have different sizes.
When str is "candidate is", the loop will be only run one time. Though str is
assigned again, it will never be used. Keep the " " will make the
output looking better.
>
> I suggest you leave "candidate is" aside in the beginning and submit a
> patch for the rest. Then you can do a separate patch to fix the other
> issue -- otherwise you'll make the same mistake as me and the patches
> will lie uncommitted for seven more years. ;-)
Ok, thanks. I will sumbit a patch without "candidate is" first. And then handle
the "candidates".
Pearly
More information about the Gcc-patches
mailing list