Ah, thanks for the suggestions. I'll revise this patch soon :)
Fabien COELHO <coelho@cri.ensmp.fr> writes:
>> Done.
>
> Ok. The file should be named "v2".
>
>> Would you like to be the reviewer?
>
> Dunno. At least I wanted to have a look at it!
>
> My 0.02€:
>
> I think that the improvement provided is worthwhile.
>
> Two questions: Why no documentation update? Why no non-regressions
> tests?
>
> As far as the output is concerned, ISTM that "btree index", "hash index"
> and so would sound nicer than "index: sub-type".
>
> I'm wondering about the sub-query implementation: Maybe an outer join
> could bring the same result with a more relational & elegant query.
>
> The "gettext_noop" call does not seem to make sense: the point is to
> translate the string, and there is no way to translate "index: <some sub
> query>". The result of the query should be translated if this is what is
> wanted, and that can only be by hand:
>
> CASE amname || ' index'
> WHEN 'hash index' THEN '%s' ...
> WHEN ...
> ELSE amname || ' index' # untranslated...
> END
>
> gettext_noop('hash index') # get translation for 'hash index'
> ...