Dimitri Fontaine wrote:
> Le mercredi 20 février 2008, Magnus Hagander a écrit :
>
>>> What about having a postgresql.conf.d directory containing a file per
>>> setting, maybe with a subdir per section. If I take Josh Berkus example,
>>> we'd have
>>>
>> <snip>
>> IMHO, if we do that it really sucks for those who use manual configuration
>> files, to the point of being completely unusable. It could be valid if we
>> want to support config only through the API, but that's not what people are
>> asking for.
>>
>> We need something that's low-impact for existing users, and this certainly
>> isn't.
>>
>
> What about having PG still able to load postgresql.conf or the tree of config
> files, automatically, erroring when both mechanisms are in use at the same
> time. This would allow for manual config editing installations and SQL
> embedded configuration setting, just not in the same cluster at the same
> time.
>
> I see how the proposal fails to answer to people wanting to edit the same
> configuration both with a file editor and SQL commands, but maybe having
> either postgresql.conf or SQL interface for configuration could be a first
> step?
>
>
No. Seriously. We need to have reasonable manual editability preserved
for all cases. The tree of files proposal just strikes me as a basic
non-starter, and, frankly, a piece of bad design. If you need structure,
then using the file system to provider it is just a bad move.
All this discussion seems to me to be going off into the clouds, where
every objection is met with some still more elaborate scheme. I think we
need to look at simple, incremental, and if possible backwards
compatible changes.
cheers
andrew