[bind10-dev] missing C API for PostgreSQL: row-by-row query result retrieval

[João Damas]

what would be the cost of supplementing the pgsql code with an in-memory table (like the rbi we already have) to allow for simple name-at-a-time selection for things like axfr. This table would only contain the names, not al the DNS info at each node. It is a crutch but perhaps one that can be put together with mostly existing code

Joao

On 2 Oct 2012, at 05:47, JINMEI Tatuya / 神明達哉 wrote:

> While doing experiments I mentioned in the other thread(*), I noticed
> that the PostgreSQL's C API misses one important feature except in the
> very latest version, 9.2: the ability of incremental retrieving a
> query result.
> * https://lists.isc.org/pipermail/bind10-dev/2012-October/003866.html
> 
> It seems to be supported in 9.2
> http://www.postgresql.org/docs/9.2/static/libpq-single-row-mode.html,
> but before that it seems we have to store all results of a query in
> memory at once and then process it.  It makes some usage effectively
> unusable (or at least very expensive) such as axfr-out or loading from
> DB in-memory for a very large zone.
> 
> If our next target for the DB backend is PostgreSQL, we need to think
> about how to deal with that, including:
> 
> - Implement everything using pre-9.2 APIs, and say that it won't work
>  well for large scale environments
> - Support both the new and old APIs based on availability, and say
>  that for large scale environments 9.2 or higher will have to be
>  used.
> 
> By the way, MySQL's C API has been supporting incremental retrieval
> for quite some time, so we don't have this particular issue with
> MySQL.
> 
> ---
> JINMEI, Tatuya
> _______________________________________________
> bind10-dev mailing list
> bind10-dev at lists.isc.org
> https://lists.isc.org/mailman/listinfo/bind10-dev