Andres Freund wrote:

> Now a) and b) are recent oversights of mine. I'd apparently not realized
> that there's detailed docs on this in buffer/README. But c) is pretty
> old - essentially 5d50873 from 2005.
> I wonder if it's worthwhile to go into that level of detail - seems
> kinda likely to get out of date, as evidenced by it being outdated for
> ~10 years.

I think it makes sense to keep a high-level overview in the README; in
particular the description of how users of this API would use it should
be there.  But the implementation details should live in comments inside
the file.  I don't think the details of the buffer replacement algorithm
should be in the README.

Álvaro Herrera      
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

Sent via pgsql-hackers mailing list (
To make changes to your subscription:

Reply via email to