On Thu, Sep 23, 2010 at 11:26 PM, Itagaki Takahiro
<itagaki.takahiro@gmail.com> wrote:
> There are no active discussions :-( I think the author tried his best, so if
> other developers think it's a bad design, alternate plan must be proposed.
>
> Also, if the syntax change is trivial, that's why we merge it at
> earlier commitfests. I saw many patch went into "No patch reviewers
> because of too large size" syndrome before. DDL changes are 5K
> lines of diff -c patch, and "select" part is additional 6K lines.
IMHO, merging a syntax that doesn't actually work leaves the tree in a
broken state. It optimistically presumes that we will later commit a
patch to make the syntax in question do something useful. I don't
wish to presume that, even if we were well along on the design of the
core functionality and especially because we are not. It's fine to
break the patches into separate chunks for review, but our main tree
is not a dumping ground for half-finished features.
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise Postgres Company