Awhile back I was objecting to Pavel's table-function patch on the
grounds that it introduced a new proargmodes value (thereby breaking
every client that looks at proargmodes) for what seemed to me to be
entirely insignificant benefits. I still consider that patch mostly
useless syntactic sugar, but that particular argument against it is
now a lot weaker in view of his variadic-functions patch, which also
introduces a new proargmodes value ... this time in support of actual
new functionality. If we're going to make clients update their
proargmodes handling, adding knowledge of two values rather than one
isn't a huge additional burden. So I withdraw my opposition to
table-functions on the condition that variadic-functions gets in too.
The question for the list is: does anyone have any serious objections
to either feature? If not, I'll get started on trying to get those
patches integrated.
http://archives.postgresql.org/message-id/162867790807100638n47c00eb4w304473a247a20aa7@mail.gmail.comhttp://archives.postgresql.org/message-id/162867790807140617s105a0a5ew61100897db52d23b@mail.gmail.com
regards, tom lane
