On Thu, Feb 13, 2020 at 9:26 AM Tom Lane <t...@sss.pgh.pa.us> wrote: > "David G. Johnston" <david.g.johns...@gmail.com> writes: > > On Thu, Feb 13, 2020 at 4:13 AM Alexey Bashtanov <bashta...@imap.cc> > wrote: > >> Please could someone have a look at the patch attached? > >> It's not just pedantry but rather based on a real-life example of > >> someone reading and being not sure > >> whether e.g. joins can be used in there. > > > Drive-by comment - I'm on board with the idea but I do not believe this > > patch accomplishes the goal. > > IMO there is too much indirection happening and trying to get terms > exactly > > right, so the user can find or remember them from elsewhere in the > > documentation, doesn't seem like the best solution. The material isn't > > that extensive and since it is covered elsewhere a little bit more > > explicitness in the DELETE and FROM documentation seems like a better > path > > forward. > > I see where you're coming from, but I do not think that repeating the > whole from_item syntax in UPDATE and DELETE is the best way forward. > In the first place, we'd inevitably forget to update those copies, > and in the second, I'm not sure that the syntax is all that helpful > without all the supporting text in the SELECT ref page --- which > surely we aren't going to duplicate. > > I think the real problem with the places Alexey is on about is that > they're too waffle-y. They use wording like "similar to", leaving > one wondering what discrepancies exist but are being papered over. > In point of fact, as a look into gram.y will show, what you can > write after UPDATE ... FROM or DELETE ... USING is *exactly* the > same thing as what you can write after SELECT ... FROM. So what > I'm in favor of here is: > > * Change the synopsis entries to look like "FROM from_item [, ...]" > and "USING from_item [, ...]", so that they match the SELECT > synopsis exactly. > > * In the text, describe from_item as being exactly the same as > it is in SELECT. > > +1
I didn't want a wholesale repetition but the whole "similar to" piece is indeed my issue and this addresses it sufficiently. David J.
