Re: BUG #13708: strange behaviour instead of syntax error
От | Pavel Stehule |
---|---|
Тема | Re: BUG #13708: strange behaviour instead of syntax error |
Дата | |
Msg-id | CAFj8pRDMbPPUkzPsKmVyFoFDiFGOP_ocL3uShc+ZE8hrdvDdnA@mail.gmail.com обсуждение исходный текст |
Ответ на | Re: BUG #13708: strange behaviour instead of syntax error (Greg Stark <stark@mit.edu>) |
Список | pgsql-bugs |
2015-10-24 14:04 GMT+02:00 Greg Stark <stark@mit.edu>: > > On 24 Oct 2015 00:03, "Tom Lane" <tgl@sss.pgh.pa.us> wrote: > > > > Hmmm ... maybe just adding parens would do the trick. > > > > regression=# SELECT ('ERROR: No uncollected prize found!' RETURN); > > ERROR: syntax error at or near "RETURN" > > > > That looks more promising ... though I'm not sure if there are > > any cases it would break. > > That does seem like a good idea. There are plennty of keywords that can > follow a target list that would lead to strange behaviour, not the last of > which would be FROM... Adding parentheses should mean only a valid > expression should be accepted. > I used same technique in PLPSM - and it is working. But If I remember well, there will be complication with multi assign SET (a,b,c) = (10,2,3) because I had other level of nesting. But PLpgSQL doesn't support thes feature. Cannot be used a bison functionality - "multiple start points" http://www.gnu.org/software/bison/manual/html_node/Multiple-start_002dsymbols.html ? Regards Pavel
В списке pgsql-bugs по дате отправления: