Received: from jupiter.jw.home (A5f53.pppool.de [213.6.95.83]) by mail.postgresql.org (8.11.1/8.11.1) with ESMTP id eAHDsms42839 for ; Fri, 17 Nov 2000 08:54:49 -0500 (EST) (envelope-from janwieck@Yahoo.com) Received: (from janwieck@localhost) by jupiter.jw.home (8.9.3/8.9.3) id EAA01528; Wed, 15 Nov 2000 04:53:02 -0500 From: Jan Wieck Message-Id: <200011150953.EAA01528@jupiter.jw.home> Subject: Re: Requests for Development In-Reply-To: <3A0AD389.647EE03A@agliodbs.com> from Josh Berkus at "Nov 9, 2000 08:40:41 am" To: josh@agliodbs.com Date: Wed, 15 Nov 2000 04:53:02 -0500 (EST) CC: sqllist X-Mailer: ELM [version 2.4ME+ PL68 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit X-Archive-Number: 200011/134 X-Sequence-Number: 165 Josh Berkus wrote: > Tom, Bruce, Jan, etc.: > > [...] > > The rest of these requests apply to 7.2: > > 2. Stored Procedure functionality, i.e. outputting a full recordset from > a function (or new structure, if functions are hard to adapt) based on > the last SELECT statement passed to the function. An alternative would > be to develop parameterized views, which might be the easiest path. That's one of my favorite requests, and I'd be glad to have a chance to start on it. Unfortunately the basic support in the parser and other parts of the core engine isn't completely planned yet, otherwise PL/pgSQL and PL/Tcl would've had this from the very beginning. > 3. Slightly more informative syntax error messages - frankly, just > grabbing a little more text around the word or punctuation that > triggered the error would be enormously helpful (I can't tell you the > number of times I've gotten "Error at or near ')'" in a huge DDL > statement. That's a general problem of a lex/yacc parser and I'm not sure how to force it to be a little more explanative. Maybe we have a chance to grab something from the lex input buffer, but IIRC that's unsafe because nobody knows how much of that is already eaten into yacc tokens. > 4. Use of named in addition to ordinal variables in PL/PGSQL functions > (e.g. $account_type, $period instead of $1, $2). Another general problem in the core engine. Dunno if we'll have named arguments in the near future. In the meantime, PL/pgSQL functions can use ALIAS to define the names for arguments at the very top (it's a precompile time only thing, so there is little to no performance impact). And PL/Tcl functions could easily do a "set account_type $1" as well, so I don't see a real problem for the readability of the functions body. To put the ball back into your yard, I'd like to make a request too. There seem to be alot people using PL/pgSQL and/or PL/Tcl extensively. OTOH there are newbies again and again asking for a good tutorial, programming examples and so on. Writing a good tutorial doesn't require a good backend developer, IMHO an experienced SQL-programmer would be the better guy anyway. During the past 4 years I've heard over and over that people would like to contribute their $0.05 if they only could code in C. That's an area where nobody needs any C experience. Jan -- #======================================================================# # It's easier to get forgiveness for being wrong than for being right. # # Let's break this rule - forgive me. # #================================================== JanWieck@Yahoo.com #