RE: UPDATE/INSERT BY NAME

>I'm not clear what the advantages of the UPDATE BY NAME operation are >supposed to be. In interactive SQL, there are no records, so the syntax >is irrelevant to dbaccess etc. ... I think we agree we are only taking about enhancements to I4GL. > ... And in I4GL, all it saves is a little bit >of typing for less clarity of program. I don't see any major advantages to >the syntax. ... Clarity comes from documentation and common use. Is the INPUT BY NAME any less clear than INPUT var FROM tag? A little less typing is definitly a good thing. In general, having to change code in only one place makes maintenance simpler and cheaper. As an aside, I have worked with a couple of programmers who will only make the change in one place. They wait for the application to crash, in order to indentify the other places need changing. > ... Against it, it would be highly non-standard and bears no >resemblance to anything I know of which is likely to occur in any future >SQL standard. I'd have to rate UPDATE BY NAME as a non-starter. >I'm even less clear of the advantages of INSERT BY NAME. > >Someone else suggested CREATE TEMP TABLE LIKE record.* -- similar >arguments apply here too. I don't feel that the argument that enhancements must be part of the ANSI standard holds a lot of water. There are database statements in 4GL that do not appear in the ANSI standard. Either that, or I missed the section on FOREACH in the X3.135-1989 and X3.194 proposal. >So all in all, I'm not in the least bit convinced. You could try running >the proposal past me again since I deleted the original as something I >wasn't going to answer, but I don't think it'll make much difference to my >views on it. > >Yours unofficially, >Jonathan Leffler (johnl@obelix.informix.com) #include <disclaimer.h> Thank you for your reply (seriously). I would be happy to restate my suggestion for a "CREATE TEMP TABLE tab LIKE RECORD rec.*" but I think your reply demonstrates a more fundamental problem. I disagree when you say that your position is simply a personal opinion. I suggest that it is the unofficial corporate position. I have heard approximatly the same response from a couple of people in tech support over the years. I think the real issue of this thread is: What constitutes a good enough reason to make an enhancement to the 4GL language ? The answer seems to be: a change in the ANSI standard, or a request from an important client (e.g.: K-mart gets Informix-RX). I think that there are other good reasons. As someone who uses the Informix 4GL language, and advocates it's use to clients, I have other priorities. I work with a I4GL because I can do more work with less code. I like the power of the statements in I4GL. The use of I4GL statements speeds development and simplifies maintenance. This saves the client who chooses Informix 4GL money. This is a good marketing point for Informix, and they should follow up on it. It is an advantage they currently have over their competitors. Perhaps there should be a committee (with users) that evaluates suggestions like UPDATE BY NAME or CREATE TEMP TABLE LIKE on the basis of their contribution to the language as a development environment. This would demonstrate to the market a serious commitment on the part of Informix to have the best database development language on the market. It would also mean less typing for me :-). ------------------------------------------------------------------------ David Stockel Internet: dccrjc9@unix.bell-atl.com of ADIA Information Technologies Voice: (215) 466-3216 at Bell Atlantic Fax: (215) 665-1329 1717 Arch Street 4-N-8 Philadelphia, PA 19103 Proudly serving my corporate masters