Comments on: To Specify the location for New Column in Oracle

By: kccrosser

kccrosser — Fri, 11 Mar 2011 21:13:06 +0000

DangerMouse – the same comment about “select *” applies in reverse for Insert statements. Instead of using

insert into tablename values (...valuelist...)

you should always enumerate the inserted column names, like

insert into tablename (...columnnamelist...) values (...valuelist...)

Then, as long as any new columns are not specified as Non-Null, your old Insert statements will still work. Using positional assumptions about database table columns is not a recommended practice with any database.

By: carlosdl

carlosdl — Thu, 10 Mar 2011 13:51:08 +0000

I don’t see how the last comments relate to the original question, which was about adding new columns to a table (not about inserting new rows).

By: dagnermouse

dagnermouse — Wed, 09 Mar 2011 22:00:21 +0000

That’s just not true. If you’re writing COBOL code that gets run through the Pro*COBOL precompiler, and you’re using VARCHAR2 or LOB data types, and you do an INSERT, and your column order as spec’d in the COBOL does not match EXACTLY the column ids of your Oracle table definition, the INSERT will flat out fail. So how in the hell do we get around this, eh? Nobody seems to know, but the problem clearly exists.

By: meandyou

meandyou — Thu, 24 Sep 2009 20:19:18 +0000

My favorite answer to questions such as “Why doesn’t company/product XYZ allow …”
is to say that during development this very topic was discussed, they flipped a coin, and it came up heads, and they decided not to …

By: carlosdl

carlosdl — Thu, 24 Sep 2009 01:08:42 +0000

Valuable points, and explanations. However, Why Oracle doesn't let you do the same, and "Is there any option/chances in the up coming versions of Oracle" are unanswered questions so far, and that's why I said "I'm afraid only Oracle could answer these questions".

By: mrdenny

mrdenny — Thu, 24 Sep 2009 00:43:04 +0000

MySQL may appear to let you do this, but what it is probably doing in the background is creating a new table with the column where you place it, moving all the data into the new table, dropping the old table, then renaming the new name.

Microsoft SQL allows you do to this, by using that exact process. It even shows you that its doing this.