Re: IDS Feature Request List (including potential new requests).
A discussion branch of an IDS feature-request thread about adding SQL-standard row/tuple comparison and, by extension, a MIN() over tuples. Participants note IDS does support ROW complex types but question what MIN(row_type) would mean, and that an aggregate must return a scalar, row or table, raising the issue of returning/unnesting tuples to the client. The proposer suggests defining it as the tuple that compares <= all others per SQL-92 rules, then backs off after reading SQL99, complaining the standard has become bloated. The exchange drifts into jokes about standards committees and Ada; no resolution or implementation decision is recorded.
Auto-generated by DrWatson from the posts below — may be imperfect; read the full thread.
"Serge Rielau" <srielau@ca.ibm.com> wrote in message news:43ef5cF1jqph5U1@individual.net... > bozon wrote: > > Oh, I thought you were against both. Sorry about that. Still it seems > > like min((tuple)) is a natural consequence of the first feature, > > because essentially min is a short-cut for "compare" all of the values > > and give me the lowest one. > But now you need to get into the world of passing back tuples as column > values to the client or otherwise unnesting the tuple. > You can't just run a fucntion (MIN) and have it spit out columns. > It has to spit out a scalar, a row/tuple or a table. > Not clear how far IDS's tuple support goes... > IDS has support for the row type complex type. Although I'm not quite sure what a MIN(row_type) really is. ;-)
Madison Pruet >>IDS has support for the row type complex type. I thought about this too. I haven't used this feature enough to be an expert at it. It is on my list of things to learn. >>Although I'm not quite sure what a MIN(row_type) really is.;-) My proposal was to define it the way tuple comparison was defined. min(rowtype(a,b,c)) would be the tuple that would be less than or equal to all other tuples. (a,b,c) <= (ai,bi,ci) where i represents any other tuple in the database. Where Tuples are compared the SQL 92 standard way. Of course after looking at the SQL99 standard I can't argue in good conscience that if it is in the standard it should be in Informix anymore. What the hell were they thinking? Take a reasonably simple and elegant language and add to it until collapses under its own weight. Was someone trying to sabatoge the SQL standard?
bozon said: > > Of course after looking at the SQL99 standard I can't argue in good > conscience that if it is in the standard it should be in Informix > anymore. What the hell were they thinking? Take a reasonably simple and > elegant language and add to it until collapses under its own weight. > Was someone trying to sabatoge the SQL standard? I think you're just too dumb to use all the clever stuff that the ivory tower brigade have dreamed up. Like me. -- Bye now, Obnoxio "C'est pas parce qu'on n'a rien ' dire qu'il faut fermer sa gueule" - Coluche did i mention i like nulls? heck, i even go so far as to say that all columns in a table except the primary key could/should be nullable. this has certain advantages, for example, if you need to insert a child record and you don't have a parent row for it, just do an insert into the parent table with the primary key value (everything else null), and voila, relational integrity is preserved. but this is, admittedly, a bit controversial among modellers. --r937, dbforums.com
bozon wrote: > Madison Pruet > >>>IDS has support for the row type complex type. > > > I thought about this too. I haven't used this feature enough to be an > expert at it. It is on my list of things to learn. > > >>>Although I'm not quite sure > > what a MIN(row_type) really is.;-) > > My proposal was to define it the way tuple comparison was defined. > min(rowtype(a,b,c)) would be the tuple that would be less than or equal > to all other tuples. (a,b,c) <= (ai,bi,ci) where i represents any other > tuple in the database. Where Tuples are compared the SQL 92 standard > way. > > Of course after looking at the SQL99 standard I can't argue in good > conscience that if it is in the standard it should be in Informix > anymore. What the hell were they thinking? Take a reasonably simple and > elegant language and add to it until collapses under its own weight. > Was someone trying to sabatoge the SQL standard? > Unfortunately standard comittees are notoriously political. I wouldn't put sabotage beyond anyone there ;-) Cheers Serge -- Serge Rielau DB2 Solutions Development DB2 UDB for Linux, Unix, Windows IBM Toronto Lab
Obnoxio The Clown Said >>I think you're just too dumb to use all the clever stuff that the ivory >>tower brigade have dreamed up. Uh, yeah. With a name like Bozon you can't expect too much. ;-) I am kind of thinking now that the designers of Ada were looking for work.
bozon said: > > Obnoxio The Clown Said >>>I think you're just too dumb to use all the clever stuff that the ivory >>>tower brigade have dreamed up. > > Uh, yeah. With a name like Bozon you can't expect too much. ;-) That's less of an indictment of you than it is of them. :o) > I am kind of thinking now that the designers of Ada were looking for > work. And found it. -- Bye now, Obnoxio "C'est pas parce qu'on n'a rien ' dire qu'il faut fermer sa gueule" - Coluche did i mention i like nulls? heck, i even go so far as to say that all columns in a table except the primary key could/should be nullable. this has certain advantages, for example, if you need to insert a child record and you don't have a parent row for it, just do an insert into the parent table with the primary key value (everything else null), and voila, relational integrity is preserved. but this is, admittedly, a bit controversial among modellers. --r937, dbforums.com