Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!cs.utexas.edu!usc!ucsd!ucsdhub!hp-sdd!hplabs!hplabsz!davis From: davis@hplabsz.HPL.HP.COM (Jim Davis) Newsgroups: comp.databases Subject: Re: variable record length Summary: *requiring* user specified maxima considered harmful Message-ID: <4777@hplabsz.HPL.HP.COM> Date: 3 Feb 90 09:52:14 GMT References: Sender: davis@hplabsz.HPL.HP.COM Reply-To: davis@hplabs.hp.com (Jim Davis) Distribution: usa Organization: Hewlett-Packard Laboratories Lines: 67 The obligatory synopsis may be found below my remarks. Michael Friedman asked: mf> Mind telling me what I could do with variable length mf> columns that I can't do with regular Oracle columns? The major capability missing is the ability to program with assurance that a datum from one column could be stored into another. If the only limit is the system limit and it is imposed without the necessity of common agreement by all concerned on an explicit value then the resulting "code" is often far better. As an example of this consider code to handle user ASCII input. Some people (and code) consider 160 characters as a reasonable maximum line length for user input. Some take 1024 or 4096 characters as a single line. The better ones will handle millions of bytes. Note that I do consider *allowing* user specified maximum column lengths to be advantagous, it is requiring them that concerns me. -- Jim Davis (Of course I speak for myself and not my company.) ----------------------------------------------------------------- Synopsis: "or you can't tell the players without a program card" Young-Mee Yang asked: ymy> How about others, Oracle, Sybase, Ingres? ymy> Do they allow variable record length? David E. Anderson answered: dea> Subject: Re: variable record length dea> Summary: Oracle supports variable length records dea> The subject says it all... Always Had It, Always Will Jonathan Krueger wrote: jk> Oh, good. Please show me your SQL statement that creates a table jk> with a text column without a specified size. David S. Masterson suggested: dsm> create table testing ( dsm> col1 varchar(80) dsm> ); dsm> The key is in knowing that the "80" is a specified MAXIMUM size. ;-) Dave Simson wrote: ds> The length in the SQL statement is the maximum length. Oracle *stores* ds> all character columns as variable length (ie not taking up more space ds> than needed for the data in question) so all character columns are in ds> effect variable length. Jonathan Krueger retorted: jk 2> Thanks for the clarification. As we see, Oracle has never had it, and jk 2> as far as we know, never will. jk 2> "It" is variable record length, not trailing whitespace jk 2> compression. The original article never asked about the latter, jk 2> the question was simply which databases support the former. jk 2> If you must put your marketing claims on the net, please limit jk 2> yourself to accurate information that answers the question put. Michael Friedman asks: mf> I'm just an applications developer. I'm not a marketer and I'm not a mf> kernel developer, so I'm really not interested in the fancy names or mf> the theoretical schtick. I'm interested in what I can produce for mf> customers. mf> Mind telling me what I could do with variable length mf> columns that I can't do with regular Oracle columns?