Roy Lyseng wrote: > Jay Pipes wrote: >> Roy Lyseng wrote: >>> Brian Aker wrote: >>>> Hi! >>>> >>>> On Aug 12, 2008, at 12:28 AM, Roy Lyseng wrote: >>>> >>>>> why do you need both TIMESTAMP and DATETIME? >>>>> TEXT should be named CLOB to be aligned with SQL standard... >>>> I am up for whatever folks want here. TEXT has always been the MySQL >>>> type. I do not mind using the standard type name (or aliasing it). >>>> >>> If there is a standard, why not use it? Especially one that is so >>> comprehensive and consistent as the SQL standard (for most parts) :) >> >> Consistent? Perhaps. Completely dated, vendor-centric, and geared >> towards legacy environments? Absolutely. > > So, when MySQL implemented subqueries, was anything but standard SQL > syntax considered? Views? Outer joins?
Yes, things outside the standard were indeed considered (although Peter Gulutzan would be the expert voice on this) and AFAIK, the parts of the SQL standard that were reasonable were used. But...subqueries were a feature that was supported on 100% of other RDBMS and had practical uses. CLOB? It's a throwback to Oracle. What I am saying is that each piece of the parsed syntax should be looked at with new eyes, and not just "let's go with the standard". If the standard is good and reasonable for the audience we are targeting, cool. If not, I'm saying, don't worry too much about the standard. That's all. >> Sticking to the standard means very little to me, so -1 on this. If the >> standard is reasonable and pertains to modern systems, sure, fine by me, >> but I am not opposed to breaking standards, especially when they were >> made by vendors 20 or more years ago and then bolted onto to fit needs. >> >> Standards are good for portability, but in this day and age, one has to >> wonder what, realistically, is the likelihood of writing portable >> applications to run on multiple RDBMS. Standard and portable SQL is a >> complete myth IMHO, and applications which try to be portable suffer >> from poor performance by aiming at lowest common denominators. >> >> Drizzle isn't really aimed at being the database for VARs or OEMs that >> want a standards-compliant RDBMS on which to write some enterprise >> application. Instead, it aims to tackle the specific problems of the >> web and cloud environments. > > Standards are not only good for portability. They also mean that a > developer can move from one database to another without having to learn > a completely new semantics. Isn't that what portability is? > Of course that does not happen in the real > world, but that is because of the vendors, not the standard. And it's the very same vendors which wrote the standard. > I do not mind extensions to the SQL standard, but incompatibilities > should generally be avoided. Most incompatibilities are also confusing > and illogical, IMHO. I find the opposite is often true. Think of the LIMIT with OFFSET clause that was developed due to Rasmus Lerdorf's insistence and put into MySQL. This syntax is 10 times easier to understand and write than either SQL Server's TOP syntax or Oracle's ROWID or even the standard's use of nested subqueries. So, like I said, each case must be looked at to see if the standard is worth preserving. Cheers, -jay _______________________________________________ Mailing list: https://launchpad.net/~drizzle-discuss Post to : [email protected] Unsubscribe : https://launchpad.net/~drizzle-discuss More help : https://help.launchpad.net/ListHelp
