Brian Aker wrote:
Hi!
On May 3, 2009, at 8:30 AM, Jim Starkey wrote:
On the other hand, if you want to start over from scratch, I've got
all sorts of good ideas...
As a few folks have pointed out, we should support quoted identifiers.
That being said though, I am still curious as to what you are thinking :)
First, a disclaimer. I designed a series of DML languages starting with
DataLanguage for the ARPAnet Datacomputer project, the Datatrieve 4GL at
DEC, and GDML at Interbase. The languages were, I believe, superior to
SQL in almost all respects. That said, SQL won, and I've tired of
swimming upstream. I've adopted a personal philosophy of SQL with "no
arbitrary changes". This radically reduces the amount of time I spend
arguing about syntax, which I regard as only slightly more interesting
than your interest in arguing about open sources licenses.
That said, I believe a new language should:
1. Have case sensitive identifiers, as is the norm for modern languages.
2. Have a record select expression (RSE) can be used for retrievals,
loop, and subexpressions.
3. Have an integral looping statement, a conditional statement, an
aggregating statement, and, of course, support nested statements.
4. Have a little "reverse human engineering" like "delete all of
<rse>" to make destructive commands clearly destructive
In short, a procedural language with an interactive and embeddable subsets.
It would be better, but in my opinion, not sufficiently better than SQL
to bother with. SQL is widely accepted and adequate.
(I'd also put in multi-table, multi-column text search and semantic
extensions to support type hierarchies.)
--
Jim Starkey
President, NimbusDB, Inc.
978 526-1376
_______________________________________________
Mailing list: https://launchpad.net/~drizzle-discuss
Post to : [email protected]
Unsubscribe : https://launchpad.net/~drizzle-discuss
More help : https://help.launchpad.net/ListHelp
