On 2022-Feb-13, Guyren Howe wrote:
> I’m back to just having no earthly idea why anyone who finds relations
> to be a productive tool for building a model would think that SQL
> being the only means to do that is Okay.
There are aspects other than technical reasons alone why some things
live on while "better" things barely survive without thriving, or even
die. For example, the fact that there are multiples companies
furthering the development of the SQL language means that there's a
group of engineers working to introduce improvements to the language
after extensive discussions and exposure to requests from users. If
Postgres decided to "fork" and go solitary with its own query language,
it would by itself have to produce all the language improvements, or be
left behind by the other products. And it takes a lot of effort to
produce those improvements. Have you looked at how SQL changed from one
version to the next?
Another aspect is inertia. The amount of software products that rely on
SQL is just too high. Suppose you came up with a technically-better
language that has all the current capabilities of SQL. Do you think a
majority of those products would immediately switch to the new language?
My guess is no, they wouldn't, because the safe bet is that SQL will
continue to work in 10, 20 years, while there is no certainty at all
that your new language would. So by ditching SQL, Postgres would no
longer be a database of choice for those products.
So, while SQL may not be the greatest possible relational language
possible, there are very good reasons for it to continue to be the
language of choice.
--
Álvaro Herrera PostgreSQL Developer — https://www.EnterpriseDB.com/
