SQL queries are pretty readable. Of course, QUEL was much better, but SQL won. I would say that SQL queries are more readable than 90% of “NoSQL” novelties.
Nah, especially as SQL can work as graph or document DB with success. At the same time most of the problems are better described with relational algebra.
Yet, here we are discussing language that is based on technology which is 40 years old (Erlang).
That is impossible dream. Data handling will always be hard, because each data is different.
There are many databases build on the beam/erlang. To the mentioned I’d add antidote db and barrel db – there are probably quite a few more. Both are quite interesting, but certainly not as popular as e.g. couchdb or riak. So the BEAM is full of options in that space. What you feel to be a favor for sql (postgres) is more the fact that databases are very complex systems, which need time to be developed and some hard questions to be answered competently. Therefore people are even more likely to stick to proven tech which they know and especially which they know works e.g. compared to a runtime programming language. Postgres seems to work for a lot of people/use cases, and the adoption does get even bigger for sql based databases, so including mysql, mssql, ….
No. It’s so 80’s (with numerous updates). SQL was not standardized until the late 80’s. I know those 20 years difference may not mean much to you personally, but when you’re arguing about something being outdated, it hurts your argument when that thing is 20 years younger than you claim.
You’re right.
This means I’m old because that’s the way my generation expressed when things were the old way of doing things.
Nevertheless, in a more serious note my point is that Elixir is all about simplicity and convenience and Relational DBs are not by themselves those, specially when it comes to the new unstructured data, human way of interact with software and specially distributed and scalable systems.
Yes, I agree with a lot of those arguments but the truth is that the future is more and more about unstructured data, relationships, distributed architectures, more than simple interfaces,…
Off course I recognise the excellent work done by Ecto to simplify the way we interact with Databases. But the fundamental problem is still there…
Yes, but those new projects seem to be paused or stopped…and making it easy to support a lot of existing relations DBs is a worrying sign as it means you’re locked into that model.
And, liked or not, new developers (not literally) don’t know SQL don’t think that structured as they use to do 20 years ago, mostly due to Javascript and Nodejs explosion, that makes it easy to copy paste a lot of code and things work. Until they don’t.
But that’s exactly why I think Elixir could embrace that problem and make persistence and manipulating data as simple as it is to tackle real-time, for example.
Rise of strongly typed languages makes this statement weird. I do not think that this is true, we still operate mostly on highly structured data.
And what makes SQL not fit in that place? It also depends on what kind of relations we speak there. Relations in sense of relations form relational algebra or joins from relational algebra? People often confuse these two.
And how is that relevant to SQL? There is hell lot of so called “NewSQL” DBs that work perfectly fine in distributed environments. All of that in the end is about CAP, where “traditional SQLs” put all their cards on CP. If Google managed with regular SQL for so long, then you, with your toy service, will also manage.
In fact I see Ecto query as a return to the QUEL which was THE query language for relational algebra (as it was created by Codd himself. It is important to remember that PostgreSQL started as POSTGRES which used POSTQUEL as a query language. So this is more like going back to where relational algebra was done by someone who know it well rather than something novel.
And I do not see that “fundamental problem”. Not at all.
I would say that this is problem with “new developers” not with SQL…
I would say that this is a reason to keep SQL, not to abandon it…
It cannot, because persistence and manipulating data is hard. It is so hard, that there is whole branch of engineering built around it. That branch is called programming. There is no way to make it simple, as there is no simple way to tell what is “data”, you cannot describe what you want to do with “data”, you often cannot define how that “data” will be encoded nor how it be stored.
Sorry if this feel a little like Linus on C++, but it has been said that SQL is the worst form of data query language except all the others that have been tried.
You might want to be cautious about generational … generalizations. Your overall point that relational databases are obsolete sounds like things I hear from people half my age
I would counter that Elixir builds atop the foundations of Erlang for fault tolerance and stability – invented by the fine team at Ericsson not long after SQL was first ratified. It’s incongruous to laud one while denigrating the other. “Simple ain’t Easy” requires a solid foundation.
By all means, I’d love to see the Elixir community explore alternate data stores, but not at the expense of first having Postgres as a solid/safe default choice.
You’re right.
