News for hackers. Stuff that matters.
Re:ORMs Are Backwards by hyperpallium on Wednesday September 04, 2019 @04:55AM (#20875270)
> But then again: Is it even a good idea to access the same database from two codebases at all?

A motivation for relational algebra was different applications accessing data in different ways. This integration remains a selling point. Further, data often outlasts applications... even languages.

Re:ORMs Are Backwards by bayesian_horse on Thursday September 05, 2019 @01:19AM (#20884039)
Still, I think accessing a single database in RDBMs like Postgres or MySQL from multiple codebases or even languages means asking for trouble.

Providing some kind of API as a library, REST, RPC or whatever would almost always be preferable, if only to centralize permissions and validation.

Re:ORMs Are Backwards by _visgean on Wednesday September 04, 2019 @07:04AM (#20875940)
the problem is that these days you need a lot of application code to check for permissions and that kind of stuff.
Re:ORMs Are Backwards by dllthomas on Wednesday September 04, 2019 @02:36PM (#20880116)
In principle permissions can go in the DB.
Re:ORMs Are Backwards by bayesian_horse on Thursday September 05, 2019 @01:15AM (#20884026)
In practice this is rarely ever done (right).

Row-level permissions are possible, but it can get more complicated when certain users are only allowed to change a certain row in a certain kind of way but not in another.

Re:ORMs Are Backwards by dllthomas on Thursday September 05, 2019 @02:50PM (#20889657)
I agree, but it does seem to become more compelling if you're doing your integration in the database.