Re:The ORM Debate Still Rages On
by diminoten on Monday April 24, 2017 @02:54PM
(#14186947)
I intuitively feel like, in $CURRENT_YEAR, this problem comes up mostly for the people who look for it to be a problem.
I work with Django's ORM on a daily basis, and I think being aware of the fact that I can't completely ignore how I interact with my database (e.g. "Well I have an ORM so I shouldn't care" <- this is bad) is enough. The idea that I throw out the proverbial baby here feels like an overcorrection, and while I get how complicated an ORM can become, I think that's more of a result of relying too much on the abstraction.
I'm sure most folks here are familiar with Joel Spolsky's "Leaky Abstractions" article. The way I see it, the ORM abstraction is leaky, and that's okay, because the alternatives are all so involved and complex. For some higher-than-fifty percentage of the time however, the abstraction works. That has value.
I work with Django's ORM on a daily basis, and I think being aware of the fact that I can't completely ignore how I interact with my database (e.g. "Well I have an ORM so I shouldn't care" <- this is bad) is enough. The idea that I throw out the proverbial baby here feels like an overcorrection, and while I get how complicated an ORM can become, I think that's more of a result of relying too much on the abstraction.
I'm sure most folks here are familiar with Joel Spolsky's "Leaky Abstractions" article. The way I see it, the ORM abstraction is leaky, and that's okay, because the alternatives are all so involved and complex. For some higher-than-fifty percentage of the time however, the abstraction works. That has value.