4 ms·
Their argument from the introduction that software can slow down a business by making changing/evolving the systems difficult and requiring programmers at every
by dmix 7y ago
Their argument from the introduction that software can slow down a business by making changing/evolving the systems difficult and requiring programmers at every step, is an interesting one:
http://blog.aquameta.com/introducing-aquameta/ http://blog.aquameta.com/introducing-aquameta/
Obviously this is has been the panacea goal in business software forever and there's a long trail of failed companies or projects who tried to do this.
IMO it's always going to require specialized knowledge of a certain level of abstraction above the machine, it will never be as simple as pushing buttons in a GUI. The question is making the languages/frameworks simpler and lowering the bar.
In practice these sorts of things work great for simple scenarios but are very brittle one you start to. Just like Excel spreadsheets used like databases it will quickly turn into a hacky maze of things forced into places where it shouldn't be.
I'd also never think 'PLpgSQL' when it comes to simplifying things. I'm genuinely curious to see if they can pull it off... eliminating the file system part is also an interesting idea.
- pm90 7y agoThe MO of SV startups seems to be to slap together something that has product market fit while ignoring performance and scaling concerns, at first. Once you have enough customers and market, you then raise funding and use that to hire more experienced system folks who can make your system performant, efficient and scalable.
- dmix 7y agoI wasn’t talking about scaling performance wise. I mean scaling the problem set beyond the initial easy ones.
- cryptonector 7y agoIt's not that using PlPgSQL simplifies things, but that putting business logic in the RDBMS does simplify things. It does so by letting you use an ecosystem of tools you otherwise could not use. There are several tools that let you serve a DB via any number of protocols, including RESTful APIs, for example. You could even give out direct SQL access to power users and not have to teach them how to maintain referential integrity. PlPgSQL is not a great language, and that's obviously not a reason to use it. But it's not obviously a reason to not use it, and more than that, it's no reason not to put business logic in the RDBMS. Once you decide to put business login in the RDBMS, PlPgSQL is just one of several languages you could use, and actually the most accessible one.