Hi Haris, PostgreSQL had arrays like forever, some 20 yrs ago I built a genealogical-based indexing (tree-like), based on arrays and int_array contrib (which also was there since 2000's) - and besides ltree being in the core distro already.
In my first emails to the pgsql-sql list , I was almost advised to stay away from arrays since they were designed for internal relations and structures, but I didn't listened to them.. still happily running the app based on GIN indexes .
That's the difference between you and me. I only use whats actually considered standard and what's considered ok or good to use. I honestly don't think people get how easy it is to setup MongoDB with all that working and no customisation required. When I picked MongoDB for Geo stuff first I found benchmarks old a year which showed faster MongoDB geo search on non indexed data in comparison to PostgreSQL and it was least old benchmark. Others were couple of years old. Second, no extra config was required to use geo tools while PostgreSQL demanded some plugin enabling or whatever. Couldn't care less if it's doable out of the box because I automated mongo in 5 minutes and it gets the job done. Works with +20k polygons quite fast. Storing sensor values which could also contain objects or array is straightforward. Why would I bother with schema? I have no benefit of it nor do I care if this DB has also support for this or that. MS SQL and Oracle if I'm not mistaking have graph like support. Probably so does PostgreSQL or will have. Why would I use it instead of Neo4j? Because some enterprise dudes told me so? I store measurements in there which have different fields/properties. Then I can aggregate it regardless of different structure. Why would I use Postgres' JSON? I mean it's by default json-like why others have need to enforce using relational stuff?
I'm not fully against using some stuff out of standard but SQL-s great for some stuff and I would like to use others that are great for those other things. Basically MongoDB has been serving me quite well and it's well hated just like Java. And frankly people still use it
Ok. arrays might not be advertised by some emails I got in 2001, but they were standard, documented and fully supported already by the same year 2001, so it was pretty legit to use, maybe the guys who answered were relational purists, I cannot know, which year are you referring to in your comment about "relational" dbs not having arrays?
What's actually considered "standard" for most people is running Oracle or MS SQL Server. You want a second definition of "standard"? It is standard to spend SUBSTANTIAL time resources in advance for design and planning , every time my programmers talk to me about "5 minutes" I get terrified about what happens in 1 year or so. Dont get me wrong, bad code and bad practices are also super possible with (postgre|my|ms|DB2)sql and so forth, heck even hibernate terrifies me as I see those awfull queries in the pgbadger reports. But at least by a good design (and LOTS OF DB CONSTRAINTS / triggers/automations/integrity checks) you are doing your best about the durability, reliability and credibility of your system and happiness of your bosses. I am not familiar neither with mongodb geospatial nor with PostGIS, but I'd say here that nothing is free in life, if mongo makes it appear so appealing against a system which has been so heavily looked after and developed then I'd place my bets that there would be surprises sometime in the road ahead.
PS
Does mongo give you some way e.g. to disable ovelapping polygons? Like it is forbidden that two different non-affiliated owners have pieces of land that overlap one another?
I don't know nor did require such feature. I only seek where do given coordinates fit in. But good thing to investigate thanks fot the question.
5 minutes for dev purpose in the cloud. Developers were not born yesterday neither was I. No need to point the obvious that thing can go wrong and nothing is free. It's as simple as it works now and there's no proof of it falling later. I know some people would love it to fail to prove their point but I'm not like that, I'm not toxic person from stack overflow. I actually use stuff if it works and don't care about the negativity because in the end of the day it all breaks and people make it work with a lot of weird things so only way to proove something is bad is when no one made it work.
For further actions, you may consider blocking this person and/or reporting abuse
We're a place where coders share, stay up-to-date and grow their careers.
Hi Haris, PostgreSQL had arrays like forever, some 20 yrs ago I built a genealogical-based indexing (tree-like), based on arrays and int_array contrib (which also was there since 2000's) - and besides ltree being in the core distro already.
In my first emails to the pgsql-sql list , I was almost advised to stay away from arrays since they were designed for internal relations and structures, but I didn't listened to them.. still happily running the app based on GIN indexes .
That's the difference between you and me. I only use whats actually considered standard and what's considered ok or good to use. I honestly don't think people get how easy it is to setup MongoDB with all that working and no customisation required. When I picked MongoDB for Geo stuff first I found benchmarks old a year which showed faster MongoDB geo search on non indexed data in comparison to PostgreSQL and it was least old benchmark. Others were couple of years old. Second, no extra config was required to use geo tools while PostgreSQL demanded some plugin enabling or whatever. Couldn't care less if it's doable out of the box because I automated mongo in 5 minutes and it gets the job done. Works with +20k polygons quite fast. Storing sensor values which could also contain objects or array is straightforward. Why would I bother with schema? I have no benefit of it nor do I care if this DB has also support for this or that. MS SQL and Oracle if I'm not mistaking have graph like support. Probably so does PostgreSQL or will have. Why would I use it instead of Neo4j? Because some enterprise dudes told me so? I store measurements in there which have different fields/properties. Then I can aggregate it regardless of different structure. Why would I use Postgres' JSON? I mean it's by default json-like why others have need to enforce using relational stuff?
I'm not fully against using some stuff out of standard but SQL-s great for some stuff and I would like to use others that are great for those other things. Basically MongoDB has been serving me quite well and it's well hated just like Java. And frankly people still use it
Ok. arrays might not be advertised by some emails I got in 2001, but they were standard, documented and fully supported already by the same year 2001, so it was pretty legit to use, maybe the guys who answered were relational purists, I cannot know, which year are you referring to in your comment about "relational" dbs not having arrays?
What's actually considered "standard" for most people is running Oracle or MS SQL Server. You want a second definition of "standard"? It is standard to spend SUBSTANTIAL time resources in advance for design and planning , every time my programmers talk to me about "5 minutes" I get terrified about what happens in 1 year or so. Dont get me wrong, bad code and bad practices are also super possible with (postgre|my|ms|DB2)sql and so forth, heck even hibernate terrifies me as I see those awfull queries in the pgbadger reports. But at least by a good design (and LOTS OF DB CONSTRAINTS / triggers/automations/integrity checks) you are doing your best about the durability, reliability and credibility of your system and happiness of your bosses. I am not familiar neither with mongodb geospatial nor with PostGIS, but I'd say here that nothing is free in life, if mongo makes it appear so appealing against a system which has been so heavily looked after and developed then I'd place my bets that there would be surprises sometime in the road ahead.
PS
Does mongo give you some way e.g. to disable ovelapping polygons? Like it is forbidden that two different non-affiliated owners have pieces of land that overlap one another?
I don't know nor did require such feature. I only seek where do given coordinates fit in. But good thing to investigate thanks fot the question.
5 minutes for dev purpose in the cloud. Developers were not born yesterday neither was I. No need to point the obvious that thing can go wrong and nothing is free. It's as simple as it works now and there's no proof of it falling later. I know some people would love it to fail to prove their point but I'm not like that, I'm not toxic person from stack overflow. I actually use stuff if it works and don't care about the negativity because in the end of the day it all breaks and people make it work with a lot of weird things so only way to proove something is bad is when no one made it work.