PostgreSQL 12 Due Out Tomorrow With Better Performance

[zboszor]

I do read the manual when everything else fails, and I do not chose ignorance. But the user space is so shitty that you are forced to resort to read the manual. A sane userspace would be well-designed and intuitive. The problem here is not ignorance, it is a poor non-intuitive userspace that places unnecessary burden on the user. It is not user-friendly.

There is no reason things shouldn't be user-friendly. There is no reason things shouldn't be intuitive.

It is very lame to have a shitty user space and then blame the user for being ignorant.

A mature project can't just rename utilities whose names are in use for 25+ years.
UNIX is also riddled with such utility names. The "cat" utility is an non-intuitive nickname of "concatenate".
A variant of "cat" that list file contents in reverse is "tac" and it's also not intuitive at all.

Or when Dennis Ritchie admitted that his greatest regret about the C language is not spelling "create" in full for the creat() function.
I am not sure he was joking or that he had to think about machines with very limited amount of RAM where every byte counted.

The notion of being "user friendly" was very different 20 years ago. There's also a thing called "historical accident". Hindsight is always 20:20.

If you are so keen on PostgreSQL being user friendly, create a bug report for them asking to rename the so called "inconsistent" utilities and see how your bug report will be weighed against the many existing users. https://www.postgresql.org/account/submitbug/

[uid313]

Even the Book of Murphy's law had a line like "when all else fails, read the manual". Another way to put it is "theory before practice". Don't choose ignorance.

I do read the manual when everything else fails, and I do not chose ignorance. But the user space is so shitty that you are forced to resort to read the manual. A sane userspace would be well-designed and intuitive. The problem here is not ignorance, it is a poor non-intuitive userspace that places unnecessary burden on the user. It is not user-friendly.

There is no reason things shouldn't be user-friendly. There is no reason things shouldn't be intuitive.

It is very lame to have a shitty user space and then blame the user for being ignorant.

[zboszor]

Sure I could, there are always things I could do, like read the source code for undocumented features, and such. But I shouldn't have to. The PostgreSQL user space has a problem with it being inconsistent and non-intuitive.

Even the Book of Murphy's law had a line like "when all else fails, read the manual". Another way to put it is "theory before practice". Don't choose ignorance.

[uid313]

Fair point.
Tho, you could check through commandline where's the binary located at or just try "man createuser" and see what would be written in it's 'man' file.

Sure I could, there are always things I could do, like read the source code for undocumented features, and such. But I shouldn't have to. The PostgreSQL user space has a problem with it being inconsistent and non-intuitive.

[aht0]

Some utilities are prefixed with pg_ while others are not. It is confusing because "createuser" sounds like something that adds a user to the system, not PostgreSQL. It is confusing because if I have MongoDB and PostgreSQL installed, then how am I supposed to know which database "createdb" creates a database for?

Fair point.
Tho, you could check through commandline where's the binary located at or just try "man createuser" and see what would be written in it's 'man' file.

[zboszor]

Yeah, that. But the tooling is also non-modern with dozens of scattered binaries, compared to tools like dotnet, apt and git. With Git you do all Git-related operations using the "git" tool.

You can just use "psql" and run the SQL commands that are exposed in those utilities. See, PostgreSQL does have a consistent interface.

[uid313]

How is the PostgreSQL license not a real license? The license lets you do anything you want with it so long as you don't hold U of C accountable. It's extremely short and readable:

https://www.postgresql.org/about/licence/

What specifically is "inconsistent" or "unpredictable" about interacting with PostgreSQL?

Because it is a license created by PostgreSQL instead of using an already existing widely spread, common license that states the same thing, such as the BSD License, ISC license or the MIT license.

Some utilities are prefixed with pg_ while others are not. It is confusing because "createuser" sounds like something that adds a user to the system, not PostgreSQL. It is confusing because if I have MongoDB and PostgreSQL installed, then how am I supposed to know which database "createdb" creates a database for?

Possibly he calls the historical accidents (that are "createdb", "createuser" and so on) inconsistent, comparing with newer utilities in PostgreSQL that are prefixed with "pg_". And yes, it's actually inconsistent but they are long set in stone. Historical accidents happen that seemed like a good idea at the time, like the "creat" function in C. You can't rename it to "create" now.

The original utility names were already there in Postgres95 (this was the version in 1995 that replaced the query language POSTQUEL with standard SQL) and I suspect even before that when the project was still called POSTGRES, in capitals, no SQL, those utilities existed. You can't change a habit with over 30 years of history. It's called tradition at this point and newcomers should just accept the way it is without calling names.

Yeah, that. But the tooling is also non-modern with dozens of scattered binaries, compared to tools like dotnet, apt and git. With Git you do all Git-related operations using the "git" tool.

[zboszor]

How is the PostgreSQL license not a real license? The license lets you do anything you want with it so long as you don't hold U of C accountable. It's extremely short and readable:

https://www.postgresql.org/about/licence/

What specifically is "inconsistent" or "unpredictable" about interacting with PostgreSQL?

Possibly he calls the historical accidents (that are "createdb", "createuser" and so on) inconsistent, comparing with newer utilities in PostgreSQL that are prefixed with "pg_". And yes, it's actually inconsistent but they are long set in stone. Historical accidents happen that seemed like a good idea at the time, like the "creat" function in C. You can't rename it to "create" now.

The original utility names were already there in Postgres95 (this was the version in 1995 that replaced the query language POSTQUEL with standard SQL) and I suspect even before that when the project was still called POSTGRES, in capitals, no SQL, those utilities existed. You can't change a habit with over 30 years of history. It's called tradition at this point and newcomers should just accept the way it is without calling names.

[alcalde]

PostgreSQL really is very full of features. Feature-wise it is really good.
I wish it has a real license, instead of a vanity license though, and a better more sane userspace (the user space is inconsistent, unpredictable and confusing with lots of scattered binaries), and good administration utilities.

How is the PostgreSQL license not a real license? The license lets you do anything you want with it so long as you don't hold U of C accountable. It's extremely short and readable:

https://www.postgresql.org/about/licence/

What specifically is "inconsistent" or "unpredictable" about interacting with PostgreSQL?

[uid313]

PostgreSQL really is very full of features. Feature-wise it is really good.
I wish it has a real license, instead of a vanity license though, and a better more sane userspace (the user space is inconsistent, unpredictable and confusing with lots of scattered binaries), and good administration utilities.