Thread: On naming of executables

As a new user of Postgres I'm wondering about the naming of certain
executables (createdb, dropuser, etc). These are fairly ambiguous
names. They could just as well apply to the OS or an other DBMS. Why
not prefix them? (e.g. `pg_createdb` such as several other scripts).

I suppose this is a case of bikeshedding and I realize I am free to
rename them; I'm just curious as to why this is.

Mark Davidson <chemosh9@gmail.com> wrote:

> As a new user of Postgres I'm wondering about the naming of
> certain executables (createdb, dropuser, etc). These are fairly
> ambiguous names. They could just as well apply to the OS or an
> other DBMS. Why not prefix them? (e.g. `pg_createdb` such as
> several other scripts).

> I'm just curious as to why this is.

I checked the earliest tag (which is over 16 years old) in our git
repository, and those executable names were already in use by that
time.  (They may go much earlier, but I'm not sure how to check
that.)  I think that in the absence of any name conflicts which are
causing anyone pain, and in the face of the certainty that changing
them would break many existing production scripts, nobody has been
very inclined to spend their limited time making the change.

In other words: a historical artifact with a fair amount of inertia.

-Kevin

Mark Davidson <chemosh9@gmail.com> writes:
> As a new user of Postgres I'm wondering about the naming of certain
> executables (createdb, dropuser, etc). These are fairly ambiguous
> names. They could just as well apply to the OS or an other DBMS. Why
> not prefix them? (e.g. `pg_createdb` such as several other scripts).

Yeah, if people had thought ahead fifteen years ago they would surely
have been named like that.

At this point, though, it seems like the costs of changing would
outweigh the benefits.  (I recall at least one fairly serious proposal
to rename them, but it didn't pass, and now there are even more years
of precedent there.)

            regards, tom lane