Re: pgsql: Add documentation for the JIT feature.

On Thu, Mar 29, 2018 at 3:00 PM, Stephen Frost <sfrost(at)snowman(dot)net> wrote:
> What we've done elsewhere when there's been similar jargon is to say
> something along the lines of:
>
> "compiling of routines (also known as Just-In-Time or JIT compilation)"

That strike me as roughly analogous to saying:

"hiding rows users don't have permission to see (also known as
row-level security)"

IOW, it's just substituting a generic description for a term of art
that has a real meaning. I can't see that as an improvement. It
would be possible to compile routines without involving JIT; as I
mentioned earlier, you could compile them to bytecode. Or, you could
compile them to machine code at definition time as, I am told by
Thomas, some database products used to do, I think by generating C
code and then shelling out to "cc". Just-in-time compilation clearly
conveys that we are generating machine code at runtime; "compiling of
routines" does not.

Besides, if we did that, what are we going to call the GUCs?
compiling_of_routines_above_cost? compiling_of_routines_provider =
'llvmjit'? Ugh.

This technique has bene called JIT since about 1983, and is well-known
under that name, mostly because of Java. Calling it some made-up name
we create ourselves is not going to make this more comprehensible to
users.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company