Re: Refactoring pgbench.c

On Sat, Jun 27, 2015 at 5:10 PM, Tatsuo Ishii <ishii(at)postgresql(dot)org> wrote:

> Main pgbench logic consists of single file pgbench.c which is 4036
> lines of code as of today. This is not a small number and I think it
> would be nice if it is divided into smaller files because it will make
> it easier to maintain, add or change features of pgbench. I will come
> up with an idea how to split pgbench.c later. In the mean time I
> attached a call graph of pgbench.c generated by egypt, which we could
> get a basic idea how to split and modularize pgbench.c.
>

I've never found the raw line count of a file to be a problem, at least not
at the 4000 line mark.

If some functions could be moved to existing libraries (like the functions
strtoint64 or doConnect or read_line_from_file or the statistical
distribution functions) that would be great, but I assume that would have
been done already if there was already a logical place to hold them. I
don't think inventing new libraries just to hold these would be an
improvement.

If you provided a suggested dissection, maybe I would find it compelling.
But without seeing a concrete proposal I think the process of refactoring
is going to impede other improvements more than the result of refactoring
it will promote them.

Cheers,

Jeff