Re: WIP: a way forward on bootstrap data

On 3/25/18, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Well, as I said, I hadn't really reviewed the .dat files, but if that's
> what you're doing I'm going to request a change.  Project style is to
> fit in 80 columns as much as possible.  I do not see a reason to exempt
> the .dat files from that, especially not since it would presumably be a
> trivial change in rewrite_dat.pl to insert extra newlines between fields
> when needed.  (Obviously, if a field value is so wide it runs past 80
> columns on its own, it's not rewrite_dat.pl's charter to fix that.)

This feature is working now. I've attached a 100-line sample of all
the catalogs' files for viewing. Note, this is pretty raw output,
without the clean-up step from patch 0002. In the most of the original
DATA() lines, there was no spacing between entries except in some
cases to separate groups (often with a comment to describe the group).
My clean-up patch tried to make that more consistent. For this sample,
it would add blank lines before the comments in pg_amop, and remove
blank lines from the first few entries in pg_type. If you wanted to
opine on that before I rework that patch, I'd be grateful.

Also, these data entries have default values removed, but they don't
have human-readable OID macros. (I'll have to adjust that script to
the 80-column limit as well).

>> -Arrange for rewrite_dat.pl to run when perltidy does.
>
> What I was thinking we should have is a convenience target in
> include/Makefile to do this, say "make reformat-dat-files".
> I'm not that excited about bundling it into pgindent runs.

I've attached a draft patch for this. If it's okay, I'll incorporate
it into the series. I think reformat_dat_files.pl also works as a
better script name.

>> -Maybe document examples of how to do bulk-editing of data files?
>
> +1.  In the end, that's the reason we're doing all this work, so showing
> people how to benefit seems like a good thing.

It seems like with that, it'd be good to split off the data-format
section of the README into a new file, maybe README.data, which will
contain code snippets and some example scenarios. I'll include the
example pg_proc.prokind merger among those.

-John Naylor