Hi,
On 2018-12-18 00:36:55 +0100, David Fetter wrote:
> On Mon, Dec 17, 2018 at 05:31:22PM -0500, Tom Lane wrote:
> > David Fetter <david@fetter.org> writes:
> > > Please find attached a run of a tool that looks for duplicated
> > > tokens. I've removed some things that seem like false positives,
> > > basically all from the stemmer part of the source, but there's
> > > still a lot.
> >
> > I thought you were talking about problems like "that that" typos,
> > but on looking at the file, what this is actually complaining about
> > is any duplicated code segments anywhere. I do not find this
> > helpful. Refactoring to the point that dozen-line code stanzas
> > never appear more than once would be incredibly invasive, likely
> > very bad for performance, and I don't think it'd improve readability
> > either.
>
> Is there a threshold, possibly much larger than a dozen lines, where
> such refactoring would actually make sense?
Sure, sometimes. But it seems unlikely that a report based on the
technology at hand here would be useful. If none of the code has changed
in recent times, it's not that usually worthy of a lot of attention to
adapt, unless we're talking much larger pieces of code. And a fair bit
of what's pointed out in that report is well known (e.g. that ecpg
duplicates a lot of code in a horrible manner). I think to be useful
it'd need to actually point out very similar code, that's starting to
diverge further - but that's obviously a much harder thing to achieve.
> > > - Would it make sense to make some kind of git commit trigger that
> > > at least warns when a new one has been introduced?
> >
> > Commit triggers are NOT the place for heuristics about code quality,
> > even if they're well-considered heuristics.
>
> Where's a better place for such heuristics? I'd like to think that
> there are opportunities for more automation than we currently have, on
> that score. Maybe a `make` target?
If we find useful targets, we can integrate them that way. I don't
consider this to be a worthy case though.
Greetings,
Andres Freund