Re: trailers: --only-trailers normalizes URLs to trailers
- From
Jeff King <peff@peff.net>
- Date
- Jun 11, 2026, 06:56 UTC
- Message-ID
- <20260611065616.GE2191159@coredump.intra.peff.net>
- In-Reply-To
- <d8f7f827-27da-41fc-af8d-72d383b24fff@app.fastmail.com>
On Wed, Jun 10, 2026 at 04:21:29PM +0200, Kristoffer Haugsbakk wrote:
Show 10 quoted lines
> > Yeah, though if you'll allow me to nitpick your subject a moment: I > > don't think --only-trailers is really the culprit here. It demonstrates > > the problem because it normalizes the "trailer" it found. But the loose > > trailer matching is the more fundamental issue. For example: > > > >[snip] > > Yeah, this is more precise. I focused a ton on the normalized output > because that’s what makes it obvious. But the fundamental problem is > interpreting URLs like trailers.
That makes sense. As the author of --only-trailers I immediately wondered if I had introduced a bug in it, so I was partially motivated by exonerating myself. ;) I agree that using it is the simplest way to demonstrate the problem.
Show 14 quoted lines
> > Agreed, though I think a rule like: ":// (with no whitespace)" is not a > > valid separator. Something like this: > > Yes, matching on `://` strictly is a better proposal. No need to care > about `http`, `https`, `file`, etc. And both of these would *still* have > to be true for this change to be a false negative w.r.t. the user’s > intentions: > > • They really input a trailer that looks like a URL, but it’s not meant > to be a URL > • They really wanted the value to start with `//` > > And again I don’t think that is likely to ever happen (with a knock > on wood).
I didn't spend much effort on the patch I showed beyond running it once. It would probably need tests and a doc update. I wasn't planning to run with it, but if you feel like doing so, please feel free to use it as you like.
-Peff