Re: [PATCH 0/3] test-suite fixes for upcoming curl 8.18.0

From: Daniel Stenberg <daniel@haxx.se>
Date: Dec 18, 2025, 12:37 UTC
Message-ID: <613s97no-7021-pp15-79s4-302o39p7n5r8@unkk.fr>
In-Reply-To: <20251218121120.GA3252258@coredump.intra.peff.net>

On Thu, 18 Dec 2025, Jeff King wrote:

> Daniel: I'm cc-ing you in case you want to double-check that curl's behavior 
> changes are all OK before the release. I think it's mostly fine, though the 
> handling of tab versus space in the third patch is perhaps questionable.

Thanks Jeff, allow me to add my comments here on the three patches:

>  [1/3]: t5551: handle trailing slashes in expected cookies output

This is all benign. As you correctly observed, we no longer keep the "original" cookie path around and only work with the sanitized version - so that's the one stored now. It was already the one used for actual comparisons so apart from the change in storage, it *should* not cause any problems.

>  [2/3]: t5563: add missing end-of-line in HTTP header

I believe I made some code checks a little stricter: header lines MUST end with at CR or LF (or both) to be treated as a valid one. Your fix for this should be good also for older libcurl versions.

>  [3/3]: t5563: relax whitespace assumptions for unfolded headers

This one is material for me to rethink.

I had to completely change our header unfolding logic because we learned that we did not apply it early enough, so some header parsing was wrongly done on pre-unfolded data. In this process, I also changed the logic that appends the following line on the previous line. To avoid having to keep a state, I decided to just append the second line onto the first one without trying to reduce the whitespace characters to a single one.

I did not fully consider the impact this might have on users such as you. Allow me to rework that a little bit further and get the former white-space behavior back. Thanks!

-- 
  / daniel.haxx.se