@jku

Hey @ramonpetgrave64, thanks for the PR!

• I'd make the fix in RekorLog.entries() though: no need to rstrip() if we never add the slash in the first place. The other use cases where the url is used look like they will keep working just fine

Yeah, I think I'd prefer a "never add the slash" approach here versus "add it and then remove it sometimes" -- IMO that'll be easier to read and follow.

(In terms of doing that -- I think we can just modifying the underlying urljoin calls that build up these different routes.)

urljoin doesn't work like other "join" functions. Removing the slashes altogether would make everything more complicated when we need to add the slashes. If you both still prefer, I could do it the way you suggest.

• https://stackoverflow.com/questions/10893374/python-confusions-with-urljoin

urljoin doesn't work like other "join" functions. Removing the slashes altogether would make everything more complicated when we need to add the slashes. If you both still prefer, I could do it the way you suggest.

Ah yeah, I forgot about the wonky join behavior even when there's no leading or trailing /. Sigh.

Given that, I think maybe we'd be better off uisng f-strings here and building up the URL that way, to avoid having to strip things post facto. Alternatively, we could use the rfc3986 package (which we probably already indirectly depend on), since it has URIBuilder.add_path and URIBuilder.extend_path for exactly this purpose.

@woodruffw rfc3986 isn't currently a dependency, so I changed to using f-strings. Also added the changelog entry.

/gcbrun

/gcbrun