Crashes if there are multiple query params

[whitebyte]

parseURL("http://example.com?test1=some&test2=more");

Crashes with the following error:
url.URLException@../../.dub/packages/urld/3.0.1/urld/source/url.d(31): domain name 'example.com?test1=some&test2=other' contains illegal character '&' at position 22

Interestingly, I see that the lib handles URLS like this in unit tests:
https://example.org/this?query=value&other=value2
but it will fail with a simler version
https://example.org?query=value&other=value2

[dhasenan]

Thank you for your report!

This is not a crash. It is throwing the documented exception when it fails to parse a URL. The error message is incorrect, however.

RFC 1738 only allows the '/' of a URL to be omitted when there are no query parameters. However, many parsers (Python3's urllib.parse, .NET's System.Uri, Java's java.net.URL, Ruby's URI) accept malformed URLs like this. Javascript's URL class does not.

Failing to parse this string is not incorrect, but it's not obvious whether it would be an improvement to parse it correctly.

[whitebyte]

I beg to differ. RFC 1738 indeed demands "/" (albeit indirectly), but it's mostly superseded by a more recent RFC 3986 that does not impose such restrictions.

Javascript's URL class does not.

But it does?

const url = new URL("https://example.org?query=value&other=value2");
console.log(url);

will work just fine and correctly define query string

[natalie2597-web]

I believe #19 is related