Talk:URL parser

From Rosetta Code

LDAP URL non-conformant[edit]

Great task! The example URLs provide good coverage, but the example ldap://[2001:db8::7]/c=GB?objectClass=one&objectClass=two is invalid per RFC2255. For solutions exercising library code that knows about more URL structures than HTTP, this is distracting. I suggest replacing it with the example in RFC3986: ldap://[2001:db8::7]/c=GB?objectClass?one which is just as parseable under HTTP rules, but won't blow up a parser that understands the ldap scheme.

--Aspectcl (talk) 03:59, 24 July 2015 (UTC)

That matches my reading of rfc2255 also. I'd say go for it (and mark the existing implementations with a task description updated tag). --Rdm (talk) 14:36, 26 July 2015 (UTC)

hostname for arbitrary schemes[edit]

The introduction claims that arbitrary URI schemes (eg. foo://) should parse with full hostname, etc. However, User:Choroba claimed that "you don't get the host from the foo:// scheme, as host is only valid for schemes that define it." (Perl section intro). One of these is incorrect, and should be amended. --Sondra.kinsey (talk) 16:00, 21 July 2016 (UTC)

Yes, appendix A of rfc 3986 defines a different syntax than the simplified statement at the beginning of this task. This is a flaw in this task's description, and someone should replace the incorrect statements in the task description with a rendition of that appendix A (and after that any task implementations which don't do it right should be fixed, or marked as incorrect). --Rdm (talk) 13:20, 22 July 2016 (UTC)