Nokogiri::XML::Schema#validate behaves differently in MRI and JRuby #1282
Labels
Milestone
Comments
|
Thanks for reporting this. Someone will take a look! |
This was referenced May 7, 2015
zfletch
added a commit
to perseids-tools/epi-doc-validator-rb
that referenced
this issue
Mar 9, 2021
There is a bug in Nokogiri (sparklemotion/nokogiri#1282) that causes `Nokogiri::XML::RelaxNG#validate` in JRuby to add errors every time it's called. To fix this, create a new instance of `Nokogiri::XML::RelaxNG` every time. To lessen the impact on the library's performance, use `from_document` and cache the parsed XML.
|
Just a note that I've started a branch to clean up XML::Schema behavior and this is being worked on. |
|
👏 👏 Thank you @flavorjones. Better late than never! |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
When trying to re-use a
Nokogiri::XML::Schemaobject for validating multipleNokogiri::XMLdocuments, the MRI and JRuby implementations behave differently. The JRuby implementation reports errors encountered in previous validation, while the MRI version does not.Please see https://gist.github.com/ylansegal/eac88b27f5369b4f8ca7 for a recreation of the problem. It essentially validates a known good document, then a bad document and then a good document again against the same
Schemaobject. In MRI, the last validation does not report errors (which is what I would expect). In JRuby it reports the errors from the previously validated document.By the way, I encountered this error because the ruby-saml gem tries to memoize the schema to avoid reading from disk each time:
https://github.com/onelogin/ruby-saml/blob/master/lib/onelogin/ruby-saml/saml_message.rb#L26
The text was updated successfully, but these errors were encountered: