correct XML escaping of illegal characters in literals

[barmintor]

Ampersands in XML content need to be escaped, and in attributes where appropriate.

[gkellogg]

Thanks, but note that there seems to be a problem with the pure- version running REXML.

[barmintor]

In transit, will follow up ASAP
On Feb 24, 2015 6:50 PM, "Gregg Kellogg" notifications@github.com wrote:

Thanks, but note that there seems to be a problem with the pure- version
running REXML.

—
Reply to this email directly or view it on GitHub
#34 (comment).

[barmintor]

Do REXML and NG have different deserializations for attributes? That's
awkward.

On Tue, Feb 24, 2015 at 6:58 PM, Benjamin Armintor armintor@gmail.com
wrote:

In transit, will follow up ASAP
On Feb 24, 2015 6:50 PM, "Gregg Kellogg" notifications@github.com wrote:

Thanks, but note that there seems to be a problem with the pure- version
running REXML.

—
Reply to this email directly or view it on GitHub
#34 (comment).

[gkellogg]

You probably hit an untested corner case. The reader uses an abstraction layer for the different libraries. I can try to look at it tomorrow, if you like.

[barmintor]

It would be great if you had a chance, will check in again tomorrow to see what I can do. We'll work from a branch in the meantime (the attribute case doesn't apply to us)

[gkellogg]

It seems that REXML XPath varies slightly for returning the attribute content. Nokogiri unescapes the result, REXML doesn't.

I separated the tests into with- and without-attribute use; fixed a bug where the triple was being output both as an attribute and an element, and restored the original simple example it was based on.

Thanks for finding and working on this!

[barmintor]

Thanks for the library!

On Wed, Feb 25, 2015 at 2:13 AM, Gregg Kellogg notifications@github.com
wrote:

It seems that REXML XPath varies slightly for returning the attribute
content. Nokogiri unescapes the result, REXML doesn't.

I separated the tests into with- and without-attribute use; fixed a bug
where the triple was being output both as an attribute and an element, and
restored the original simple example it was based on.

Thanks for finding and working on this!

—
Reply to this email directly or view it on GitHub
#34 (comment).