Add additional tests for borrowed input #509
Conversation
Codecov Report
@@ Coverage Diff @@
## master #509 +/- ##
==========================================
+ Coverage 61.33% 61.35% +0.01%
==========================================
Files 33 33
Lines 15393 15393
==========================================
+ Hits 9441 9444 +3
+ Misses 5952 5949 -3
Flags with carried forward coverage won't be shown. Click here to find out more.
Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here. |
| /// Deserialization of all XML's in that module expected to fail because | ||
| /// values requires unescaping that will lead to allocating an internal | ||
| /// buffer by deserializer, but the `Borrowed*` types couldn't take ownership | ||
| /// on it. |
There was a problem hiding this comment.
Is there any concept of opportunistic zero-copy with Cow<&str>? As someone not terribly familiar with serde.
There was a problem hiding this comment.
If I understand you correctly, Cow<str> will borrow whenever possible i. e. when deserializer can provide data of underlying source by calling Visitor::visit_borrowed_str. The opposite case when the deserializer allocates String by itself and calls Visitor::visit_string. The intermediate case when deserializer provides borrow to the data, that it owns, via Visitor::visit_str.
In our case we can either provide access to the XML source by calling Visitor::visit_borrowed_str, or allocate a new String (because we in any case need to allocate to do unescaping and/or decoding) and give it to the Visitor. If XML in UTF-8 and does not contain escaped data or mixed text/CDATA or CDATA/CDATA content (after solving #474), the data will be borrowed
Borrow input in attributes is working now, I suppose, after rewrite in #490. Now that is ensured by tests. Closes #328