psDecodeEntities
is no longer passed numeric character references (e.g.,  
, A
) and the predefined XML entities (&
, <
, etc). They are now handled by the parser. Both of these construct classes only have one spec-compliant interpretation and this behaviour must always be present, so it makes no sense to force user code to re-implement the parsing logic.0x
or 0X
like &0x20;
were accepted. This was not in compliance with the XML specification and it has been corrected.#
that are not character references are no longer allowed and will be parse errors.data-default
for data-default-class
, reducing the transitive dependency load. For most users, this will not be a breaking change, but it does mean that importing Text.XML.Conduit
will no longer bring various instances for Default
into scope. This will break code that relies on those instances and does not otherwise see them. To fix this, import Data.Default
from data-default
or one of the more specific instance-providing packages directly (e.g., data-default-dlist
for the DList
instance).rsXMLDeclaration
; setting it to False
omits the XML declaration.tag
function no longer throws an exception when attributes don't match #93many_
combinator to avoid building results in memory #94Consumer Event m a
to ConduitM Event o m a
to allow yielding valuestakeAllTreesContent
with takeAnyTreeContent
, that only consumes one treeNameMatcher
type to refactor tag parserstake*
functions to stream events rather than parse themignore*
functions to comply with naming conventionSupport monad-control 1.0