Re: cvs: php-src /ext/simplexml simplexml.c /ext/simplexml/tests 025.phpt
| From: | Marcus Boerger | Date: | Tue, 01 Nov 2005 11:57:51 +0000 |
| Subject: | Re: cvs: php-src /ext/simplexml simplexml.c /ext/simplexml/tests 025.phpt | ||
| References: | 1 2 3 4 | Groups: | php.cvs |
| Request: | Send a blank email to php-cvs+get-34935@lists.php.net to get a copy of this message | ||
Hello Rob,
Tuesday, November 1, 2005, 12:59:02 PM, you wrote:
> Marcus Boerger wrote:
>>>When doing a recursive get, if the namespace is redefined further down
>>>in the subtree, the redefinition is what ends up in the array rather
>>>than the closest namespace to the element. Just to be clear this also
>>>only adds namespaces that are used, not those defined but not used
>>>(which is probably ok with how they are used in SimpleXML).
>>>
>>>
>>
>>Easy interface. If that leads to problems do it manually or use dom.
>>Wans't that our standard answer? :-)
>>
>>
> Works for me and also answers previous question on redefining default
> namespace.
I added code to add default namespaces as "" now, looks good, see the test.
>>>When calling getDocNamespaces recursively, if there are any
>>>redefinitions in the tree, what ends up in the array all depends upon
>>>the LAST node that get processed by the sxe_add_registered_namespaces
>>>function.
>>>- probably better to do this one manually and walk the elements in the
>>>tree checking the nsDef and adding them. Will be faster - one tree
>>>traversal rather than multiple times (it ascends the entire tree every
>>>time right now starting from the position of the node passed to
>>>xmlGetNsList - which is a killer when called recursively).
>>>
>>>
>>
>>Yes it is slow, i thought of the recursive useage only for debugging
>>purpose.
>>
>>
> Concern here wasnt as much about speed as you had mentioned it was only
> for debugging, but that a node buried in a document can alter the
> returned namespace array can make debugging a bit harder. I am going to
> change it to search manually which not only will increase speed, but the
> results will be consistent. Since the prefix is going to remain as the
> array key, imo it's probably best if the first encountered namespace in
> document order is used and any dupe prefix skipped.
Ah, skipping dups makes sense in the recursive mode.
marcus