On Tue, Oct 28, 2008 at 6:32 AM, Vincent Massol <[EMAIL PROTECTED]> wrote:
>
> On Oct 28, 2008, at 2:22 PM, Thomas Mortagne wrote:
>
>> On Tue, Oct 28, 2008 at 3:12 AM, Vincent Massol <[EMAIL PROTECTED]>
>> wrote:
>>>
>>> On Oct 28, 2008, at 11:07 AM, Sergiu Dumitriu wrote:
>>>
>>>> vmassol (SVN) wrote:
>>>>> Author: vmassol
>>>>> Date: 2008-10-28 09:19:23 +0100 (Tue, 28 Oct 2008)
>>>>> New Revision: 13864
>>>>>
>>>>> Added:
>>>>> platform/core/trunk/xwiki-rendering/src/test/java/org/xwiki/
>>>>> rendering/block/XMLBlockTest.java
>>>>> Modified:
>>>>> platform/core/trunk/xwiki-rendering/src/main/java/org/xwiki/
>>>>> rendering/block/XMLBlock.java
>>>>> Log:
>>>>> XWIKI-2785: The order in which XML attributes are printed when
>>>>> converted to XWiki syntax is not fixed
>>>>>
>>>>
>>>>> + public void testAttributeOrder()
>>>>> + {
>>>>> + Map<String, String> attributes = new HashMap<String,
>>>>> String>();
>>>>> + attributes.put("aaa", "value1");
>>>>> + attributes.put("bbb", "value2");
>>>>> + XMLBlock block = new XMLBlock("element", attributes);
>>>>> +
>>>>> + Iterator<String> it = attributes.keySet().iterator();
>>>>> + Iterator<String> newIt =
>>>>> block.getAttributes().keySet().iterator();
>>>>> + while (it.hasNext()) {
>>>>> + assertEquals(newIt.next(), it.next());
>>>>> + }
>>>>> + }
>>>>
>>>> I'm not sure this test is enough,
>>>
>>> it's definitely not enough.
>>>
>>>> since the order for two attributes
>>>> could happen to be the same. Why not add 10 attributes (aN = vN,
>>>> where N
>>>> goes from 0 to 9)?
>>>
>>> I've just tried this:
>>>
>>> public void testAttributeOrder()
>>> {
>>> Map<String, String> attributes = new HashMap<String,
>>> String>();
>>> for (int i = 0; i < 10; i++) {
>>> attributes.put("key" + i, "value" + i);
>>> }
>>> XMLBlock block = new XMLBlock("element", attributes);
>>>
>>> Iterator<String> it = attributes.keySet().iterator();
>>> Iterator<String> newIt =
>>> block.getAttributes().keySet().iterator();
>>> while (it.hasNext()) {
>>> assertEquals(newIt.next(), it.next());
>>> }
>>> }
>>>
>>> And the tests till passes even when I revert the implementation for
>>> XMLBlock (i.e. no usage of LinkedHashMap).
>>>
>>> Any more idea? Could it be that the test is wrong or the
>>> implementation not correct? (I don't se anything wrong).
>>
>> It's the implementation :
>> change
>>> this.attributes = attributes;
>> by
>>> this.attributes = new LinkedHashMap<String, String>(attributes);
>>
>> does not change anything, the most important is what is the initial
>> Map implementation of provided "attributes":
>> - if it's HashMap then there is nothing you can do here to have the
>> same order user set in the wiki syntax
>> - if it's already LinkedHashMap then you have nothing to do
>
> I'm not sure. I thought about this too. Here's my understanding: we
> should return the map keyset in the same order as provided by the
> user, i.e. if the user uses a HashMap then we return it in the same
> order as the HashMap. If the user uses a LinkedHashMap then we return
> it in that order too. I also think we should create a new map and not
> use the one provided by the user. I was also hesitating about adding a
> addAttribute() method to the class.
Yes it's better to create a new Map object by security (that way the
new Map will not be modified when modify provided attributes Map), I'm
just saying that just for the order of Map elements this does not
change anything.
>
> Note that I was not talking about end users but about users of XMLBlock.
>
> Last, I've also fixed XHTMLMacro locally so that it uses a
> LinkedHashMap when creating XMLBlock elements.
>
> Thanks
> -Vincent
> _______________________________________________
> devs mailing list
> [email protected]
> http://lists.xwiki.org/mailman/listinfo/devs
>
--
Thomas Mortagne
_______________________________________________
devs mailing list
[email protected]
http://lists.xwiki.org/mailman/listinfo/devs
