Allen Gilliland wrote:
> 
> 
> Elias Torres wrote:
>> Allen Gilliland wrote:
>>> a few more little odds and ends ...
>>>
>>> 1. I'd prefer we not call the aggregate data "clouds".  I think we can
>>> find a better name that describes the data.
>>
>> +1 I have a bean called WeblogEntryTagAggData for aggregated data. I
>> think you are right, since I think that cloud is not the only/future use
>> of this table.
>>
>>> 2. Instead of updateTagSet() I think we should follow the getter/setter
>>> bean pattern, so the method should be setTagSet() or possibly setTags()
>>> because setTagSet() sounds weird?
>>
>> I'll move to setTags() and getTags() if possible  (most likely), but
>> setTags is used by hibernate to set the collection for the bean. I'm new
>> to hibernate so if you have work-arounds or I'm simply wrong to think
>> such a thing please let me know. That's why I'm using a different method
>> for "updating" the tags, because I can't completely replace a reference
>> to the internal set because hibernate throws up. Additionally, I don't
>> want to lose the first time you used a specific tag because all we are
>> currently sending to the client is the tag names.
> 
> When I ran through my proof of concept version of tagging a while back I
> used getTags() and setTags() without any problems.
> 
> Also, looking at it I had a setTagsAsString() and getTagsAsString()
> method which was useful for centralizing the task of bridging between
> the XXXTag object and tag strings from the UI like "foo bar etc".  May
> be useful.


Those methods are useful and I already have them in the code. However,
they are not in the pojo because I don't want to decide deep inside how
we split the tags: commas or spaces. Maybe make that pluggable.

> 
> 
>>
>>> 3. I think the parameter for the feed url should be "tags" instead of
>>> "tag" for consistency.
>>
>> I'll make a note of this.
>>
>>> 4. The PageModel is inherently constrained to a specific weblog, so
>>> rather than getWebsiteTags() I think the method can be just getTags().
>>> But regardless of the naming I am still only okay with having this
>>> method if it has some form of a constrained result set.
>>
>>
>> Sure thing.
>>
>>> 5. I'm not sure I'm onboard for the tag search form thing.  How is that
>>> supposed to be used?
>>
>> Dave added that I believe. My TagServlet is for the edit-ui so you can
>> have type-ahead of your own tags. I'm not really sure what Dave meant
>> from it or maybe he was trying to interpret my messy proposal.
> 
> I think we should talk a bit more about these last 2 items before you
> commit to an implementation.  I'd definitely like to hear more about the
> purpose of the tag search form thing.
> 
> About the type-ahead feature, I think that's an important feature and
> something we definitely want to do, but I'm wondering if we should
> consider a general architectural approach to ajax tools like this. Right
> now we already have the UserData servlet which is a one-off of a similar
> situation, and I'm sure we'll want to keep adding things like this, so
> maybe we should organize this work a bit more so that we don't end up
> with a huge assortment of one-off servlets for providing data to ajax
> widgets?
> 
> We could start by thinking about a consolidated url architecture where
> all of these things operate under some kind of RESTful service stemming
> from a single location.  maybe ...
> 
> /roller-ui/authoring/ajax/users (info about users)
> /roller-ui/authoring/ajax/tags (info about tags)
> etc, etc.
> 
> or perhaps we can look at multi-purposing Jeff's AAPP stuff?  the only
> danger there is that there are security implications since the AAPP
> stuff has the ability to modify data.
> 
> -- Allen
> 
> 

We need to have this discussion.

>>
>> -Elias
>>
>>> -- Allen
>>>
>>>
>>> Elias Torres wrote:
>>>> Thanks Dave! The page is much more readable now.
>>>>
>>>> I have updated the page with answers to most of your todo's. Regarding
>>>> my progress I now have edit-ui for tags, tags tab in frontpage, hot
>>>> tags
>>>> in sidebar. From a code perspective this means, pojos, weblogmanager,
>>>> sitemodel and struts together with .vm/.jsp changes are done.
>>>>
>>>> Things to do:
>>>>
>>>> - entries for specific tags (and their pagers)
>>>> - feeds
>>>> - type-ahead servlet
>>>> - parameterize the tags tab in frontpage
>>>> - .. and code thread to optimize the aggregation for large sites (i.e.
>>>> pre-compute tagclouds)
>>>>
>>>> Any more feedback on the proposal? Should I email asking for a vote?
>>>>
>>>> -Elias
>>>>
>>>> Dave wrote:
>>>>> Oops. Here's the link:
>>>>> <http://rollerweblogger.org/wiki/Wiki.jsp?page=Proposal_WeblogTags>
>>>>>
>>>>> - Dave
>>>>>
>>>>>
>>>>>
>>>>> On 9/18/06, Dave <[EMAIL PROTECTED]> wrote:
>>>>>> I review the tagging proposal and (with Elias' permission)
>>>>>> reformatted
>>>>>> it for readbililty. I also added some comments and a couple of TODO
>>>>>> and ISSUE markers to indicate places that need clarification.
>>>>>>
>>>>>> Please review and provide feedback. How close are we to agreement on
>>>>>> this proposal?
>>>>>>
>>>>>> - Dave
>>>>>>
>>>>>>
>>>>>>
>>>>>> On 9/18/06, Dave <[EMAIL PROTECTED]> wrote:
>>>>>>> On 9/15/06, Elias Torres <[EMAIL PROTECTED]> wrote:
>>>>>>>> I have begun writing some of the tagging support for Roller 3.1
>>>>>>>> and I
>>>>>>>> have a basic unit working. I've created WeblogEntryTagData and
>>>>>>>> added
>>>>>>>> methods to WeblogEntryData to support Tagging. I've also written 10
>>>>>>>> tests in WeblogEntryTest.java and of course added support in the
>>>>>> editing
>>>>>>>> UI to manage tags.
>>>>>>> Personally, I'd like to see a complete proposal before work
>>>>>>> begins --
>>>>>>> i.e. a list of the specific methods w/signatures that will be
>>>>>>> added to
>>>>>>> the managers and page models, a list of the macros that will be
>>>>>>> required, etc. Listing those page models and macros could reveal
>>>>>>> some
>>>>>>> data model deficiencies.
>>>>>>>
>>>>>>> That said, I'm not opposed to committing partial work as long as we
>>>>>>> have consensus on the proposal (I think we do) and there is a
>>>>>>> commitment to finish it.
>>>>>>>
>>>>>>> BUT... Please wait until I create a Roller 3.0 release branch
>>>>>>> from the
>>>>>>> trunk today (I just reviewed the changes since RC1 and the
>>>>>>> non-sandbox
>>>>>>> ones look good for 3.0).
>>>>>>>
>>>>>>> - Dave
>>>>>>>
> 

Reply via email to