Your Feedback on Needs / Use Cases for Alfresco WCM

[kvc]

As you know, Alfresco is working on adding robust, easy-to-use WCM services to the Alfresco platform.  And, as ever, we are keen to make certain we are (a) building the OOTB capabilities you as a development community require and (b) are providing all the necessary hooks you need to extend or customize our solution to fit your needs.

If you have not already taken a look at our wiki, we outline the different components we are focused on, with particular attention to detail around our updated versioning model and the production model (dev –> preview –> staging) associated with it.  For those who have not seen the wiki, you can find it here:

http://wiki.alfresco.com/wiki/New_Web_Content_Management_Plan

To open this forum, we would like to open with an open question:  what are YOUR key priorities for WCM?  What services are essential for deploying in your organizations, and specifically what types of websites would you look to be using for?  We would appreciate any feedback on needs, use cases, and examples … we want to ensure that we are building you the right product.  Let's get the discussion rolling, and if you have any specific feedback on what's already on the wiki, fire away!

[kvc]

You completely grok where we are going with WCM!

First, for de-coupled delivery, we will support numerous options:

1.  Deploying completely statically managed or generated sites to a naked FS
2.  Deploying a dynamic site by deploying application resources (JSPs) to the
     file-system and content to a run-time Alfresco repository
3.  Deploying any web app or script-driven site to a native file-system, with
     option to customize deployment of content to any sort of non-Alfresco
     repository

Our goal is to build a full-fledged content distribtion product.  This will handle export and import of content from Alfresco server instance to Alfresco server instance, to file-systems, etc.  When you create a website you will register a deployment model for the site, and the model can support deployment of a site to one or more target servers (if you recall, our engineers were part of the original core Interwoven team, so we are familiar with complex deployment models).

For DreamWeaver-based sites, we will be adding support for the CIFs interface.  Goal is to drag-and-drop existing sites to get benefits of workflow and version control, and over time migrate to structured content publishing.

I'll take as a requirement adding details to the wiki on deployment models.  Basically, where you are expecting us to go is bang dead-on.

Kevin

[dking]

I'll take as a requirement adding details to the wiki on deployment models. 
Kevin

That would be great. Hopefully it will make clear how URLs are defined and made search engine-friendly.

Also important for understanding how to use the WCM will be documentation on how to write the XSLTs and JSPs. Mainly, how do I get a reference to the information architecture/taxonomy so I can build navigation menus and breadcrumbs? (Again I hope I'm not missing something in the Wiki/forums, where I did look. Have I mentioned how important good documentation is? Presumably the supported versions will include professional docs; I'm willing to pay for this.)

BTW, I re-read the thread here and realized (as you clearly stated) that the browse feature was envisioned for December but not necessarily as part of the WYSIWYG editor. To me, WCM is to a large extent about enabling non-technical users. If they have to understand how to copy and paste image dimensions from the properties of the image they have to right-click to get to and then copy the URL (without introducing a specific server domain which then breaks the sandbox concept, and which introduces links that the system can't update if a file name is changed) then you're missing an important user group. Every product I've seen has this, and the better ones enable search; Ingeniux's is cool, with autosuggestions for files matching the name pattern as you type it. I know you get this, but want to underline it because I probably can't use the product without this. Internal reference integrity and ease of use is not optional.

[rrahman]

WCM–excellent! I'm representing federal agency interest in enterprise web content management and am looking forward to share our requirements and discussions on web contenet management systems.

Thanks, I'm very excited about this addition to alfresco!

[kvc]

Also important for understanding how to use the WCM will be documentation on how to write the XSLTs and JSPs. Mainly, how do I get a reference to the information architecture/taxonomy so I can build navigation menus and breadcrumbs? (Again I hope I'm not missing something in the Wiki/forums, where I did look. Have I mentioned how important good documentation is? Presumably the supported versions will include professional docs; I'm willing to pay for this.)

Yes, got it.  This is actual already in progress for the GA release.

BTW, I re-read the thread here and realized (as you clearly stated) that the browse feature was envisioned for December but not necessarily as part of the WYSIWYG editor. To me, WCM is to a large extent about enabling non-technical users. If they have to understand how to copy and paste image dimensions from the properties of the image they have to right-click to get to and then copy the URL (without introducing a specific server domain which then breaks the sandbox concept, and which introduces links that the system can't update if a file name is changed) then you're missing an important user group. Every product I've seen has this, and the better ones enable search; Ingeniux's is cool, with autosuggestions for files matching the name pattern as you type it. I know you get this, but want to underline it because I probably can't use the product without this. Internal reference integrity and ease of use is not optional.

We've updated WCM-34 to include this.  We're getting this feedback from a number of sources.

Kevin

[dking]

Also important for understanding how to use the WCM will be documentation on how to write the XSLTs and JSPs. Mainly, how do I get a reference to the information architecture/taxonomy so I can build navigation menus and breadcrumbs? (Again I hope I'm not missing something in the Wiki/forums, where I did look. Have I mentioned how important good documentation is? Presumably the supported versions will include professional docs; I'm willing to pay for this.)

Yes, got it.  This is actual already in progress for the GA release.

Kevin

Great, I'll look forward to reading it and trying this out. Let us all know when a draft is available; I'm assuming you'll want feedback on this part of the functionality before GA, and I'd like to see what the approach is since I'd be spending many hours up to my elbows here. You see a lot of tangled stuff in some products when you get down to this level.

[dking]

Hi Kevin et al.,

One important aspect of Web site management is keeping language versions in sync and managing the links between content items that represent the same information in different languages. The Java web app frameworks I'm familiar with almost enforce good internationalization using properties files for messages and page text, but many of the WCM systems I've seen leave it to the site developer to decide what the data structures are and how localization will be managed (OpenCMS has a nice concept of siblings so that a content asset is aware of its translated equivalents). Which approach will you take: allow implementations to roll their own because deployments vary; or design an internationalization strategy to make this easier in at least some scenarios?

In my case, I plan to have one site with three languages available rather than having three different monolingual domains, and I'd likely benefit from a facilitated i18n/l10n strategy, or at least a discussion of suggested implementation options.

Thanks!

[rdanner]

I have been thinking about another kind of folder.  One you might call a Targeted folder, or Query based folder / Dynamic Folder.  As properties of the folder you would set a Query and a Template.  When the folder contents were traversed they would execute the query and render the results with the transform.

This would allow me to manage my document in one way but allow a certain group of users to view them as organized according to their needs.  For example If I kept URL information as metadata about documents I could use the query and the template to render a folder structure that looks like a docroot.  I can manage the content in a structure which is less temporal and allow users with managing for a specific channel to work in the context they are familiar with.

As Kevin has said, we've had the same notion.  It's a very powerful idea, but one that needs to be crafted carefully.  The most exciting form of dynamic folder is one that, as you say, executes a query on traversal.  The problem I see with that in the general case is performance.  A query can be arbitrarily complex and therefore can be expensive to process.  How would you feel about semi-dynamic folders to start with?  By semi-dynamic I mean a folder that is populated by a query, but only explicitly, by a 'refresh' operation.  This wouldn't preclude fully dynamic folders.  We could still provide that service, with the proviso of "be careful with your dynamic queries", but it might save headaches if we didn't have to. The scenario I worry about is someone setting up a fully-dynamic folder with a 6 way join that takes 30 seconds to execute, because you've got 50,000,000 documents in the system. Then the performance complaints come in to the Alfresco administrator.  It's always a delicate balance between bullet-proof and flexibility.

Please let me know what you think.

I dropped the ball on this.  I think this is absolutely a strong idea and in addition to that I think we might consider having the ability to "Materialize the folder" where the fodler turns into a folder containing the links to the documents that are part of the query result projection.

[marcus]

Another system I used to work with extensively had a feature very similar to the "query folder" idea, but it was a bit more general and called something like "Record Set Data Source". It had different flavours to it, which meant that as well as a search query, there were instances that could generate the list based on an SQL query, RSS consumption, SOAP call etc.

[kvc]

Great, I'll look forward to reading it and trying this out. Let us all know when a draft is available; I'm assuming you'll want feedback on this part of the functionality before GA, and I'd like to see what the approach is since I'd be spending many hours up to my elbows here. You see a lot of tangled stuff in some products when you get down to this level.

Will do.  Email me next Friday and we will arrange time with our developers to review our plan here.

Kevin

[smcclelland]

There are many ways to support this in the product but it was not clear to me "out of the box". A simple case is a product catalog or photo gallery. I would like to have a page that allows related pieces to be added at once.

HTML data describing the item
large catalog image
thumbnail image

this would allow a quick catalog or photo gallery to be constructed. I want a non tech user to be able to add, edit, or remove something like this in one screen. In the way that you have the +/- buttons for adding paragraphs I'd want users to add images.