the Design Experience Weblog

The OpenACS/.LRN Fall Conferece is scheduled for November 1-2, 2006 in Boston, Massachusetts. see the wiki page http://openacs.org/xowiki/en/Fall_Conference_2006

An addition day or two of hacking/bug-fixing activities are tetatively planned.

Some folks are putting together ideas to disucss about OpenACS and there is a list of people who are interested in attending

I am planning on talking about automated testing, and the benefits of adding testing to your design and development process, as well as discussing the future directions for OpenACS with anyone who is interested.

I am writing some code to generate PDFs using OpenReport (which calls ReportLab) and I needed a list of the built-in fonts that can be used. Here are the names of the Adobe Core Fonts:

  Courier

  Courier-Bold

  Courier-BoldOblique

  Courier-Oblique

  Helvetica

  Helvetica-Bold

  Helvetica-BoldOblique

  Helvetica-Oblique

  Symbol

  Times-Bold

  Times-BoldItalic

  Times-Italic

  Times-Roman

  ZapfDingbats>

Joel Aufrect has stepped down as release manager of OpenACS. Since I had recently read the release documentation and worked on helping Joel test and get OpenACS 5.2.x out the door, I volunteered to take over. Thanks alot to Joel for his years of hard work. The process he worked out and the documentation he wrote made it much easier to take over the volunteer position.

I plan to look for ways to make releases of OpenACS better quality and more frequent. I am starting a personal campaign to clean up the opeancs.org bugtracker.

OpenACS 5.2.0 has been released. This is a big release with many changes, mostly to the core data model to enable much easier handling of objects and tieing then into system-wide servies. It also incudes the tsearch2 driver for full text search with PostgreSQL. OpenACS 5.2.0 supports Oracle 8i and 9i and PostgreSQL 7.4 and 8.0.

OpenACS is a community focused web application toolkit. It's main focus is developing collaborative web applications. OpenACS provides a wide variety of collaboration tools including discussion forums, weblogs, news, file sharing, photo albums, faq builder, web based presentations, surveys, and more.

There has been a bunch of conversation regarding OpenACS (more on the OpenACS forums) and the differences between Ruby on Rails and OpenACS. It is very interesting and brings up some very good points on both sides.

Lars worked on OpenACS and came from the original Arsdigita Community System. Few people have more experience with OpenACS than Lars, so he has built more than a few web applications with OpenACS and its predecessor.

If the applications you are building are not very similar at a domain level, then reuse is very hard. That is, OpenACS is not a generic web platform that will work for every web application. Here is where some people diagree with Lars, that because many web applications are not appropriate for OpenACS, perhaps reuse of OpenACS components is not a good fit for web applications.

There is another side to this.Reuse can occur on a more domain specific level. If you are developing sites that are very similar, reuse makes more sense. The unit of reuse is really all that is different. That is, Rails is obviously a reusable framework, so it has common components that are needed for web applications. It sets the domain very narrowly and is successful, everyone would agree. If we just look at the OpenACS core, which is still huge, compared to Rails, it still has a limited set of features, and no applications per se, in the core. It also has a set of reusable components for web applications. It just sets a different domain, a different kind of web application that requires closer control of data, and more specific requirements on auditing and control of interaction with the application.

This is the idea that struck me the most. In a comment from Mark Aufflick:

Just to return us to the topic to finish off, letﾒs see what Glass has to say about whether reuse-in-the-large is at all feasable:

Mark posts a quote from Richard Glass

OK, so reuse-in-the-large is a difficult, if not intractable, problem. Is there any way in which we can increase the odds of making it work?

The answer is "yes." It may be nearly impossible to find components of consequence that can be reused across application domains, but within a domain, the picture improves dramatically

...

Software people speak of "families" of applications and "product lines" and "family-specific architectures." Those are the people who are realistic enough to believe that reuse-in-the-large, if it is ever to succeed, must be done in a collection of programs that attacks the same kinds of problems.

So if you are building a application for a certain domain, reusing it is possible, and even makes sense. This is where we think of OpenACS and systems built on it less as a framework, and more as a domain-specific language. Rails simiarly is definitely a domain-specific langauge. OpenACS and Rails just are targeting different domains.

I agree there are places in OpenACS where reuse is not the best plan, but for the projects I have worked on the core of OpenACS has helped me build an application that works well for the clients needs. And that is what it is all about, getting people the software they need.

Over at Creating Passionate Users is a great post on getting happy users by making the write thing easy and the wrong thing hard.

The most relevant part to me was "An API design which exposes the highest-level interface rather than a huge pile of lower-level calls (which could make it way too easy, for example, to call the right things in the wrong order), and whose methods/operations are named well! Half the reason our books sell so well is simply because some of the Java API designers used names that practically beg you to do the wrong thing."

I can say OpenACS APIs could use some work in this regard. I have been trying to create better high level APIs for many of the most important functions, but I think getting it right, and making it obvious the correct way to use the APIs including combining them to create software, is very important. APIs clearly need to be designed to be put together as building blocks, not in isolation. I'll have to think about this some more whenever I work on the OpenACS code.

Don Norman says google isn simple, it just hides complexity. I think this is not quite the whole truth. Norman goes on to say that MSN and Yahoo, for example, are simpler because they show you all your options on the home page. I seriously doubt that Norman recommends his clients show every possible option on their web site home pages.

He also says, to get to google news you have to click several times. When I want Google News, I visit http://news.google.com. No additional clicks required. When I want a map, I got to maps.google.com. When I want to check my email I go to gmail.com. There are seperate services, all provided by one company. Google is not trying to be a portal, although you can personlize the home page. This actually refutes Norman's entire premise, since he commends Yahoo for offering a customizable home page where you can get news etc on the main page.

Another way google helps is by "knowing" what you want. If you search for something, it gives you the right thing most of the time, without your having to know if its from the "news" site, or "maps". If you search for an address, the first search result goes to Google Maps. If I search for "news about hurricane katrina" the first result goes to google news. Seems to me that Google is simplfying things by giving me what I want, without my having to choose from hundreds of possible links on one page. This leads me to recommend a book, Don't Make Me Think, by Steve Krug.

Dave Tosh has a presentation about e-portfolios, Computing is Connecting. Sounds like this fits right into my idea to focus on connections, "the connections are the content".

Reading learing technology weblogs and just paying attention to current general weblog trends, networking, connections, social software, etc are very important topics. This all comes done to connections between this and that. So I am coining a new pharse "the connections are the content", after my previous "the community is the content". I guess connections are a generalized form of community in some graph theory sort of way.

Another way to look at this is how learning is about making connections and recognizing patterns. This is closely related to Connectivism (more), or a theory a learning by finding the meaning in connections between ideas, things, and information.

This is also how e-portfolios are starting to look, as a dynamic set of information linked, or connected, instead of a static list of accomplishments.

Lou Rosenfeld picks up on the Long Tail idea and applies to to search results. He notices that few searches are very common, and there are a large number of less common results. Most often an information artchitect or someone else will optimize the results for those most popular queries. Let's not forget about the rest of the queries. What is missing in those unanalyzed queries and results. It makes sense to get the big picture about what your users are searching for. He says perhaps studying the search queries long tail would help use find hidden users and potential market segements, ongoing stable information needs, as opposed to fads, and learn about rising trends by analyzing the long tail over a period of time.

Sounds like its a good time to do interesting work with search. Luckily I am developing the next (or first in some cases) generation search engine for OpenACS. Adding best bets, and other cool search features is a little way off, but having good tools to look at past search queries, and what sort of results users got when they tried to search seems pretty valuable. Obviously the value depends on what type of site you are running, but overall users needs to find what they are looking for no matter what kind of site it is.

I was reading the Polar Bear Book and just starting thinking about their various definitions of information architecture. I like the definitions about the goals or IA, that is, making it possible to find and use the information you need. This is somehow related to knowledge management, besides being a buzzword, of course.

IA stresses categorization, and metadata, usually of a more formal nature. And KM also has some of those aspects. I would like to see how to take more organic techniques for finding data such as Google, tags, and RSS and make them fit into the concepts of IA. Knowledge management seems to have a little more thought in this direction, in the area of Personal KM, usually related to weblogs, and more now with e-portfolio type applications.

I searched for "information architecure knowledge management" and found a course of study with the same name. It seems to combine library science, computer science, visual design, and communications. I'll have to explore that some more. I like the idea of combining visual design with the other subjects. This reminds me of Edward Tufte and the basics of designing information visually.

John Hiler at Microcontent News gives a great overview of the evolution of hierarchy vs. search for finding information. I have been thinking about this since I read The Inmates are Running the Asylum.

Well, really I have been thinking about it since I have had to organize large amounts of email. When I started filing things in folders, it was a pain to find something. Once I saw gmail, I adapted that model to my IMAP email with mutt. Now I have everything in my inbox, over 3000 items now, and search. It is still working fine. Maybe its not the best solution but it works for me.

Adapting these ideas to content management, organziation, and findability of web site data, I have been using search, combined with categorization, and other metadata filtering of search results. I think this is a good way to go, but the trick is figuring out what metadata is useful, and how to present it in the user interface.

Dan Brown writes about the underlying model behind CMS products, taking into the the concepts of workflows, roles, etc. He says the models are an Idealized Cognitive Model. The model resembles a factory. We can do better if we model a CMS to work like real people do. This is a great opportunity for Open Source content management to fill a need that real people have. Instead of modeling the content authoring and editing process around the coporate hierarchy, model it around the real process of real people writing. This is obviously why weblogs, and the models they are based on, are so popular.

I have been thinking about this in much simpler terms in the back of my mind for a while now. The first assumption most CMSs make is modeling the site in a hierarchy of folders. This works for people who understand hierarchal filesystems, which if you read "The Inmates are Running the Asylum" is not everyone. How can we model the structure and organization of content, and the authoring process, around how people think about their web site. I have a feeling its a combination of tags and some yet to be discovered organic metadata that doesn't look like metadata to the content authors.

I was over on the #openacs irc channel, answering questions about how OpenACS differs from some random open source CMS. It was a typical day. I said something like "OpenACS focuses on user interactions and community first, and content second unlike more open source CMSs which focus on content first." Then I had an ispiration. The community is the content. A nice short explantion, if still a bit last year's buzzword sounding. I think online communitiy was in style as a buzzword a year or 3 ago.

OpenACS and is predecessors where pioneers in online community toolkits and OpenACS is still the online toolkit that builds everything from the model of users, groups, and how they interact.

I did a google search for the phrase "the community is the content" and found some interesting results.

23 results. (wow looking now, my entry about the phrase is #2 in the results since I posted it yesterday.) The first result is a PDF flyer from MIT Press that mentions MIT CogNET which is an online system that was built on the ArsDigita community system which turned into OpenACS.

Next I searched for "MIT CogNet Arsdigita" and found a great gallery of sites built on ACS from the past. OpenACS should revive this type of feature and showcase current OpenACS based sites and how they use the unique features of OpenACS to solve interesting problems.

Gregor posts the readability of serveral CMS web site home pages. Looks like some are a little trickier to read than others. I ran OpenACS through and got a Flesch Reading Ease score of 67.30, which is approximately 5th grade reading level. I think its a good compromise for a web site that targets business decision makers and hackers all in one place.

Over at TheFeature.com I found a nice ratings interface with two options. Plus or minus, thumbs-up, or thumbs-down, whatever you call it. It adds a yes or no vote and calculates the rating from all of them.

Philip Greenspun talks about using SQL to check how long a text column you'd need to store the average discussion forum posting. That's neat, but his most interesting comment is that it could be a problem if your database doesn't handle large text objects. He is most likely referring to Oracle's worst-in-class handling of large text over 4k in CLOBs. CLOBs in Oracle require all kinds of special handling. That is why I recommend PostgreSQL for all but the most demanding applications. I haven't yet found a web application that overreached PostgreSQL's capabilities. The shining light in PostgreSQL is the "text" datatype. It has brains. Using a varchar of the appropriate size if that is appropriate and storing a compressed text object for larger text, all transparently. It "just works."

"f you want to do something that's going to change the world, build software that people want to use instead of software that managers want to buy." - Jamie Zawinski

Today I read two interesting takes on using tags for categorization and places where they might need to be enhanced or might not work.

Folksonomies: How we can improve the tags from Lars Pind and The Obligatory Folksonomy Post from Michael Feldstein.

I need to dig into these and come up with something useful to say. Right now I'd say that they both mention looking at the resource first, and finding out how different people have tagged it. That is the opposite of finding what other resources have the same tags, and could produce another set of useful metadata.

A year-end summary of OpenACS accomplishments was posted on the openacs.org forums:

* We released 9 versions of OpenACS, beginning with the first release of OpenACS 5.0, all the way to OpenACS 5.1.4

* Major changes and improvements to the toolkit, including:

o internationalization
o automated testing
o rapid development of tables (list-builder)
o external authentication
o support for Oracle 9i
o improvements to automated testing

* We've put in place (well, this is largely Joel's doing, I believe) a rigorous new release process for packages and for core. It took a while for it to start being used, but not packages are being released fairly frequently, and communities are starting to grow around particular packages.

* The installation process isn't much better than it used to be, but people are starting to work on it. The documentation is also very high quality, much better than it used to be.

* Registrations for the site are WAY up, better than they've been at any time since the inception of the project. I think we're finally starting to market ourselves better.

* We finally got the website onto OpenACS 5.x, thanks to Joel and Malte.

A little late, but OpenACS 5.1.4 was released on January 9, 2005. It includes:

• Bug fixes.
• The missing CR TCL API has been filled in, thanks to Rocael and his team and Dave Bauer.
• This release does not include new translations.
• See the Release notes for a full list of changes.

Download OpenACS 5.1.4 or upgrade via the repository.

I developed a quick wiki for OpenACS based on the content repository using the formatting code borrowed from Wikit. Using the new Tcl API for the content repository it was straightforward to write the forms and code to add and edit the pages. Hopefully this will make it easy for folks to add simple pages to their OpenACS based web sites.

Robin Good explains Why RSS Search Feeds Based On Web Searches Are Important, partially in resposne to Stephen Cohen's Why RSS Feeds For Search Engine Results Are Useless.

Stephen says that since the top 10 results for a particular query don't change very often RSS is not useful for results from a search engine such as Google or MSN. I don't think you only have to look at the 10 ten results of course. There are probably some interesting items in the top 50 or so. I don't see any reason you can't ask for the second or third page of results as RSS just as well as the first.

This is probably more true if what you are searching for is covered by regular news media, or weblogs, or basically any web site that has regularly updated content. You can't dismiss search based on the fact that a basic query will return fairly static results. You can try an advanced query. If you restrict to pages that have changed in the last number of days or use other criteria to narrow the results I suspect RSS would be more effective.

Robin seems to focus more on the fact that somewhere in Stephen's post he mentioned new sites such as news.google.com. He mentions that there are many valuable sources besides news outlets, and this is true. He does also address the fact that simple queries will not return different results, and of course, you will need a more complex query to get valuable results.

This reminds me of the work I did for a client recently where the entire web site emitted RSS to share data with other affiliated web sites. The RSS feed generation was built on an advanced search concept. Full text indexes of content was one search criteria, but in addition, language, date, and a categorization system were also used to generate RSS feeds.

Why not return search results as RSS? Its as good as any other XML format. And who knows what interesting uses for the data can be imagined once its in a easily machine readable format.

Over at web-graphics.com is another post about user defined tags. Andreas asks, now that we have these tags, how do we associate them with the content. There hasn't been a push to semantically refer to the tags in the HTML of the pages where the tags are used. Usually they just show up as a list of links to categories. It will be interesting how this is solved. I suspect some variant of Dublin Core will become popular similar to the way it is used for defining trackback data for weblog entries.

Louis Rosenfeld suggests that user supplied tags, popularly mentioned as "folksonomies" lately, would be a good complement to controlled vocabularies. This makes sense. User supplied tags would be useful to figure out how real people think about the categorization of their content. This feedback from users could drive the definition of the controlled vocabulary.

One thing he says is that right now there are 600 photos at flickr tagged with the word "summer" and that after a while there will be 6000, or 60,000 and it will be difficult to pull out the ones that are relevant. Why just use the one word? What if you searched on "summer beach" or some combination of terms? This is how I use del.icio.us to manage links I want to remember.

I am not sure if I use the tags at del.icio.us in the same way as everyone else. I don't make up new tags by combining words, but just put in 3 or 4 words I think will help me find the items in the future. This is some sort of wacky variation of facted classification I think. Where I take out the most important aspects of a item and pick the word that defines that aspect. This means if you look at the list of tags I have, I have a huge number of unique tags, some with only 1 or 2 items in them. But if I search for an intersection of tags I can narrow it down quickly.

Anything that gets people to think about metadata is good. Maybe once you manually tag a bunch of items, the computer could examine those, and compare them to new items and suggest potential tags for the new items.

Tim Bray's series On Search is a reasonably concise collection of the most important concepts in web full text searching. I recently rediscovered it, and felt it was time to read the whole thing.

One particular entry that was relevant to my current thinking was about metadata and search. He mentions that collecting metadata from humans is difficult and if you want to try, don't try to collect too much. I think this is where flickr and del.icio.us are very smart. The freeform tag approach, when used on large enough scale, does seem to allow for easy, useful metadata, particularly on subject on web pages.

It also mentions automatic categorization, and designing an API for managing search related metadata. All areas I plan to work on for OpenACS.

Back to reading for now...

One of the things always in the back of my mind is search, categorization and discovery of information for web sites. Lately I have been focusing on full text indexing for OpenACS based web sites using the tsearch2 module for PostgreSQL.

tsearch2 works reasonably well and is configurable with different dictionaries that you can build yourself based on ispell word lists and snowball stemmers. It has a parser built into the database and so far has made it easy to integrate full text search into various SQL queries.

One additional way folks seem to like to organize content is categorization. Generally this takes the form of a dropdown or multiple select widget when you can pick from a predefined flat list, or hierachy of categories. An additional kind of categorization based on free text "tag's has been recently popularized by sites suck as de.icio.us and flicr. I think taking both of these ideas, and full text indexing the category names and synonyms, and full text indexing the free form tags would make a neat combination in addition to full text indexing of the "content" of a particular page or item.

The art would be in weighting the category names, tags, and content to return the most relevant results. Of course, it depends on who is doing the categorization. I think it would be a fun experiement to try out categorization vs. free form tags vs. full text indexing on the OpenACS discussion forums database.

One advanced idea might be to extract the most common tags from the forum posts. That is, once a sutiable library of the most popular tags is established, it might be a good idea to apply that list to the text of a forum posting based on the words in the posting. A domian specific idea would be to prepare a list of OpenACS package names and tcl procedures and to tag those as well. This way, a visitor to openacs.org could go to the online API documentation and find a tcl procedure or package name and then pull up the list of forums postings that refer to those packages or procedures.

All this is possible if I ever have a free minute to play with these ideas.

OpenACS 5.1.2 has been released. Mainly this is a maintenance release. For full details see the changelog. You can also download it.

This post on making a better open source CMS is making the rounds. I responded to the points on the OpenACS forums and cross posted it here.

"Make it easy to install" - Ok, we probably lose more than any other CMS package on this one. If we can get RPM/DEB/etc packages this could work. We do have a web based installer, although somehow the config.tcl needs to be edited to reflect the few settings required, but a pakcaged installer could use a default for ip address and database name and get something running quickly.

"Make it easy to get started" Here Openacs and all the existing CMS like packages fall down. There all special tasks that need to be done, I guess a preconfigured CMS install.xml would be the thing here.

"Write task-based documentation first" Good idea. Figure out what someone using a CMS might want to do, and then write the code to let them do it.

"Separate the administration of the CMS from the editing and managing of content" This is also a good idea, and I think most of the content focused packages that exist do provide the content management interface and allow content editors to not also be site wide admins.

"Users of a public web site should never -- never -- be presented with a way to log into the CMS" This assumes that the contnet management system has no comment or other interactive features. Sorry, if you have a totally non-interactive web site, don't use OpenACS.

"Stop it with the jargon already" Ok, this is a good criticism. A good UI would make sense to the kind of user who is a content editor, not a web nerd.

"Why do you insist Web sites have "columns"" Well. None of the existing content apps for OpenACS has any sort or restriction on how to setup templates for content. We win on this count because all the templates needs to be coded, most of the built-in ones just have one "slot" for the main page content, the user can add anything they want anywhere they want.

So, it looks like OpenACS is a very flexible framework to address these user interface issues in a content management system, and they will be good ideas to keep in mind for anyone working on such an application.

Mayomi is an online mind-map creation tool. It also is a community that allows users to share their maps with others.

Quite interesting. This is sort of what I was talking about, but it still uses the mind map style tool for creation. I wonder if there would be a way to import linked data into it.

It would be interesting to find a tool that could show and let you browse a web site through a mind-map style interface such as that in FreeMind. Freemind could probably even do it. I think I'd prefer a browser based interface, but then you might need Flash or something to render it. Maybe it already exists.

My main motivation for thinking up this idea, which I am sure is not original, is that I really don't like the Freemind interface for entering data. I am more learning towards a more Wiki style for entering the kind of linked data used in an app like Freemind. I think the visualization possiblities are very interesting with that type of interface.

I'll have to do some research to see what already exists.

I mentioned mirroring a Wiki to a PDA using Plucker, but that really isn't enough. It is only one-way. A read-only Wiki is good for reference, but to capture thoughts and ideas you need to be able to write as well as read.

Importing some other Wiki intoWikiPad or one of the other Palm OS wiki's might be a solution.

Points more and more to an internet-ready PDA.

As soon as I posted the previous entry I though of some more. The trick is knowing what to index, and what is important in something such as an email. I suspect a machine might have to learn this, or at least that would be nice. Any time I have to specify what to index or how, is a chance I could screw it up.

I will have to go out and find some research on this topic, I am sure it is out there.

Instiki is a wiki built on Ruby by the developer behind Basecamp. I tried to install on debian and it was not quite as easy as the author claimed.

Here is what I had to do.

apt-get install ruby
apt-get install libwebrick-ruby
apt-get install libstrscan-ruby
apt-get install rdoc
apt-get install libzlib-ruby

Download instiki http://rubyforge.org/frs/?group_id=186

You also need to edit instiki.rb to point to /usr/bin/ruby or run it as "ruby instiki.rb"

Then it works. Apparently if you install ruby from source it comes with all those goodies.

More later after I actually try it.

I will just ask though, why has noone figured out that it would be nice to synch a wiki to a PDA?

I guess the answer is an internet enabled PDA.

I had a crazy idea browsing the web site for VoodooPad. VoodooPad is a neat little app that basically gives you a desktop Wiki with all the Max OS X coolness built it.

Well, I don't have a Mac, otherwise I'd probably just buy it. Instead I thought it might be fun to build a little desktop app using Kupu as the editor. That way I don't have to write one. It would be an interesting experiment. In addition a neat feature of VoodooPad is an API to edit Wikis on the internet. If your Wiki supports the API, you can edit it remotely with VoodooPad. This would be interesting if you could run Kupu locally as an editor for a remote web site.

Over at CMSWatch I found and articleWeb Services and Conent Mangement by Travis Wissink. This is a good, pretty detailed description on how web services might be used with a CMS.

He mentions that some CMS packages offer web services, and wonders if it would be cool for a CMS to consume web services. Taking this to a logical conclusion you could build a CMS from a variety of external parts. This is the small pieces, loosely joined or Frankenstein CMS concept.

I think this would be really great for a smaller scale web site, and for back-end operations for building a site. It doesn't really scale for content delivery, but I guess that is not the point.

OpenACS has a "service contract" concept that was originally designed to support WSDL definition of services. It would be interesting to see what sort of parts could be connected together to build an Open Source CMS from various packages.

A good example of a web application API is Flickr. They inlcude an authentication API, so that external services can log a user into a Flickr account and access their contacts and profiles.

This is definitely an intersting space to watch.

So far, client side content management can offer some features and ease of installation that web based systems can't. I have seen Macromedia Contribute, and it has a nice feature where after editing a page, you can send it off to someone to have it reviewed. Now, there isn't any way client side to integrate with some arbitraty CMS on the other end to see who could approve an article.

I am imagining a way to do with with WebDAV. Finding out who can perform actions on a page, and changing who can do that from the client might be a good fit for WebDAV ACL.

Along with other the rest of the WebDAV stardards this could be a great way to build a rich client-side content management interface.

davfs gives filesystem support for WebDAV shares to linux. So far it seems to be working fine.

Exploring the Mnemonic user interface(pdf) looks into the future of computer user interfaces.

I just had the idea of adding a caching feature to the OpenACS templating <include> tag. This would probably require support on the including script, and the included script. Includes in OpenACS can take arguments, so it would make sense, I think to cache on those values. In addition an includable page fragment would probably need a way to advertise whether or not it was cachable at all.

Then again, maybe its best to just cache at the database level. That is generally going to be the greatest bottleneck in an OpenACS system. The code that renders the template usually runs very quickly, and the round trip to the database is almost always the right place to improve peroformance.

One thing that would make sense is that caching decision could be made at the page level. So if one page were slow, and it had a number of included parts, it could just be cached by changing the template without diving into the underlying code.

To make it work it would be nice to finally allow a page contract for includable elements that is similar to ad_page_contract.

After writing that, I realized that the include caching is very similar to the partial rendering idea that the content repository and old aD-CMS supported. That is, a datasource could be combined with a template and written to the filesystem. Maybe that is another way to explore caching of includable content the old-fashioned way.

Tom Hoffam thinks about scaling a weblog installation across an entire school. He asks if traditional weblog software can scale to provide a weblog for every class, or even every student in a large school.

I think this is where something like .LRN really can fill a gap. It is designed for managing course and school wide collaboration for universities. It is built for thousands of users and hundreds of course. .LRN is by a high performance web application server and RDBMS and provides fine-grained permissions, security, and privacy controls. Centralized install of software is not necessarily a bad thing. The central install can provide all the tools students need to collaborate.

I have been thinking about content management for the web for a few years now. I started by looking into the CMS package developed by ArsDigita way back. It originally was a standalone system, but later parts of it were merged with ACS 4. Eventually all that became OpenACS 4. Unfortunately the concepts of the CMS package, Content Repository, and the rest of OpenACS were never reconciled, and to this day, there isn't a comprehensive content management solution bundled with OpenACS.

I want to focus on the fundamental needs of content management listed by Karl Goldstein, author of the original CMS from ACS/OpenACS 4.

Always keep in mind the core function of a web CMS from the user's perspective: the ability to easily create, edit, organize and manage and publish chunks of HTML text, where in most cases each chunk equals a page.

That said, authoring content is really the heart of the user experence for a CMS. And the coolest CMS editing experience is not the one that gives authors complete ability to go wild with formatting; it gives them the flexibility they need to present their content while constraining them enough to ensure consistency and reusability across a site.

A CMS needs to make is easy to write to the web. This needs to be kept in mind all through development. This is where OpenACS usually falls down. OpenACS makes it great to design a robust data-model, with complex permissions, workflow, and a powerful templating system. All the tools are there. They just need to be presented to the user in a way that makes sense.

The coolest editing experience also gives the authors to ability to insert and configure reusable dynamic components into their pages

This is where it starts to get tricky. It is easy enough in any web toolkit to add in dynamic parts to a page. Figuring out how to present this to the user is a challange. The editing experience needs to make it simple to add these elements to a page. I think, mostly, it is best to add the dynamic elements to a template, into which the actual content is then placed. This is good for navigational elements, links to "related items" and that sort of think, but it leaves out the ability to add images. Of course and img tag can be supported, but users really want to be able to look at a bunch of images and then pick one to insert into their content.


The other extended features Karl mentions are: a tool to configure navigation, link checking, themes to change the colors/look without editing template code, and asset management, which I just mentioned. Managing image, audio, video, and other resources is important for most users.

I think a major part of the user interface solution is Kupu, the new OSCOM backed project for a rich editing user interface that works across Open Source CMS platforms. It offers structured editing of content, and will include the ability to select and add images and links to content. I am excited about the development of Kupu, and I believe it is the best option to offer a great experience for content management in OpenACS and other Open Source platforms. Now I just need to make it work.