O'Reilly Interview with the Plone Founders

Alexander Limi writes "Just in time for some light weekend reading, O'Reilly's OSDir.com has published a byte-sized interview with the two founders of Plone. This is a nice follow-up to the earlier discussion on Slashdot, and covers a lot of the unanswered questions people directed to us earlier as the surprise winners of the O'Reilly COMDEX competition."

Comments

I am going to sue these people.

[Knight55]

This word sounds like Phone, I am representing my client thomas edison and taking a 100% commision for you trampling all over his IP. My dad from the SCO conned me into doing this.

Re:I am going to sue these people.

[martyn s]

Alexander Graham Bell.

Re:I am going to sue these people.

[paranoidsim]

I too read Plone as phone. Consequently, the headline to me read, "O'Reilly Interview with the Phone Flounders". I got a laugh.

Re:I am going to sue these people.

[BlueGecko]

Yeah, Bell may have made the telephone, but, dammit, if Darl can sue people for using Linux, then Edison should be able to sue people for using the telephone!

A bit telling

[Saven Marek]

This says a lot about good documentation and decent ease of use. I've seen many perfectly intelligent people come up against the brick wall of Zope's usability, and sit there scratching their heads going "wtf?". Luckily zope IS very powerful, otherwise it'd never end up being used.

While it's testament to the skills of the plone team that now there's a solution, and indeed that's the OSS way - if a solution is needed someone will write it - the years that zope's existed WITHOUT some kind of help it desperately needed is telling.

Re:A bit telling

[Malcontent]

While I agree that plone is better documented then zope I disagree that it's "easy".

Zope is just confounding. Plone makes it easier to get some things done. But sooner or later you are going to have to create a template or a script and then you'll be scratching your head and saying "wtf".

When I was first using plone it would take me hours to find where some text on the screen was being produced or where to go to change it. I am still perplexed about where the actions for the user bar are for example. And of course sooner or later everybody will get a visit from the "spammish aquisition".

I am waiting to see what zope3 and plone 2 are going to be like. I hope they make it easier for mere mortals to use them.

Re:A bit telling

[jalet]

I agree that beginning to use Zope without any documentation at hand is a bit confusing (to say the least). But really you should be able to setup and use Plone to create a basic working website with publishing workflow, membership, themes, i18n, etc... within minutes, and without having to type any line of code.
Then of course to go further, you have to read some doc, but just like with all powerful software.

Re:A bit telling

[the_rev_matt]

I've been using Zope for 4 years now, and I still don't get this view that Zope is inscrutable. I have been trying to work with Plone for over a year now and it's damn near impossible to create your own product for it.

While Zope documentation can be kindly described as minimalist, Plone documentation simply doesn't exist and what little there is is 100% wrong. Hell, I've had some of the plone developers send me solutions to problems and their solutions don't work (usually because they've skipped 2-3 major steps in their directions, assuming that I know as much about their undocumented product as they do).

I think Plone is a great project, and it will likely become an integral part of Zope, if you want to do anything other than slap a different skin on it you are SOL. I'm particularly bothered by the fact that they override many of the default behaviors of Zope/CMF and there is NO way around it so it is not possible to port a Zope/CMF product to Plone without completely rewriting from scratch.

Size does matter.

[cgranade]

Problem is, a "byte-sized" article would be one-half of a Unicode character, or shorter than all but two words in the English language. Not probably a constructive article.

What's Plone?

[Eric S Rayrnond]

Plone is built on top of the open source application server Zope and the accompanying Content Management Framework which have thousands of developers around the world supporting it.

Plone is ideal as an intranet and extranet server, as a document publishing system, a portal server and as a groupware tool for collaboration between separately located entities. A versatile software product like Plone can be used in a myriad of ways.

Who uses Plone? Many organisations. NASA / Jet Propulsion Labs, Lufthansa, the Austrian Government.

Re:What's Plone?

[dracvl]

Actually, the popular MAESTRO Mars Rover site is also a Plone site.

Does anybody else find it slightly amusing that Plone is running on Sun's network - on Sun hardware no less - and no Java in sight? ;)

Re:What's Plone?

[CapnKirk]

Ok. Here's the long description:

Zope is a web application server. It is written in python. It has a builtin web server or you can run it behind apache, squid, whatever. It maintains all data in a object-oriented database. What zope does is generate dynamic html pages on the fly. You write a page template and when the user GETS that page, the variable information is inserted and then returned to the user.

Another feature is that a Zope website -- and much of Zope itself -- is managed via the web. And it has a very sophisticated security and permissions facility. It uses the concept of "roles" to which permissions and access to objects can be attached.

CMF (Content Management Framework) is a zope application that creates a set of services for the website developer: navigation, calendar, new items, workflow, etc. It also provides the basis for css-based "look-and-feel".

Plone, as was noted in the interview, started out as a CMF "skin." It has evolved into kind of a "CMF best-practices". It's philosophy is -- in part -- to permit the creation of *sophisticated* web content in a collaborative environment by users who know little or nothing about html, etc.

There's lots more to be said, of course. But I've been using Zope for two years, and Plone for nearly a year. My preferred scripting language is Ruby, but Zope/CMF/Plone is so valuable, I went out and learned Python in order to read the source code. Today, most of my work involves writing a page template and maybe some snippets of python code to go along with it -- often less than ten lines. Simple.

If your need is collaborative web content creation/management, web portals, etc., and Joe Sixpack is your user, then Zope/CMF/Plone is the way to go.

A very satisfied user,

Kirk

Plone is self documented

[axxackall]

While I agree that plone is better documented then zope I disagree that it's "easy".

Plone is still in a deep lack of being documented. For example "Plone for web-designers" is till missed. Many API details I still have to get from the source code itself.

Also, one of the best Plone's documentation is a set of already existing and still being actively developed Plone applications.

But in general Plone still keeps guiding app developers, and thus leaves them more chances for future interoperability.

I wish that one of issue collectors/trackers in Plone will stabilize. Currently I use CollectorNG, which can already beat Bugzilla. I am not sure what PloneCollector developers want to achive by completely rewriting CollectorNG. As for official Zope/Plone issue collectors - they are kind of primitive.

Another wish: Zope and Plone sites will have forums with functionality of CMBoard, which I think is beating PHPBB already.

Re:Plone is self documented

[kalistra]

The single most significant improvement of PloneCollectorNG over CollectorNG, IMO, is that the Plone version lets you define your issue schema dynamically, through the web, using a very simple interface.

Also, because PCNG is built using Archetypes, the hooks for TTW schema editing have been backported to the Archetypes project, allowing for simplified development of TTW schema editing for ANY Archetypes based pieces of web content.

Plone 2, Archetypes

[supton]

I think things are getting better - much better, much faster. Soon-to-be-released Plone 2.0 and Archetypes are contributing to this - the learning curve for all aspects of Plone is getting flattened.

Zope has a "Z-shaped learning curve" -- or so goes the saying; this is becuase Zope (well, Zope 2) has a deep tree of class inheritance - a "deeply object oriented" system (to borrow a phrase from Jon Udell, not sure if that's his intended meaning). When someone tells you to "read the source" -- that's usually becuase Python is remakably easy to read -- but you still you end up with a task that's fairly involved and somewhat academic (not to discount this - once you get it it is quite rewarding).

What the CMF and Plone do is put a "wide-not-deep" framework on top of the Zope app server to abstract most of that tedious, academic learning curve for serious developers. The CMF hard-codes a really simple MVC-like design-pattern for best practices for component-oriented development, where lightweight components interact (global "tools" like search/catalog, workflow, etc and content objects in folders/containers (the model) - and UI/automation skin code (view/controller)). Each component is lighter-weight and pluggable (with defined interfaces and unit-tests), and CMF, Plone, and unrelated Zope 3 development are working towards not just pluggable components, but user/admin configurable components. The Plone 2 control panels are a good start towards making this more human. The ease-of-development and deployment story is getting better. The UI is also more configurable in Plone 2 via CSS.

Getting better by the minute: Archetypes is the secret weapon for Plone's future success; Archetypes makes schema-based development for content items, along with relationships among content items, not just easily possible, but much less tedious. It's architecture, in many ways (though it is still maturing), is superior to the same concepts in WinFS in M$ Longhorn. Archetypes will make development of content types easier to learn and develop day-to-day, whether you as a developer prefer to live in Vi (or Emacs), UML modelling tools, or a web-based schema editors. Simple, usable, documented examples for Archetypes development in Plone are popping up every day. Developing global CMF tools (singleton services/utilities for all objects in the site) has always been trivially easy, but underdocumented. Plone 2 is making the UI easier to customize, and I expect that forthcoming books and improved documentation on Plone 2 will make this straightforward.

Keep in mind, the Plone/Zope/Python stack is much less complicated and easier to learn than equivalent technology stacks in Java app servers (and less messy than inline web apps in PHP/ASPX/etc). And seriously, if you have to say WTF, say it on #plone on freenode or the plone-users list - there's a high likelyhood that someone will have an answer to just that question... ;)

Re:Plone 2, Archetypes

[Malcontent]

My biggest wish for plone and zope is a more productive working environment. I really miss my filesystem tools like grep. The find button does not work great at all and takes too long.

I keep thinking it would be great if there was a cvs like tool where you could check out the subtree to your hard disk and work on it with your favorite editor and then check it back in.

If not that then an eclipse or a jedit plug in would be awsome.

litte plone commercial

[2.246.1010.78]

I really liked them talking about increasing 'mind share' in the interview, because when I had to set up a new website for my local political party youth organization, the hardest part was convincing people that plone is exactly what we wanted - they never heard of it and it takes quite some time to explain the whole thing to people that aren't geeks at all.

what we wanted was:

• something with decent security - this rules out phpnuke
• something that is standard compliant, xhtml and the stuff
• something that is once set up and than easily managable by joe doe
• something that serves as a depository for information and relates the information to other relevant information
• something that seperates content and markup, plone1 does this halfway decent

Plone fit the bill and I think we can be quite happy how well it works, if you never tried it out: take the time and toy around with it a bit. The learning curve is a bit steep at the beginning (at least for the person that sets the whole thing up), but afterwards it is really a beautiful piece of software.

What exactly *is* Plone?

[0x0d0a]

I'm trying to figure out exactly what all of the hubub is about. From what I can tell from a quick Googling, Plone is a piece of software that helps people use Zope more easily, by slapping a GUI on it. Zope is a Python-based content management system. Content management systems are apparently something vaguely along the lines of Slashcode -- they store data in a database and let you more easily generate webpages from said data. They would be used by someone who wants to set up an e-commerce website or a blog-capable website.

I may be utterly wrong -- I'm a little surprised that I couldn't immediately turn up a simple explanation of what these things are on the web. If I'm missing something here, can someone clarify?

Re:What exactly *is* Plone?

[darnok]

OK, I'll go first - I'm sure people will correct me if/when I'm wrong...

Zope is a Python-based, Web application development system. It runs on *nix and Windows, and I'm pretty sure Macs as well. One of its key strengths is that it allows Web page designers, content generators and Web logic coders to work together without stepping on each other's toes - that's a big challenge with most Web application tools. You do all your work within Zope using a Web-based GUI, which is another unusual feature. There's a lot more to Zope than this, but that's enough for starters.

CMF is a Content Management system that runs on top of Zope. Content Management is for those sites where you want relatively non-technical people to be able to contribute "content" without having to worry about HTML and other nasty techo stuff. Think of people providing articles for your local school's newsletter - they should just be able to supply ASCII text, and someone else deals with typesetting and page layout. In this case, the "someone else" is CMF. There's more to CMF than that, BTW...

Plone sits on top of CMF, and adds extra tools such as workflow to CMF. In the school newsletter, you would probably have an editor who checks all the incoming articles, fixes typos and ensures nobody's said anything nasty. The contributor of the article would send it to the editor, who would then either accept or reject it. The "workflow" in Plone lets you implement this editor-type role in software. Again, there's a lot more than Plone than that...

Hope this helps a bit. I really like Zope, but as many people have said, getting your head around it is a bit challenging at first. Unlike many tools, it's difficult to "start with the easy stuff and learn the tough stuff as you go along" - Zope doesn't really lend itself to that approach, which I think is where many people struggle with it.

Re:What exactly *is* Plone?

[Qbertino]

Err, well, you've missed the mark a little.
Here's my quick standard reply:

Zope is an open source Web Application Server, developed allmost exclusively in Python (some speed parts in C) with an integrated object relational database, aka ZopeDB and a web frontend with access to all interal components.

That pretty much summs it up for Zope. :-)
Now for Plone:

Plone is a CMS and a content syndication system programmed for and with the Zope Appserver. These Zope Applications and 'addons' are very easy to develope and install on Zope (naturally, if you consider the description above) - think 'plugin' - and are called Zope 'products'.

So Plone it a 'tad' more than you're standard CMS, be it slashcode/e107/Nuke/whatever, since it can very easyly utilize the vast power of the underlying Zope and other products, like Webshops, syndication mechanisims or webcrawlers and data-mining bot's, just to mention a few. Zope actually severely blurrs the edge between database, application and frontend and leaves it completely to the developer where to draw the line between those components.
Imagine an appserver where you can just drop of data for storage at whim without having to mess with DB abstraction layers, conectors and stuff, that comes with a full featured web interface where you can track and modify the inerts of your appserver either by custom coding (in whatever language you fancy that has conectors to Zope, Perl for instance) or by using the interface options and elements - which you can of course provide with your own extensions.
That's what Zope and thus Zope/Plone is all about.
That one can't exactly say what Plone is in standard terms actually shows the power of Zope. Basically it's whatever you make of it.

Re:What exactly *is* Plone?

[stuntpope]

Good description, but I disagree with your assessment that "it's difficult to 'start with the easy stuff and learn the tough stuff as you go along' - Zope doesn't really lend itself to that approach."

On the contrary, with built-in authentication, roles, template engine, through-the-web editing, etc, etc, Zope makes it easy to build simple sites quickly with little digging into documentation. Heck, the outdated built-in tutorial is good enough for that, and the Zope Book (free online) is very good at getting you headed down the road to more complex web applications. Plone has done a great job of taking the framework that is the CMF and polishing it into a more 'approachable to mere mortals' product. That said, I'm one of those who don't believe Plone is the answer for all Zope application problems. Also, if you've never done web development using PHP, ASP, JSP or the like, then you'll have more than just the Zope learning curve to climb (thus you'll hear the complaint, "where's the content on the screen actually coming from?").

From what I've seen as a long time Zope user/developer who has management 'buy-in' and other developers now using Zope, the Perl hackers and people comfortable in a *nix environment take to Zope well. They know how to dig for answers. The developers raised on a diet of VisualBasic and who think Access is database development are flummoxed because they no longer have an IDE telling them what to do, and they don't grok using the source (by the way, there's also the DocFinderEverywhere product, quite helpful). I know this comes across as a flame, but it simply is what I experience at work. I don't deny that Zope suffers from confusing, often outdated, inconsistent and even contradictory documentation, but there are plenty of resources. The resource sitting at the keyboard is also critical.

Zope - a dream come true.

[Qbertino]

I've been dealing with Zope for quite some time now. What has been said here and in the interview about the weedy unfinished stuff and the (still) inconsistent documentation of Zope and it's Products ('Product' is a technical term in the Zope Appserver) is generally true.
If you don't have a knack at OOP *and* aren't willing to read through some messy, redundant and unfinished third party code experiments you're gonna have some hard time getting going with it.
Beyond that Zope is nothing less than the ultimate refrence for the way all server side stuff will be done in the future. Zope comes with a fully integrated object relational Database, runs with and on, what I call the fully GPLd equivalent to Java, Python and is an absolute breeze to develop with.
Technology wise Zope makes BEA, .Net, SunONE and IBM WebSphere look like some IE plugin in beta stage. Shure it can be a serious slowpoke on standard PC's, but nevertheless have I and some people I work with bet on Zope. When Zope - 3.0 is going to take a big step - reaches maturity in terms of documentation and community standards for developing, computers will be fast enough to make Zope the tool of choice for any server side thing one can think of.

The problem with Zope (and Plone)

[gunga]

Plone is cool when you begin to use it because it seems to work immediatly, has a ton of functionality and looks good. I don't doubt it's a fine piece of software but there are big problems with it.

First, Zope's internals are overly complex and sometime guided by ideological choices (the object database IMHO). It's a closed world with its own culture and logic. Its culture doesn't promote interoperability (again that's my feeling). It takes you out of the Web and into Zope Universe (after all this year, there are still problem when you want Apache and Zope to talk together, eg. Digest authentication).

Zope is often dubbed (I think Jon Uddel first said this) "Python's killer app" but I find it very non-Pythonic: overly complex, non-explicit and un-welcoming. Plone adds another layer on top of CMF, Zope, the object database... it is very difficult to understand.

I think that the best web setup is still a light and fast frontend (and PHP is good at that), a solid Database (PostgreSQL is better than a lot of people believe) and a third "business logic" tier which can be a separate application or shared between the frontend and stored procedure in the database. It's not the perfect theorical model but it's manageable, it stays simple (if you work hard enough to keep it simple) and you can evolve a simple website towards this model without restarting from scratch each time the requirements change ("embrace change", remember?).

I'm fascinated by Zope and Plone because they do so much and frankly, I don't know if I'd be able to write such a piece of software. But I think it goes in the wrong direction: the application server direction. It tries to coerce the light, simple and stateless nature of the Web into the heavyweight transactional world of corporate applications, just like the Java world does (Java Server Faces seems to make it worse). It is difficult to make a good Web application, but it's even more difficult when you fight against the Web and the way it works.

Re:The problem with Zope (and Plone)

[sirReal.83.]

In theory, I strongly agree with what you're saying. In practice, however, having set up my own Zope/CMF/Plone side on Debian unstable (yes, unstable) and Apache, I'm gonna have to disagree.

It took me a while to set up Zope/Plone. There's a nasty bug in Debian's distribution of Plone, but thankfully there's a super easy workaround in the BTS. It also took a couple visits to #plone@irc.gnu.org to set up an actual Plone instance, but in retrospect it wasn't that hard. I got the Apache passthrough working too. Now, Plone's setup phase is done. Anything I need to change is done via GUI, and the Structured Text system is perfect for marking up content without obfuscating it. Oh yeah, and the Structured Text was the only thing I actually had to look at the Plone docs for. It's probably even easier when you use them start to finish.

I've used e107, PostNuke, XOOPS, Slash, Scoop and probably a few others. They're all neat, but I've had way more problems with them all, from just plain failure to strange errors to lack of features. e107 does have a TON more themes available, though.

You say Zope is going in the wrong direction, but I fail to see how. With Zope, you can build webapps into your Plone site - just don't ask me how. With most PHP-based CMS's you still have to install an SQL server of some flavor, and I doubt if you can build webapps in. You complain about Zope/CMF/Plone being three-tiered, but really CMF is just an addon to Zope - it doesn't add complication. And I think it's pretty sweet to be able to manage all your Zope stuff through one interface - including all your Plone sites and whatever else you've got going. You also complain that Zope is a "closed world" because of the object database. Yeah, it's about as "closed" as any other website that has an FTP backend - i.e. NOT.

And as far as speed, I haven't noticed a bit of difference between Plone and e107 (the only other CMS I've ended up using for real). I'm not pretending I get any real traffic, though. But to balance that, I've got a horrible setup - at the moment a P4-1.4 with 256/RAM and a cable modem with an upload cap of 256Kb.

In short, Plone rules. Not sure how it does it, but it does.

It's dangerously unstable

[tylernt]

It stores all of your data in one big fat file. I had a guy at work using it for a small workgroup. That one big file got a little bit corrupted, and he lost the entire thing. Everything was gone.

It would make far more sense, to me, to store things in separate files. That 'all your eggs in one basket' thing, you know.

Re:It's dangerously unstable

[geniusj]

he should have had backups.. you can take exports of it as well in both Zope export format and XML.

straw person...

[supton]

There are not "default" behaviors of Zope/CMF - CMF is a framework - NOT NOT NOT a product. CMFDefault is just an example application. Plone is a CMS application on top of the framework an app server product. It is not "impossible" to port of CMF product (i.e. a product designed to work with CMFDefault) to Plone - there are very few differences. I do this all the time. I'd like to point out that CMFdefault is rarely used for any serious (paying) work or production products by anyone (not Zope corp, not independent CMF developers and consultants); Plone on the other hand codifies best practices learned from real installs.

Also, use archetypes - it is easy to create your own content type products with Archetypes. This isn't that hard, and the documentation is getting better.

Database Connectivity

[MartinB]

Interestingly, you don't have to use the native object database - there are connectors to most sensible third party databases, including PostgreSQL.

For All It's Merits...

[AShocka]

and there are plenty, as stated by many here, the problems I see with it are;

• Why isn't the Plone site a bazaar for the Plone community? Most other sites like Drupal show a very active community, but Plone.org seems like a ghost town compared to the others; no online discussion forum, which is a much better interface than email, especially when a new user wants to just check in and see what's happening. This scares a lot of users off, because this should be the centre of activity for a community. Zope/Plone/Python developers seem to tend to be focusing on consulting businesses, that's great, there's nothing wrong with that, but if that is the case, then Plone will remain a niche CMS because it does not function like other OSDNs. As a consequence it is not going to get the community involvement as other OSDNs do
• Resource hungry and no ability to write to a static htdocs dir. I know there are caching mechanisms in the Plone book, but it still does not really address this need. This may not be an issue for many of you that can run your own server and connection, and just throw resources at it, but for those of us who want a solution even running on a dedicated server (which usually have very little RAM), it can be a resource hog. Then, on top of that, having to dynamically serve content, that might just as well be served statically, is a huge price to pay on your server. There should be a way to use this app as a publishing tool and write the production version in static html as an option.

These are two issues I think the Plone/Zope community need to address if they want to be seen as a very accessible solution, because a lot of users will see it as a great solution to implement, start building something, then when it starts to really get difficult, the only real solution left is to hire a Python/Zope developer or consultancy, cause there just is not enough real solutions in online forums or documentation.

You can use a lot of other CMSs without having to become an expert under the hood, but everyone here is saying it's easy if you just read a bit of code. That's okay if you are either a great coder or dedicated to this CMSs path, but not for someone that may be just involved in web publishing and comes to Plone/Zope hoping for a complete web based interface solution. After all, isn't this the intention of Plone/Zope?

This scares off a lot of potential adopters. Even Cocoon and Axkit are better documented and have a lots of user discussion (admittedly on email lists)

Re:Incorrect on both counts...

[AShocka]

If you look at that IRC count and rank that as popularity, you could be right.

But someone else might look at that and see Drupal:16 Plone:76, the Drupal users found what they needed in the docs and forums so didn't need to go on IRC, where the Plone users needed too.

Again, for all the good points you are presenting, why is this not clearly documented on the Plone web site. Then users would know what type of CMS and developer community they are getting. There might be a larger adoption base and larger customer satisfaction rating.

It may be great for developers. But what if I am a manager of a small to medium publication (wasn't Zope first developed for a newspaper publication?), I'm looking for something. I do want real information about this product. I look through the web site and download the Plone book. Really, the book nor web site do the Plone community any justice as a true knowledge base.

If it is such a good product, and if it really should be deployed more, then what everyone is saying here needs to be presented on the web site and in the book.

Also, I see a lot of people really frustrated by the Zope books out there. Maybe there is need for just one good Plone/Zope book.

Industry is littered with a history of superior products failing, why, because often they were not presented or marketed appropriately.