HTML Logo by World Wide Web Consortium ( Click to learn more about our commitment to accessibility and standards.

Moving forward with Composr

ocPortal has been relaunched as Composr CMS. ocPortal 9 is superseded by Composr 10.

Head over to for our new site, and to our migration roadmap. Existing ocPortal member accounts have been mirrored.

Nose to the grindstone

Nose to the grindstone As discussed previously, Composr/v10 is a relaunch, not just a pile of new features. One big part of this is refreshing all our documentation.

Some of my staff recently spent about a month of man hours between them re-doing all our documentation and addon screenshots, which is about 500 images, including about 30 schematics.

I've just spent most of the last few days resolving outstanding issues related to this, and now it's all committed to our git repository.

It's hard to think of anything less sexy in terms of the work. In fact, honestly it was incredibly tedious for all involved. However, this is a part of our commitment for quality with v10. Not only were the screenshots we had terribly outdated (about 10 years old in most cases!), they had "ocPortal" slathered all over them.

Comparing the new and the old screenshots, it really shows just how far we've come in 10 years. This reflects our own advancement, but also the massive changes that have happened in the web design sector. The web has gone from a place where hobbyists made "home sites", into the media centre for all our lives. I used to opine that web design fidelity was getting comparable to glossy magazines – now I can see in many cases websites now far exceed the design quality of magazines.

(To be honest, I think I was remiss to leave the documentation screenshots unrefreshed from v2 to v9, many were very ugly looking and out-dated and it may have been the cause of some "ocPortal is ugly" complaints. Then again, updating them has been a pretty big cost, so we can't repeat that regularly unless we get volunteers.)

When we launch Composr I will be incredibly relieved (the development has been arduous to say the least, and we're still not there yet – lots of small or medium sized tasks to do), but also very proud of our clean and modern platform, cleaning up after a 11 year product evolution.

The fact that we've been able to re-take all these screenshots says something – Composr is now pretty stable! All the primary functionality and non-bundled addons had to be running well enough for us to go through taking these quality screenshots. A few hundred bugs had to be fixed for that. This all said, I need to insert our usual caveats that Composr is not done until it is done, that we're not working to a release date, that we have had a big staff shortage to contend with, that I have ended up too much as a bottleneck in the process, and that nobody should plan around it until we hit beta. Very occasionally people start swearing at me for not getting things done fast enough so I need to be clear here about what expectations should be.

My epilogue to this blog post is to talk about fallen non-bundled addons.

While going through everything, essentially doing the first full re-test in quite a while, sadly some of the addons were found to no longer be viable, or no longer worth upkeep. One thing observable over recent years is the corporatisation/professionalisation of the web. A lot of consumer-focused widgets and APIs have been dropped because the target market for them has declined (fewer non-professional homesites), technologies have changed, or web companies have just decided they are not financially viable to maintain.

1) lastfm. Lastfm had deprecated the API we used to use. I tried to sign up to their new APIs, but their site had become a massive mess since they relaunched a few months back and is significantly broken. Their old site and code (which includes the APIs) seems to have been left in a zombie state.

2) amazon_wishlist. Amazon have dropped their wishlist API. However, the affiliate sales block is still viable and I have rewritten it to use their more modern widget code (same goes for justgiving and ebay blocks - rewritten, so viable again).

3) xmppchat. Killing this was not forced, it was my decision. XMPP as a technology has declined seriously over the years. The commercial companies using it (e.g. Google) have withdrawn, and proprietary chat systems have taken over (Facebook Messenger, Skype, Snapchat, …). Ordinary users expect more from chat systems than XMPP can deliver, such as video chat, or at least convenient file transfer. Maybe XMPP could deliver, if major companies were behind it with implementations and Open Source code, but they're not, so whatever standards may exist have become irrelevant. IRC has absorbed the users XMPP had, and has shown to be the legacy chat protocol that has the staying power. Honestly I never liked the xmppchat code either: it required thousands of lines of complex and ageing JavaScript, a dedicated server to run an XMPP server, very complex server configuration and maintenance, and seriously bloated up our git repository. I don't think anyone ever used the addon except us, and even we did away with it and switched to iframe'ing a third party chat service, due to the temperamentally of it all (basic things like proxies could kill it). Future chat efforts are expected to centre around our own chat system (for an integrated on-site solution), and also in us recommending integratable third party systems that do a better cross-platform/cross-device job, particular involving video and modern web standards (so WebRTC stacks).

View all


There have been no trackbacks yet