Moving to Jabber in the workplace

Jabber Journal Issue #21 is out, and with it come lots of links to Jabber servers and a link to r.i.pienaar’s post about Jive Messenger (a free, Java-based Jabber server).

There are a good number of easy-to-install Jabber servers these days (Jive Messenger is incredibly easy to install on both Linux and Windows machines, and I’d recommend it without reservation), but providing decent clients is a different issue.

If you want to roll out an internal IM-system, but can’t afford MS or AOL, then Jabber is probably the next-best choice in terms of servers, but no Jabber-only client I’ve seen can match the usability or feature-set of the current MSN client (let alone the ease of install and built-in access to Active Directory).

Before we starting crossing clients off our list, lets run through some features that an IM client in the workplace should have before they can be adopted:

  • MUST run on Windows 2000/XP
  • MUST be user-friendly to the level that non-technical drones can pick it up as easily as they use Windows/MSN Messenger
  • MUST support client-client file transfer
  • MUST have a low-cost pricing system (if not free)
  • MUST be able to provide SSL communications
  • MUST provide local logging of conversations (chat history)
  • MUST be able to deploy the correctly-configured client easily
  • SHOULD support video/audio transfer (e.g. webcam chat)
  • SHOULD have a server-based update mechanism to force upgrades
  • SHOULD be in active development

Well, that’s a pretty hefty list, so whilst I’m not really expecting anything to actually meet it, I’m certainly not letting the clients off the hook 🙂

Despite everything else in the list, ultimately the user interface will be the deal breaker. Not pretty enough? You’re out. Not consistent? Failure. Too complicated? No chance.

If we look at the most popular Jabber clients for Windows, we see Exodus, Rival and Psi; none of then have particularly beautiful user interfaces, and are all quite hard to get started with. Of the three I prefer Psi, because it’s very clear what’s going on: “here is your roster list, click on it and do stuff”, but it’s inconsistent in places, minimal and very unlikely to win over any MSN users.

a screenshot of pandion

Looking further afield, there are two other noteworthy Jabber clients, Pandion and neosmt. Pandion is excellent; its user-interface is very friendly to those coming from an MSN background; it’s free for internal use in a company; it’s also brand-able because the UI is provided by HTML and CSS rendered by IE. It has a limited implementation of the XMPP standard, but it doesn’t really matter too much, because all the core features are there but it does have a very nasty bug which disconnects your client if, for example, your roster is too big and takes too long to deliver. neosmt, on the other hand, is a very good client with a fuller protocol implementation as well as other great (but proprietary) things like whiteboarding and audio/video sharing, automatic update checking and so on – all things which would be useful in a company. The user interface is pretty good, but suffers a major blow by actually having the main menu hidden away under a cryptic ”>>” icon on the title bar which doesn’t even look like a button and not (properly, at least) supporting JISP iconsets, so no MSN emoticons for all our poor we-fear-change clueless employees.

a screenshot of neosmt

If we cast out net wider to the multi-protocol clients we see things such as GAIM, Trillian and Miranda. In their default states, we take one look at GAIM and Miranda and instantly throw them away (Miranda can be made to look much prettier, but not without tweaking and so on, and no-one’s going to configure and rebuild the installer the way I want for me). Trillian is, er, definitely acceptable by the MSN crowd, but sadly the free version doesn’t support Jabber and Trillian Pro is $25/client, so pfffff to that.

Which leaves us back with neosmt and Pandion. neosmt is clearly the superior client (with more of the features listed above than any other client I looked at), but harder to use and a sea change from the clients most people are going to be used to. Pandion has the basic messaging and user interface down to a T, but little else (and that nasty bug to boot). Clearly a middle ground would be absolutely perfect, but is pretty unlikely.

In the end, selecting a client will come down to the needs of the users who are going to be using it, but I think giving neosmt a run and falling back to Pandion if neosmt proves unacceptable seems like a reasonable position to take for rolling out Jabber clients on a Windows network, and will probably be the approach I’d take.

Calendaring and information gathering

Just some stream of consciousness:

The Hula Server was announced the other day. Hula is a calendar and mail server.

They’re planning on using CalDAV as the backend for the calendaring system, which is a nice idea, but probably not realistic. I don’t think I know of any WebDAV-enabled servers I can use, and I certainly wouldn’t install it on my own server. I can understand why they’d want to do it. There is no free decent alternative to Exchange etc. on Linux (I’ve posted about this a few times before).

Apparently their original idea had been a whole groupware-scale “solution”, but fortunately jwz put the kybosh on that (that link is worth reading regardless – it’s great content on how to make software that people like, and want).

I’m not really sure why a mail server needs a mail client in it really. Surely if you just make sure that Thunderbird, KMail and Evolution can use it you’re fine and don’t have to put the early effort in to make the web interface, better doing it somewhere else?

The calendar stuff is interesting. I have my own half-solution based on a hacked version of my Tellybot, but it’s not really terribly good (I have a Python script which outputs my Outlook calendar courtesy of Aaron Johnson but it’s not hooked up to anything) – it sends me reminders, but that’s about it. I’m thinking about having it output .ics files and letting PHPiCalendar pick those up and display them as well as emitting RSS files. RSSCalendar.com is another possibility, but doesn’t have an API, so as much as I like it (in stark contrast to my intial hating of it), I wouldn’t use it for my own stuff (I would however recommend it if my family wanted to do stuff, and then pull down the .ics files myself and put them into my own calendaring stuff).

Along with all this groupware-style stuff I clump in tools like Dashboard so that in some interface I can type in someone’s name and have all sorts of info about them returned to me, like in this screenshot (click for bigger):
which I should also point out is a copy of this image from the Dashbaord website, reproduced without permission).

At the moment I’m doing some preliminary work on a sort of personal portal like the kind mentioned by barb dwybad that includes my flickr photos, my books, my del.icio.us bookmarks, my music and my IM history (parsed from the Psi and MS Messenger history files), which is just a PHP page using Magpie (because all those things are available in RSS format) but I can’t see a reason why this could be extended to work for anyone who had those details (i.e. those URLs) specified in a FOAF file. Well, actually I can, and that’s that I can’t think of a way for machines to recognise that a URL is a link to an RSS feed of bookmarks rather than anything else, other than by detecting the “del.icio.us” in the URL, which is incredibly lame since I thought RDF was supposed to free us from that kind of thing.

Of course, a lot of people don’t have FOAF files, or even if they do, don’t link to all their other feeds and so on (we can’t all be Leigh Dodds), so how acceptable is it to create those files for people (whether they’re then published privately or publically)? For example: Jason Kottke – who has a has a weblog, listens to music, takes photos, reads books and so on.

So, in summary: Hula looks nice, but I don’t think it will take off; Dashboard is great, but mainly unavailable; FOAF can provide us with Dashboard-like features; calendering is hard; writing FOAF files about people is rude.

I thank you 🙂

Firefox and banks

At some time in the distant past my Dad worked for Playboy Casinos and recently he heard that there was going to be some kind of reunion for ex-staff, but not knowing any details he decided to look it up on the internet (where he uses Internet Explorer). The first thing he did was to type ‘Playboy’ into Google. You can imagine the rest of his day, closing popup after popup whilst trying to find the right site. In the end the popups won – they were appearing faster than he could close them so he gave up, turned the computer off and didn’t bother looking again.

Whilst visiting few weeks later, and having heard this story of woe, my little brother decided to try and help. First of all he installed Ad-aware and scanned the computer which removed around 250 ‘suspicious’ items. Then he installed Firefox and helped my dad find the information he was looking for, popup-less. He set Firefox to be the default browser and left my dad to it (a few weeks another relative visited and removed the remaining ~50 items that had hidden themselves).

The problem is that, as non-tech savvy as my dad is, he does actually use internet banking and his bank, Nationwide, don’t support Firefox and so he had to return to IE.

On Nationwide’s “technical requirements” page (oh yes indeed) they say that Mozilla (up to version 1.6 not including Firefox) is Functionally supported. Mozilla 1.6? That’s pretty recent isn’t it? According to http://www.mozilla.org/releases/ it came out in Jan ’04, and a month later Firefox was at 0.8, so I can’t possibly imagine why they don’t support Firefox (were there really major differences between the two?).

Whilst he was happy to make the transition to another browser, there’s absolutely no way he’s going to spend his time trying to remember which sites work in which browser, so it’s back to IE he goes, armed with a popup-blocker. Of course, that won’t solve all his problems, and at some point the machine will lock up again due to some IE exploit (they never download the Windows updates because they’re on dial-up) and someone will need to go down and fix it. And all for the want of Firefox support (ahem, mostly :).

A bit of further investigation (although, admittedly, not too deep – I don’t plan on opening a bank account just to find out what’s broken, and my dad lives over 200 miles away) revealed this little gem on Martin’s blog:

The Nationwide online banking site works fine in any browser. In fact, I highly recommend the service. Not only have they continually improved the interface to make it more simple and less graphical (there’s hardly any images), but you can now transfer money between your own accounts instantly (no clearing), pay bills and set up standing orders.

The only thing that doesn’t work on Mozilla Firefox is a javascript dialog box which appears when you log off prompting you to restart the browser. However, as this is probably only designed to avoid security holes in ie, I’m not bothered that it doesn’t work in Mozilla.

posted by: Frankie Roberto on July 10, 2004 10:21 PM

Could they really be not claiming support for Firefox just for that? It would seem so. Walking my dad through installing a user-agent spoofer over the phone is out of the question, so it’s IE only for him from now on, and a Firefox user lost. Things like this must happen every day, and it’s really tragic that someone who could really be helped by the features of Firefox is denied by the writers of the website.

To be fair to Nationwide, they’ve clearly set out what the requirements are, and do actually provide some support for Gecko browsers (albeit a browser that’s a year old, although that’s nothing in IE-years!), as well as handheld devices (not much luck for Opera though – they support all versions except 6.01 to 6.04 and 7, so that’s all the good ones then), but with the ever-increasing number of Firefox installations, can they (and other major websites) continue to justify not updating their sites?

Having to recode for syndication?

I’ve just realised a flaw in my feeds.

When I provide a blockquote, I always fill in the cite attribute, and then use JavaScript to pull it out and display a link to the source after the quote itself, but of course this doesn’t work in aggregators, and they’ll never see the link to my source, let alone know where it’s from! They’d need to either look at the source of the feed, or visit my actual webpage to find out, but there’s nothing in the content of my post which hints that they might not be getting all the information.

What can I do about this calamity? My first thought is stylesheets, but nothing doing – the aggregator does what it likes!

Does this mean I’ll have to start including an explicit link to my source when I quote someone? Surely there’s something else I can do?

Also check out Invisible Information: Machines get all the fun by Paul Haine.

Gosh, Gravatars are useful after all

Since the time I first saw them I’d thought that Gravatars, Favatars and FOAFatars (OK, I made that last one up but this is what I mean) were cool, but useless. I appear to have just proved myself wrong.

I regularly read Rich Rutter’s blog Clagnut, and he recently implemented Gravatars on his site. On some of his recent posts a guy called Rob Mientjes had left some comments, and he’s got a Gravatar enabled which I’d noticed and thought “hey, that’s a nice-looking Gravatar”, and then forgotten all about.

Tonight I was just idly browsing Flickr, looking at all of the photos tagged with “London”, when I saw a photo of the Make Poverty History rally, so I clicked the link to see all of the photos tagged with “make poverty history” and saw a photo by Neil T (who I happened to know was Neil Turner of neilturner.me.uk fame, and reasonably local to me).

So I browsed around his photos a little, to see what was new, and came across this one where one person had left a comment on that photo: Rob Mientjes. I would never have recognised the name, but his Flickr avatar is the same as his Gravatar and was instantly recognisable, so just by seeing the picture next to his comment, I knew exactly who he was, and why he was probably looking at Neil’s photos. Astonishing. Quite frankly, I’m almost more amazed that some part of my brain is actually dedicated to remembering these things 🙂

Gmail invites


Gmail appears to have gone crazy and started giving out masses of invites. Perhaps they just bought a load of new hardware or something? Anyway, if you want a Gmail account, drop me a line, and you’ll be arbitrarily judged 🙂

Posted from flickr.

Hacking extensions

The thing I really love about Firefox extensions is just how eminently hackable they are.

I already know XUL, JS and RDF, but even if I didn’t, changing extensions to do what you want is so easy that anyone could do it.

Take Feed Your Reader for instance – it’s a marvellous extension which replaces Firefox’s default feed-disovery behaviour (click the orange button and subscribe to feed as a Live Bookmark) with the ability to subscribe to a feed in an external aggregator.

The selection of aggregators is pretty good, and seeing as most support the old Radio Userland http://127.0.0.1:5335 method of subscribing, it’ll work in some that aren’t even listed.

Sadly though, I use JabRSS as my aggregator, so there’s no such support for it built in to FYR, as it requires a Jabber client to do the actual subscription. So, five minutes of coding later, and voila I have a JSP page which accepts a URL as a parameter, logs in as me, and sends a subscription request to JabRSS (using Smack). Another five minutes later, and I’ve hacked FYR to pass feeds through to my subscription page, so I now have JabRSS subscription from within my browser. Brilliant! I’ve actually been meaning to do this for about three months, but it’s one of those things that wasn’t important, just nice to have and so has just sat at the back of my mind, but now I’ve got it, it’s great! I can’t imagine why I possibly didn’t do it sooner!