Well, it’s here: after far too long an embarrassing silence, I’ve finally taken the Two Worlds web site into the technology and presentation that matches my thinking and working practice – while I can argue that I’ve been too busy with vision, engagement and strategy for clients to have time to address my own, that would be but a small part of the truth. Finally though, pragma and need have coincided, so here is the first public iteration of the Two Worlds site. It’s new, it’s but as yet sparsely populated, so please do come back regularly as I develop the social, commercial and technological thought themes behind my business, add current and historical information around the Ubiquity model of the enabled society and then leaven the whole with a little humour, technocracy and random digression. Alternatively, subscribe to any of the site’s XML feeds (see pretty buttons to the right) to be kept informed of changes and updates.
Here’s an opinion: The Web is about being accessible to all – it is not, nor should it be, the domain of any one operating system, organisation or web browser. There are a good set of international standards which determine how information is delivered to and presented by browsers. Most – no, make that, “nearly all” – browsers are compliant with those standards, within a few degrees of buggishness and interpretation. So making a site work with these is a matter of tweaking by degree, not kind. There is of course one notable exception, and that (again, “of course”) is Microsoft. And here we do appear to have a combination of conspiracy AND cock-up: Microsoft are trying to drive/keep the industry in thrall to a proprietary browser and the related server architecture. They are also guilty of producing a product that, in terms of compliance to standards, is full of bugs, ommissions and misinterpretations of best practice. Whether those are driven by corporate decision, gratuitous disregard or blind ignorance is another matter. By whatever means though, its browsers display a moderately cavalier disregard for standards and are of such a bug-ridden nature that making a site work consistently requires delving into an underworld of hacks, tweaks and rewrites that are sufficient to cause apoplexy or death-by-boredom in any thinking organism. Approximately 40% of the development time for this site has been spent in trying to implement fixes and work-arounds for Microsoft’s browsers. In comparison, tweaking for all other browsers has been, in most cases, a matter of minutes.
In order to tread the fine line of compromise between high-handed disregard for poor design and monopolistic practice and preventing the many users of such products from actually accessing these sites, we’ve gone for the “greatest good of the greatest number” and made everything work with W3C DOM-based browsers and the later versions of Internet Explorer, on Windows and Mac. But please do consider this, by preference, an ABM site: Anything But Microsoft. If they ever learn and decide to create standards-compliant browsers, then that’s just fine and dandy. In the meantime, I look forward to the day when the world’s web designers bring a class action against them, to claim for the time, brain cells and money lost in trying to make their bloody browsers work. Me, I’m off to ride my motorcycle.
Our core web technology platform (the Two Worlds vServer) is aimed at communities and enterprises which require universal content and collaboration services that can be updated and managed in multiple ways, from mobile devices and in strange and exotic places where bandwidth and means of access may be decidedly limited. Many users have English (if at all) as their second language, so I want them to be able to both post in their own language, for those posts to be avaible in other languages and to all users with direct translations of any site content. Here I call in the excellent service at freetranslation.com, which it is possible to call directly from a web page to carry out a machine translation of that page. In fact, if you subscribe to their Platinum service, at an astonishly reasonable $3.95 a year, you a) encourage them to keep going and b) get the facility to refine the dictionary for informal language and automatically follow links in the target language once you’ve translated the first page. So…
The vServer that hosts our own and our clients’ web sites is based (loosely) upon the Movable Type blogging engine. In the versions we’re currently deploying (2.661/3.11), there appears to be no embedded way to get at the current page URL for any page on the site – MTEntryPermaLink will give you the effective URL for an individual entry, but that’s a special case. So what to do? Enter php, stage left, with smug grin: Now I’m a complete newbie at php, so there may well be much easier way of doing it, but what the hell – this one works, at least in my production environment, which is, FYI, Mac OS X/X Server 10.3.5/Apache 2.0.47/php 4.3.2/MT2.661.
What and Why?
I am not now a programmer and haven’t been one, real or surreal, for at least fifteen years. So WHY have I just spent two days solid trying to get a single perl module compiled for my server? Rhetorical question – any smartarse answers should be placed in the usual wormhole on a piece of EBCDIC punch-tape.
The Challenge: install the perl module Crypt::DSA on a shiny new Movable Type 3.11 installation, to allow us to make full use of the features and get best performance from the comment registration mechanism. So I go to CPAN and try installing Crypt::DSA. It tells me I need the Math::Pari module first and then falls over in grand style on trying to install it. That was four days ago. I have just got it all working (or installed, at least) and this is for anyone going through a similar process – hopefully you’ll find this before you acquire significant bruising on your forehead, much like the one I’m currently nursing.
Firstly, useful resources:
The starting point for all this is Benjamin Boksa’s description of installing Math::Pari, taken together with some hints and pointers found on a Mac thread on the Pari site. That tutorial was then corrected by David Jacob’s blog on installing Math::Pari on Mac OS X.
Whilst pointing me in the right general direction, none of these actually worked. So here’s what did, with my system configuration of:
- Mac OS X Server 10.3.5 installed and up-to-date (according to Software Update)
- Xcode Tools installed and up-to-date
- Perl 5.8.1 (default Mac OS X installation)
- Movable Type 3.11
- Pari 2.1.5 – as of today, this is the latest stable release of Pari. If you’re using an earlier or later one, you’ll need to change the process to suit the names. D’uh…
So, for those who are trying to install Crypt::DSA, or anything else that needs the Math::Pari module and libraries, here’s what worked for me, presented in step-by-step fashion, with minimal geekspeak, although it does require some familiarity with the Unix command line. using the tcsh shell. Some of this may be overkill, but I’m being deliberately pedantic here and describing every step that led me to a working installation.
During the development of the Two Worlds vServer and my subsequent iceclimb (with crampons and two axes) up the North Face of the CSS learning curve, I came across a bug in the Mac OS X version of Internet Explorer, up to and including 5.2.3. Nothing new there, except that appears to be unique to the Mac version and doesn’t appear to be desperately well documented (at least for this CSS newbie).