Club Linux Coffs Harbour

Nice to hear from you too, Ruben!

Things have been a bit hectic around here with child raising and renovations... have to be stupid to do open-source as well! Too real for comfort really!

Maybe I could commandeer you into trying to open the xanjax.org site on a Macintosh - if you're still at the Uni - there must be at least a few laying around. I still have absolutely no idea if the site will work in Safari. I would be very grateful if you could, and post the result, good or bad, to the xanjax blog.

The Gentoo experience was a good one for me. It's very good to install just what you want, have a configuration which is not someone else's idea of what a configuration should be; plus you don't ever really need to do another system install as Gentoo is capable of transmogrification over time into whatever one considers the 'latest' version to be. Finally, as monopolisation extends ever deeper into private territory, binary only distributions may one day become illegal. I hope it never gets to that! Even the big guns will not ultimately be able to extinguish privately held source code because it will be held by commited people. Although you can download the source of other distributions, few actually do, myself included, so Gentoo effectively provides a bit of insurance. Finally, emerge and portage. Emerge is a bit tricky at first, but if you keep your installation you do it once only. Portage is as easy as any linux software updater and better than most. The documentation on their website is great.

I had a quick look at your website - I miss the trips we used to make to MinnieWater. I had no idea you were into nature walks otherwise I would have invited myself along!

Cheers

David - openPC Labs - Xanjax.org

I hope you ran Xanjax through the lynx (mangle?), considering it was your comments that led me down that path! :~]

As far as the AGPLv3 goes, I have to admit I'm not really sure if it was the right move, so I have retained commercial rights so xanjax could potentially be re -released on another free licence in future if a. it ever becomes popular and useful, and b. I made the wrong licensing decision.

The JavaScript being downloaded onto everyone's PC seems to me to be the crux of the issue - any website using licenced JavaScript software is without doubt 'conveying' - the actons of the end user are largely inconsequential, most will not even be aware of any use of software and cannot re-convey it unless they know what they're doing. The object is to get web authors to do the right thing, and to hopefully contribute sometimes too.

Once I get past the basic framework reliability hurdle, lots of your other comments will be answered.

I intended for example to do the typical AJAX loading indicators but decided to err on the side of 'release early' and 'release often.' The other thing is that I'm not sure if that's supposed to be part of the framework or not, but should certainly have been built into the website.

I use one of those ultra cheap shared hosting sites in the US against my better judgment - latency, even a few seconds deadtime, is too often a problem now, although it wasn't before the operator got too many customers; time to move on I suppose! That might be why xanjax was slow for you; occasionally it is for me too, especially in US daytime. Xanjax can help a lot with a hosting situation like this, but of course can't correct seconds of deadtime. Probably I should get a better host so people don't think it's xanjax otherwise they won't see the point.

The pre-caching idea is in mind already; perhaps it could be automatic based on content popularity metrics. This is already being done on many commercial sites to force feed victims with commercials... yuk! But let's use it for worthwhile puposes. From what you say I don't know if you're suggesting the user creates the demand list. I have thought of using a mouse over to begin/cancel an update but haven't tried yet.

If/when I find time, these features will happen.

As far as libraries like jQuery go, they have some very useful routines, but I think they miss the point a bit. Loading a static 14K (or even 7K like xanjax) is a huge time hit on a slow connection. One day, xanjax will be capable of intelligently loading compatible routines and CSS. Instead of a static compatibility 'layer' the AJAX approach to compatibility should be implicit - what is needed is loaded; nothing else. At the moment, I deal with compatibility as far as possible by narrowing my code to commonly supported functions; there's very little browser specific stuff. This is surprisingly effective and not as limiting as one might imagine.

I haven't looked too carefully at the typical 'stack' style analysis of what xanjax might look like yet, quite purposefully. I don't want to get limited in my approach by a technical analysis which probably is more useful for technicians and end users. Some things 'stack' quite nicely like hardware and software layers in network design. I'm not convinced that web design neatly fits into such a model, although it certainly may be useful as a mental tool.

Your comment about the 'M' layer becoming http is interesting - is this a RESTful principle?

It seems to me that there are two sharp ends to the web design dilema, both opposing (of course). One is user demand for easy, unlimited use, the other is scalability (at the server end). Oodles of excess processing power already exists at the client end whereas quality server power is both limited and hugely expensive. The fixation on centralisation, even for content, is one largely brought about by the monopolisation desire of capitalism. It's in our thinking. Free sofware offers the potential to escape some of these limits. Perhaps, for example, a 'future' website might exist everywhere and nowhere, as an array of self migrating nodes on thousands of private PC's; something of an expansion of the idea behind Bit Torrent. Getting the browser to do most of the work would be fundamental to such a scenario. (And wouldn't it worry the hell out of the copyright cops!)

So, now, maybe we need a 'C' (co-ordination) layer instead of an 'M' layer, to enable the database to be replaced with http?

If Xanjax makes even a micro-step in that direction, or inspires someone else to make it work I'll be very happy. But we're right at the beginning and maybe it won't even get off the ground in a small way.

I've recently transformed the HTTPRequest routine into an object so we can now run multiple download requests simultaneously - now we can begin to go beyond a basic website into more dynamic micro content. That's the next step (I think), after which I probably need to think about a more structured approach to dynamic module loading and API's... hmmm!

Cheers

David - openPC Labs - Xanjax.org