Tuesday, 28 June 2016

Wiki 001

I think we're agreed that we want a wiki for at least some functionality.

I also think we're agreed that we want to use free, open-source, non-proprietary software. This narrows things down considerably.

I'm willing to shelve the idea of leaving open an obvious pathway to getting the data into 'semantic web' triple form. I think this means we don't need to consider or discuss this aspect of things any further. I've also shelved the detailed proposal I was working on, as I think it would only be a distraction from getting started.

Technologies considered so far are: MediaWiki, Waliki, Tiki and Xwiki. If anyone else has found anything suitable, obviously we can consider that. I've haven't after a fair bit of research.

Of these, I think there's general agreement of those expressing an opinion that MediaWiki is too Byzantine. Waliki is in early development and doesn't have a DBMS, using github as a back-end instead, so I'd strongly prefer not to use that. The other two seem viable.

Chris, via one of our round robin emails (I assume it's OK to publish this):

From playing around with it, XWiki looks very powerful, but possibly a bit intimidating for us as administrators to get set up; there are a lot of options available. PHP vs. Java probably shouldn't massively inform our decisions, except in that it might be slightly simpler to get PHP apps set up with a web-hosting company. However, I don't foresee this being a massive issue either way. I will say that I have more experience with Java than PHP,  so that does make me lean in its direction a little
I agree with this, and think Tiki offers us the best chance of getting started quickly and using features 'out of the box' while affording us the ability to customise functionality as we choose. Php has apparently been substantially improved as a programming language in recent years, and is definitely more widely supported by hosting providers etc so allows us more option there should we need them.

So I'd like to propose that we go with Tiki as our best option given that we want to get things rolling quickly.

I am in the process of getting a server set up which we can use as a live system under the labourroots.uk domain. Is Windows acceptable as an OS? Tiki will run fine on it. I'm not trying to steamroller this through, just to get things moving forward after a brief (apparent) hiatus.

One issue we need to consider is how comments/talk pages are to work. In wikipedia, talk pages are an unstructured nightmare. Somewhat better formats are available as extensions to MediaWiki, and in other more recent wiki software. However, especially in the context at least of using a wiki for policy development, we should certainly be thinking about how this is actually going to work, and whether talk pages are likely to descend into long-winded slanging-matches.

Tiki allows forums to be used to discuss Wiki pages, which is better than the unstructured nightmare of MediaWiki talk pages, but not sure it offers the kind of tight integration between edits and discussion I'd ultimately like to see - but not sure any other Wiki software does either so that is a wash. Ditto creating tighter and more easily interrogable tracking of exactly which contributions were made by whom.

I'm going to go ahead with setting up a Tiki installation for now, but obviously any objections, suggestions, etc can be made and addressed here.





5 comments:

Tim Wilkinson said...

Comments invited on http://strategy.labourroots.uk/2016/06/wiki-001.html

Chris MacMackin said...

Well, wearing my free-software-zealot hat I'd rather Linux than Windows, but from a technical perspective, I don't think it should matter for something like this. Linux may sometimes be cheaper to get from a hosting company, though.

I agree that no wiki really seems to have a tightly integrated and organized discussion system as we might like. I don't know if anything does. GitHub would be the closest approach, I think.

Chris MacMackin said...

That comment shouldn't be taken to mean that we shouldn't use Windows, by the way.

Tim Wilkinson said...

1. re: Windows - you're probably right. I was just being a bit lazy as I am used to using it. I'll see about Linux.

2. re: GitHub - do you think we ought to be (re)considering Waliki in that case? Did you happen to see what xWiki/its extensions do with respect to this area? I didn't get to the bottom of that entirely

3. Another issue I think is worth trying to get right from the start if possible is references & sources. Most material we produce, e.g. educational, polemical and reference material as well as policy documents, needs to be properly sourced. This means making it mandatory (and therefore also easy) to supply proper references in some way that permits cross-references, re-use of refs etc. This is conceptually reasonably simple - the only issue is modelling and developing a concrete implementation. I have some ideas about this which I'll put in a separate comment later.

Chris MacMackin said...

You don't need to use Linux--that would just be my preference if I was acting as a sysadmin (both from an ideological standpoint and from the fact that I am more adept with it). Go with whatever you're most comfortable with/can afford.

I wasn't meaning to suggest Waliki, no. In any case, it's not based around GitHub (the website) as the backend, it's based around git (the version control software). Any decent wiki system will have version control, so we should be fine in that regard. My reference to GitHub was more of an ydiscussion from going off track, but I don't think any software platform can overcome that.

I didn't look thoroughly through the xWiki extensions. There may be something in there, but I don't know. I just thought it looked like an easier platform to extend if we needed to. Of course, with increased extensibility comes increased complexity to manage, so there's a trade off. I'd be fine with trying Tiki, at least for an initial experiment.

I don't know what to contribute in terms of references. For my academic research I take the approach of keeping a massive BibTeX file which I copy citation information into (actually, I use a frontend called KBibTeX) as I identify new papers to read. Then I can just reference it as I typeset my manuscripts in LaTeX. However, to work nicely, this relies on the fact that there are massive academic databases of papers which can provide the BibTeX citation for each one--and the fact that I use LaTeX, which I wouldn't expect of our users. Still, perhaps something like a central database of referenced works could be useful?