Several days ago we stumbled upon interesting problem to solve “Abstract enough User permissions & roles model so that it can sit on top of everything within a enterprise system”. There are several implementations already in java, php, C# and etc. unfortunatelly there was missing something required in our special case: dynamically added “targets”, ie today we have user permissions upon ‘files’, but tomorrow we need them for ‘projects’ as well.

So simple as it is:

Notes

Access rights, user permissions and etc. is always related to a target as my colleague Simeon said while we’ve been discussing the above problem’s crafting.
So the solution is better described as sentence

Who has what kind of permissions upon which target data/action/tool/feature & etc…

The same structure above can be even cleanly represented as this object graph/map:

The above I suppose can be well implemented in any kind of DB and optimized to work perfectly, however I would use graph-like DB (neo4j) if I go for a stable/scalable solution at first place.

I’m mighty pleased with wordpress… It tends towards better UX by evolution done right. I’ve been able to check 5 free public available themes (one with several custom options) within 5 minutes.

And here is it obi.wizartworx.com with a new theme… Still polishing some bits & peaces of the theme, but it is wonderful well done job by ravisama

Just wanted to log down the time and hour of seeing how far wordpress had gone so far… In the beginning it didn’t updated automatically at all, today it gives you a feature to browse/preview and to try hundreds of themes… And that is not “wordpress” success only, it is to the ‘crowd’ itself.

Yesterday while I was going to do some core development for the techbrowser guess what happened. Github went down. This caused me very little to be worry as git is decentralized and I was able to continue development. However I’ve been curious to see what is happening all around the world with github’s outage.

1 Hour later I was still clueless what is the public reaction except that when I opened a twitter search on “github” I was flooded with various tweets all around one thing -> “github’s outage”.

After 30 mins of searching I found several solutions:

• http://trendistic.com/ – is cool but not realtime enough.
• http://mashable.com/2009/04/04/twitter-trends/ – has a nice collection of tools but unfortunately none of them feels like the one.

So I decided to mockup one of my own:
twitter trended search chart

Source code is of course on github

The solution is as simple as this:
1) on regular intervals (currently on every 1s) a request is made upon twitter search
2) returned results count is captured as Y axis value and the elapsed time from the initial submit as X axis value. All other requests been made to twitter search use “since_id” attribute to fetch only the new coming results on first page, where if the last since_id is equal to the current received one: it is assumed that the incoming results count is a new hit.

Based on the above mockup it is fairly easy to track & monitor twitter trend in realtime. I’ve tested the web app during the github outage and it prooved to be useful (there was a peak of around 60 tweets at that time)

However there is still something which should be kept in mind – the trending search chart is a proof-of-concept, not real life utility

PS. I would like to point that the mockup is build using componentjs which proved once again that using component based development in html/js is possible and highly effective.

**Update**
I’ve found one other service providing at least charts of the activity in twitter per trend, however it is still not realtime:

• http://www.crowdeye.com/

Would like to see at least one of the services targeting twitter as platform to implement realtime trend activity charts. TweetDeck for example

Today:

• clumsy looking SimpleChrome Extenstion for in browser based adding but without the searching UI

• clumsy looking SimpleUI for web based adding and searching.

• Decentralized logic architecture but centralized database location for the tags <-> links relations
• Tags autocomplete support
• Tags auto identify support when adding links
• open source

Tomorrow*:

• Realtime stream via twitter (streaming as @devsunited until more appropiate account is choosen)
• Realtime stream via XMPP (should be invited at XMPP servers remotely first)
• Technologies/links dependency system (still in construction…)
• Technologies genome view (realtime graph-like map of technologies and their dependencies, similar to http://geni.com & http://familyone.de but for technologies)

Tomorrow is ironically used. I’m fully aware of the scope for the above statements as well as that contributions,ideas and united work are more than welcome.

Use twill + twillRest extension.

Here is how single RESTful test looks like using twillRest:
assertGetContains something http://domain/user/1
assertPostContains something http://domain/user/1 field1=value1&field2=value2
Features:

• Quickly write down functional tests
• Add support for state maintaining via custom response parsers
• based on twill (and thus supports twill-fork)

There are of course others functional and unit test frameworks for python but as I needed something simple and just working – twill looks like the one. Once twillRest is upgraded with true REST client like (http://code.google.com/p/python-rest-client2/) it will became even more usable (alghtough I’ve used it in production already)