Chamindra de Silva

Learning Software Development the Open Source way: A student guide

2011-01-27T10:43:00.000-08:00

A book I co-authored with Professor Allen Tucker and Professor Ralph Morelli targeted mainly as a semester course for students was recently published by Chapman and Hall/CRC as part of their innovations in software engineering series and is available on Amazon titled "Software Development: An Open Source Approach"

Background to book and value to students interested in a career in IT
Participating in Open Source is a fantastic way for computer science students to get some hands on experience in a live Software Engineering project, especially before they start a professional career in Information Technology. Participating in an Open Source provides invaluable experiences that closely emulates what they have to face in the real world of professional software engineering, which a pure academic project or simulated mock project cannot offer. Additionally internships with software engineering firms often does not sometimes provide the full range of experiences needed as students are rarely permitted to participate in the critical parts of a billable client project thus their access and exposure is often restricted. Open Source projects however welcome contribution and the sky's the limit in terms of what you can contribute. You are valued more for the quality of what you can do, rather than how many years of experience you have under your belt. Participating in a global Open Source project also is not just about coding, but also about getting exposure to some of the invaluable soft skills needed to becoming a well rounded professional software engineer or architect. This includes interacting with diverse people from developers to users, documenting for those different audiences, understanding software usability, software intellectual property boundaries and learning how to promote your product and yourself. And if the student does well, it also gives them valuable credentials, referenceable experience and a global recognition that could serve to rapidly develop their career. Invaluable not just for students but even professionals who have still not got into Open Source. The book is a hands-on guide for you to get involved and becoming a valuable part of a Open Source community. Hope you enjoy it and do send us feedback for our next edition.

Proprietary can learn from open source licensing world, especially for the cloud

2010-11-21T14:13:00.000-08:00

Proprietary licenses have evolved greatly since it was initially forged by IBM and Microsoft many years back. Every proprietary vendor has their own unique license and the royalty model can come in many different forms. Revenue for the proprietary vendor can be licensed proportional to the number of users, nodes, processors, bandwidth consumed or an annual subscription model. Furthermore each vendor has their diverse models of discounting that can result the licensee in a complete state of confusion if he does not seek proper legal advice.

Unfortunately software licenses lives in the nexus between a legal document and technical jargon, which often renders either profession incapable of deciphering the actual right of the licensee. There have been times when I have found that even the sales team of the vendor is somewhat unclear. Standardising the licensing models is certainly something the proprietary world can learn from the Open Source and Open Content world. Over 50% of the Open Source projects fall under one license, the GPL and the other common one Open Source ones LGPL, Apache, BSD are used across at least thousands of Open Source projects. With such pervasive use, this greatly simplifies the problem of understanding licenses. Compound with that is the most Open Source licenses are kept extremely simple. The BSD license is but one paragraph that would hardly take a quarter of a document.

Unfortunately, if the proprietary world does not move into standardising their license models, they are soon going to be left behind as more applications get deployed in the cloud, with providers like Amazon and Rackspace. This is because the cloud model is on-demand model. You pay as you use and the cloud decides how many resources are needed to serve you need. This might mean it needs more processors, nodes and users per node and this need to happen flexibly and dynamically for it to work. As there is so much diversity in the proprietary licenses this becomes very complex as applications are often build from multiple components. If each component has a different revenue model, it becomes very hard to meter the total usage to charge the client in turn. Open Source licenses do not suffer this fate as such limitations are against Open Source principles of freedom and thus you have any number of users, processors, bandwidth usage, nodes all at the same base cost for software, zero.

So my recommendation to the proprietary world is simply to standardise and reduce the complexity of licensing such that more focus can be given to the value of the product or risk being left behind in the cloud world.

Published In:

OSI Days, India and Article on the Open Source License Compatibility for the cloud

2010-11-21T14:03:00.000-08:00

Recently wrote an article based on my comments at a panel discussion at the OSIDays conference in India. It was also published in the Financial Times. I welcome comments (next blog post)

OSI Days was held in Chennai, India and is an evolution of Linux Asia, re-branded.
Picture is of a panel discussion I participated in on programming languages

The conference schedule was done based on input and it seems that PHP is still the most popular language of choice especially in Chennai.

Panel Discussion on Open Source for the Cloud with some of the Open Source Business leaders in India

This picture is with the two project leaders of the two most popular PHP frameworks, Zend Framework and Symphony

Picture with leading Drupal developers and the author of Adminer and PHPMyAdmin alternative.

Emergency Information Systems Interop Workshop for better preparedness @ ISCRAM 2011

2010-10-20T06:34:00.000-07:00

I would like to announce that we are organizing an emergency/disaster information systems interop workshop for the next ISCRAM 2011. The interop workshop will be an environment where software vendors and information systems solution providers can bring in their systems for testing against popular information exchange standards such as PFIF, CAP, GeoRSS, EDXL Standards, GIS Standards, etc. The structure will be based on the Apache interop workshops. More importantly we will want to test, improve and validate information exchange between systems for certain popular pre-defined use-cases utilizing these standards. Thereby the results of this workshop will serve better preparedness by assuring that the systems that participate are able to exchange data effectively and work more efficiently in partnership at the time of disaster response.

The interop workshop will be part of the information systems interop standards track, where you can also submit research papers, work in progress papers, practitioner presentations covering best practices and demos for inclusion in ISCRAM. We are also welcoming presentations on new, existing and emerging interop standards for educational purposes to help improve awareness, facilitate feedback and adoption.

Participants interested in participating for the interop workshop can submit their solutions/products/demos/

functional prototypes as a demo for this track.

Further submission details can be found at:
http://iscram2011.lnec.pt/pdf/tracks/information_systems_interop_standards.pdf

For information of the ISCRAM 2011 conference the main website is:
http://iscram2011.lnec.pt/

We look forward to your participation to this much needed exercise. If you have any suggestion on this track please do not hesitate to contact myself, Renato Iannella or Tom de Groeve. We also want to form a steering committee for the interop workshop, so do let us know if you would like to join that.

Why Virtusa supports QA teams volunteering on OLPC?

2010-08-03T05:15:00.000-07:00

Virtusa has been applying and contributing to Open Source R&D for quite a long time now, from Apache contributions on Web Services, to the Sahana Disaster Management project in the wake of the Tsunami and most recently to OLPC. In the case of OLPC through with our experience with Open Source, we realized that there is much opportunity to contribute to Quality Assurance (QA) as most often Open Source volunteers are motivated rather by the research and development side of the project and there often is not enough focus on QA. Yet projects like OLPC and Sahana have a global impact and in the latter case needs to be mission critical, thus the quality and stability of the system should be a very important part of the project.

Why OLPC? Well despite the challenges OLPC foundation has had based on their policy for deployment (e.g arrangements only with Govs and not with the private sector for deployment), their mission to empower children is certainly very honorable and something that should be supported by all of us to help bridge the digital divide. It also has to be noted that some people have mis-understood the OLPC. It is supposed to be a tool that will supplement (and not replace) existing education systems and empower especially children in rural communities, who otherwise would not have access to IT or IT teachers for learning, enabling them to learn for themselves. OLPC also has had a much broader indirect impact, as it has been a flagship product that greatly helped bring about the netbook revolution and simply the existence of the OLPC and it’s $100 target has certainly helped drive down costs and have got people thinking about other low-cost solutions for educating children in rural communities. This competition is healthy and it will certainly progress further with the upcoming releases of the OLPC 1.5 and OLPC 2.0 (touch based, iPad like laptop), embodying a lot of lessons from the deployment of the first OLPC 1.0s.

Overall Virtusans volunteers have spent about 40 man months on the project so far delivering about 800 test cases. A good deal of time was spent learning how the system is supposed to work, especially as Open Source projects typically do not have well defined requirement specifications and Use Cases, which are normally used by our teams to derive test cases in client projects. But now we do have sufficient knowledge to quickly nurture new contributors and we presently have volunteers in India and Sri Lanka contributing off our spare QA capacity on the project. One area we are specifically looking at now is test automation on the Redhat based sugar operating system as a mechanism for providing more efficiency for testing new builds.

Virtusa and Virtusans will continue to help OLPC achieve their goals by supporting the team with QA contributions as part of our Corporate Social Responsibility in a initiative we call Tech Reach.

Related Articles:

CSR Wire
http://www.csrwire.com/press_releases/30163--Virtusa-supports-One-Laptop-per-Child-Program
Andhra Business
http://andhrabusiness.com/NewsDesc.aspx?NewsId=Virtusa-to-back-one-laptop-per-child-programme.html
ITVarNews
http://www.itvarnews.net/news/11238/Virtusa-Supports-One-Laptop-per-Child-Program.html
ITPro
http://www.itpro.lk/node/2058

Idea: Building in mobile phone beacon technology to help search and rescue operations

2010-08-01T07:08:00.000-07:00

Telcoms infrastructure is often down or saturated during a disaster, however even if a mobile device cannot be used to dial a call or send an SMS, they inherently have another capability. Mobile devices are effectively radio transmitters that with very little modification should be able to act as locating beacons to locate trapped people in the locality of say about within a 200-500m radius.

Mobile phones are quite pervasive now with high penetration even in developing nations . These days most people carry a mobile phone wherever they go. During my visit to Taiwan, I got an opportunity to speak to a leading OEM chip chip manufacturer, who’s chip functionally ends up in a great amount of phones today and they informed that this very doable.

Please do not confuse this with locating mobile phone using cell towner triangulation techniques as that lacks the granularity to locate people in for example a trapped building block. With this you should be able to say the number of phone in for example a collapsed building and make a guess at the amount of people trapped. The natural attenuation of the signal also should help build a suitable “metal detector” like tool to direct a search and rescue operation to the mobile phone and the trapped people or bodies with them.

You might also counter saying that GPS technology can be a better locator, however GPS technology requires a clear line of sight and even to SMS their location it all requires active input by a potentially unconscious person and a level of literacy on mobile functionality beyond the basics. If anything the only input required would be to say someone is OK, so that their signal no longer contributes to the noise of multitudes of signals of the number of people with mobile phone in say a 500m radius. Or they might annotate their signal with a request for urgent help or with other information if they so wish, but that would be optional. One issue is battery consumption as most of the time mobile phones are not transmitting, but are passive receivers which require consumes far less battery power. Transmitting a beacon will have to be done in very energy efficiently way. Like a periodic beep and powered for say a maximum range of 500m.

Yes you might worry about privacy or abuse in non-disaster times, in people being able to locate you, and there should be a way for the user to turn it off when needed or it is something that is turned on only on a cell broadcast by a tower. Detectors, which would be specialist directional devices that can locate people with such granularity has to be restricted to emergency use though legislation.

Additionally in more advanced phones you might be able to use it as simple short range walkie talkie to permit those that are trapped to communicate with the rescuers and broadcast a help message.

I am going to follow up on this possibility, which I thing will have a significant impact for search and rescue. Thoughts?

UN APCICT/ESCAP Publication on ICT for Disaster Risk Reduction includes Sahana

2010-07-04T23:57:00.000-07:00

The recent UN APCICT/ESCAP Case Studies on ICT for Disaster Risk Reduction includes the Sahana project focusing on the experiences deploying Sahana for the Haiti Response. It includes a lot of other case studies and best practices in different countries in the Asia Pacific Region and is a worthwhile read. You can find the publication on the UNAPCICT website at http://www.unapcict.org/ecohub/ict-for-disaster-risk-reduction-1. I co-authored this case study with Mark Prutsalis. Thank you to APDIP for continuing to recognize Sahana contributions in this sector.

What makes a truly Open Standard?

2010-06-30T10:59:00.000-07:00

As far as I am concerned, for a standard to be truly open it should have the following going for it in order of priority.

A liberally licensed specification - Not all standards are free and some require royalties to obtain. Licenses should be aligned to the Free and Open Source principles.
An Open Community Process - Not all standards are defined and refined in an open community process. Some require expensive exclusive memberships before you can participate, which limits the participation by organizations / individuals in poorer developing nations.
An Open Source implementation - This is not a must, but it certainly helps assure that the standard has been implemented transparently. It also help propagate the standard as any one can adopt these libraries into their products. TCP/IP, is a good example of the success of a standard through this model.

However we need to work with the reality of the proprietary nature of most standards and give priority to the ones that have greater adoption or we would not be serving the critical need for having a standard in the first place.

Virtusa donates to Rehabilitation Efforts

2010-04-12T02:39:00.000-07:00

I have been involved in a Virtusa CSR initiative to support the rehabilitation efforts in Sri Lanka for the Commissioner General Rehabilitation's Office. There are a lot of rehabilitates who are interested in a vocation in ITES/BPO and thus Virtusa was invited to help guide a program that would help meet those aspirations. Virtusa decided to make a significant donation at this critical post-civil war period to help the rehabilitation efforts. The donation included a computer lab with 30 computers (running Ubuntu) in Vavuniya, support for defining a program for those interested in a vocation in ITES/BPO and the development of a software solution to help better manage the training and rehabilitation efforts. The lab was opened on the 1st week of April by the President of Sri Lanka, Mahinda Rajapakse and it was a great honor to be part of that ceremony.

A further1300+ ex-combatants were released that day to their families and it was a heartwarming sight to see those who have suffered so much due to the civil war being reunited with their families with a fresh outlook for a peaceful future. Some of these kids are so talented and it is sad to see that they wasted so much of their lives being trained instead as lethal weapons. We hope they will have a better future in peaceful Sri Lanka.

The event also gave me another opportunity to observe the President quite close and I continue to be impressed with his charisma and down-to-earth nature with people, especially in this environment, where he was surrounded by ex-combatants, who not too long ago would have taken every opportunity to do him harm (as mandated by their previous commanders).

I was also very proud of our Virtusa team and leadership that gave their full backing for this initiative, given it's relevance to peace building post-civil war. I also had to opportunity to work with the Sri Lanka Army these past few months (from end of last year) on this initiative and I consistently found that everyone I met was very professional, ethical in their approach and disciplined on delivery. It was a pleasure and a honor to work with them and we would not have been able to contribute on our CSR goals without their vision for rehabilitation and encouraging support of our contributions.

There is still a lot of work to do and you still find remnants of the war scattered around, yet you can see the progress being made on the ground. We hope and pray for a peaceful and prosperous Sri Lanka of the future.

The UN WFP Food Request Portal based on Sahana goes live

2010-03-20T22:08:00.000-07:00

The UN World Food Program Food Request Portal based on the new SahanaPy framework went live last week and I hope it will serve to bring efficiencies to food distribution between WFP and their Implementation Partners in Haiti. This was however yet another instance where we had to rapidly change Sahana and build a totally new module for a live requirement and it certainly was not easy given the availability of most of our people is part-time. As part of this deployment hover we have learned a lot of new best practices that we should apply in the future.

Ushahidi and Sahana will work closely together in disaster response

2010-03-20T21:19:00.000-07:00

I had an opportunity to meet Juliana Rotich, Co-Founder and Program Director of Ushahidi during the III Congress in Spain and we had a great discussion on how Sahana and Ushahidi could collaborate more moving forward building from our strengths and partnering on disaster response. We had been collaborating recently during the Haiti Response. The way Ushahidi and Sahana organizes itself as projects are very similar based on Free and Open Source Source principles and we also even share the technology stack (PHP), thus integration would be quite easy. We have also arranged for our development teams to meet for an integration hackerton. Juliana also became a good friend and here we posing for that "formal" landmark handshake between Ushahidi and Sahana to pledge work closely together in disaster response.:-)

Speaker and RT participant at III International Congress, Spain

2010-03-20T21:07:00.000-07:00

I had the opportunity to represent Sahana and Sri Lanka at III International Congress event in Gijon, Spain. The event was about building better corporation between “north” and “south” countries and for learning about the innovation in ICT that was coming up from the south. The audience was represented by policy makers in Europe and the UN and innovators from Africa and Asia. I was invited to speak as well as be a round table participant and I took that opportunity to talk about Sahana, H-FOSS and also brought some awareness to the ICT industry in South Asia. In the Round Table discussions keep points I mentioned was that innovation from the south too often get’s tagged with an ICT4D banner for novelty and this hides it’s true potential to be applicable as an globally applicable innovation in its own right. I also am not a fan of the "North/South" term and any demarcation that creates artificial barriers for collaboration. In cyberspace at least we are all equal and social networking is working to eliminate such divides as it no longer matter where you live. Overall I met some great people and made good friends amongst the participants and speakers and look forward to further collaboration with them. Notable for Sahana we made some good links to collaborate with Ushahadi, TachicalTech and UNDP.

References:

III Congress Coop 2.0 Event: http://encuentro2010.fundacionctic.org/en/ponentes/
Presentation: http://www.slideshare.net/cooperacionarroba/evolution-of-the-sahana-open-source-project
Newspaper article: http://www.lne.es/gijon/2010/03/02/creamos-software-libre-sri-lanka-gestion-desastres-tsunami/880526.html
Newspaper article 2: http://www.europapress.es/asturias/innova-00297/noticia-innova-fundacion-sahana-presentara-cooperacion-20-herramienta-gestion-ayuda-desastres-20100123120539.html
Newspaper article 3: http://mas.lne.es/siasturias/etiquetas/iii-congreso-internacional-de-tic-en-cooperacion-al-desarrollo/

10 questions to ask when selecting open source products for your enterprise

2010-01-03T06:39:00.000-08:00

Was recently invited to write an article to TechRepublic on their "10 things" articles on 10 questions to ask when selecting open source products for your enterprise. This is also based on the research I am doing at Virtusa on Open Source maturity models and governance. I wrote this as a brief guide for enterprises when selecting Open Source products and it basically covers 10 checks you should make on an Open Source project to better assure you are realizing the benefits that have been promoted by Open Source. By no means is it supposed to be comprehensive, but here are my personal tips as somewhat of an Open Source insider on how you can select between suitable Open Source products.

Ref: TechRepublic

ApacheCon Asia 2009 Roadshow was a success!

2009-12-16T15:45:00.000-08:00

The ApacheCon Asia Roadshow which was a dual event this time between China and Sri Lanka went off quite well. We had about 200 participants for the two days, which was followed by a unconference on the beach! Prof Mohan Munasinghe, a Noble Peace price winner gave a compelling talk on climate change and I had an opportunity to chat with him after. Also we had some facinating conversations with the current president of Apache Foundation, Justin Erenkrantz and long standing director and x-Chairman, Greg Stein. Virtusa was sponsor for this event and my role was as a moderator of the panel discussion, apart from being in the organizing team. Checkout the pics.

Sri Lankan President Launches OLPC program

2009-12-10T16:54:00.000-08:00

I was invited to the Sri Lankan President's residence at temple trees today for the launch of the OLPC program in Sri Lankan. 1300 laptops are to be distributed to 13 rural schools. A lot of partners were involved in getting this off the ground including many members of the FOSS community. Virtusa specifically sponsored the server machines for the schools as part of their CSR "digital reach" effort. We wish the Ministry of Education, who is at the center of this effort the best success as it certainly will do wonders for kids in rural areas. I remember the amount of keen curiosity and interest the kids had when we did our field tests in the schools. Apart from that, it was an great opportunity to see the president up close, as he was sitting just two chairs just in front of me. I am generally a fan of our president as whatever said and done, he lead this country out of a devastating civil war, the magnitude of problem, militarily, economically, socially and politically (particularly international politics), which no other president before had to face or ever had firm a resolve to overcome. All our kids have a chance for a future in this country because of the end of this war, that deteriorated our nation for decades. So the president can rest on his laurels as far as I am concerned at least for now, as he delivered on this important promise to his fellow Citizens only just few months ago. Now it is time for all of us to contribute to heal the wounds left behind after the civil war and ensure that the reason for the civil war does not rise again.

Ref Article: http://blog.laptop.org/2009/12/10/sri-lankan-president-launches-a-national-xo-program/

Sahana at Relief 02 at Camp Roberts, California

2009-12-03T15:47:00.000-08:00

We had a fantastic two weeks at Camp Roberts and accomplished a great deal for Sahana. This included the use of a $400 (two years ago price) eeePC/Netbook as our primary data collection server, pulling WMS layers from mosaiced and geo-referenced UAV and satellite imagery of the Camp Roberts experiment site, configured an SMS gateway running on a Windows-based server using Cygwin and SMSTools and a Nokia 3220 phone to send and receive SMS messages from Sahana, an Android application to send in structured SMS messages to Sahana with embedded GPS coordinates, exporting of a KML feed from Sahana to Google earth to name a few of the new functionality we added.

The new SMS capabilities which I worked on gives Sahana the ability to serve as an incredibly powerful crowdsourcing and disaster situation awareness application. New SMS functionality included:

A new format for a structured SMS message to be sent from any cell phone to Sahana.
The ability to register a user name to a cellphone number.
The mapping of the SMS messages to a DHS symbolset of incident information based on feature class reported by the Android and SMS message.
We developed the ability to poll the Sahana server and pull information about the last known location of a registered Sahana user
Or the search of the last know report based on a keyword

This all was applied in an integrated field experiment and was also present to some key representatives from organizations such as FEMA and DHS

You can find the full details and pictures on Mark's talksahana post here:

http://talksahana.com/?p=303

Moved homepage after Google stopped googlepages.com

2009-12-01T14:56:00.000-08:00

Decided to move my homepage to blogger.com since google decided to discontinue the googlepages.com hosting and have migrated everything to their google sites. Also this might be a good opportunity for me to finally get into a bit of blogging, which I have been keeping away from more becuase I prefer to post my thoughts and ideas on mailing lists, where it would be more visible and open to discussion. Let's give this a try :-)

Sahana Phase II / H-FOSS Design Principles

2009-03-14T01:33:00.000-07:00

It is time to move on from the good ol Sahana Phase II framework that has served us and been used in disasters for 4-5 years now. I would thus like to take this opportunity to capture the key design principles that I would like to see being applied in the next generation of applications in this domain. Sahana Phase II was built to work in the following constrained environment, we found time after time during disaster response:

Telecoms and Internet access is either down or intermittently available.
Bandwidth is often at a premium so every character counts
Power can go out at time or not be availability
Any central data center or infrastructure might have been affected by the disaster
People have little time to get familiar with new systems
Off the shelf systems have often to be customized for the requirement or risk not capturing aspects of gathered data
Local developers have very little time to learn and support the system
There are many existing legacy systems and loads of spreadsheets with valuable data
Data will come to you with different levels of granularity, validity, redundancy. It often needs to be cleaned up.

Here are the key design principles, I had when building the Sahana P2 framework. Some of them you might find a bit odd and even contrary to popular practices building enterprise applications. However it is exactly for these reasons that large enterprise grade applications fail during disaster response. Here are the key ones:

The System has to work in a low bandwidth high latency environment

Even the most developed nations will find themselves in this environment because immediately post-disaster the networks are saturated with people trying to get through to each other. This has been where a lot of the more complex systems fail at a time of disaster, as they are often build to work at a time when internet throughput is a given. Sahana P2 spits out a very lean XHTML structure with all the visual look and feel coming from static CSS stylesheets that is downloaded and cached only once. This has two corollaries:

Heavy AJAX or RIA can be a showstopper

Rich internet applications that use heavy AJAX calls and frequently go back and forth will inevitable have issues in this environment. For example a simple AJAX pop-up due to latency will take much longer to appear and as a result the user might think the system has moved on, when it hasn't. If you are including AJAX based functionality make it work in such a way that it improve usability but is not a dependency for the completion of an action. The exception to this rule is mapping and that is the only one as it is something you cannot do without. Sahana Phase II avoided heavy AJAX in favor of simple HTML forms for this reason.

Let the developers be in close touch with what is going out on the wire

Developers can go off in a tangent in terms of functionality and forget about how much their implementations is costing in terms of throughput. A lot can be said for abstraction, but the more a developer is in close contact with the HTML output and character cost of his implementation the better. Sahana P2 let developers stick to PHP without much abstraction except in the form of template helper if needed and Web2Py/SahanaPy has a templating system that should be used more favorably when building applications.

Design Uncluttered Intuitive Interfaces:

The system was build to have a clean, uncluttered and simplistic interface which a responder could understand intuitively. This is important because often we do not have time to train new users immediately post disaster. More than most system a lot of help text was provided on the home pages and on the modules and additional help text and tool tips are provided to help the new user get used to the system. As a result there was a lot of good feedback on Sahana Phase 2 interface from users who found the system very intuitive and easy to learn.

Plugin architecture and a focus on automation

Sahana P2 had a lot of dynamically pluggable namespace automation both at directory, file and function level. All you had to do was drop in a new file with a specific name (main.inc, admin.inc) for the framework to recognize there was a new module and it's admin page. A lot of things were automated, greatly reducing the learning curve for an administrator. To move out a module all you had to do was move the directory. So there was very little in terms of configuration files that needed to be written. This is in contrast to what I feel is badly practiced in the Java world of needing multiple complex XML configuration files to get something to work.

KISS Design and Learning Curve for Local Developers

The people that would often be most committed to supporting a system post-disaster and will be in it for the long run would be the local developers. As a result a system needs to be easy to understand and customize. PHP and Python as scripting languages are some of the easiest languages to learn, though I might give the edge in terms of ease of use to PHP as it is focused as a web development language. In Sahana P2 we also did not add a whole new layer of learning for someone to create a module. A module writer could use vanilla PHP to build his module and use the template helpers only if needed. If he wanted to escape the framework in certain ways he could use function overrides to completely change the layout. There were many instances where we found during multiple disaster, even during Haiti, Sahana P2 downloaded and customized heavily independently without any help from the community developers. This was a testament to this design principle.

No Complex Workflows and Security please

Avoid introducing complex workflows and security and creating dependencies on people to approve. Just post a disaster most people are overwhelmed and if a system dictates that person X should approve person Y before he can do action Z, most likely person X will become a bottleneck or will end up wasting his time with having to monitor approvals. Rather build systems that are open by default and have minimal workflows if at all needed to deliver results. The risk of not having access to good information is more of a problem here than the risk of unauthorized access. This is why by default Sahana P2 comes without security enabled. However if you do need it possibly later on in the relief phase, you can enable it at any point and customize it to your hearts content.

Portability and Data Synchronization:

Sahana was made available as a portable application, and thus moving the program with it's data is simply a matter of copying a self-contained directory to a USB disk. This has the added benefit of being pre-configured click and play instance of Sahana and an ideal way for gathering data from the field and synchronizing it to a central machine, when there is no internet access in the field. The data import and export functionality also performed an important function of importing data from spreadsheets.

Depend on the OS and Database for your scalability

Sahana Phase II depended more on the proven LAMP stack for it's scalability. PHP by default is stateless and make it very hard for you to introduce state. Every new call is an new instantiation of the Sahana application. Whilst this can seem very inefficient, it is an architecture that was designed that way to be horizontally scalable. This stack is actually one of the most popular stack for horizontally scaling large portals and is tweaked further and used by organizations such as yahoo or google, so we knew we had little risk of having problems with scalability in P2 that could not be resolved.

Resilient easily debuggable Architecture and Database:

Tensions are already high in the middle of a disaster that you cannot afford to depend on systems that have little tolerance to race conditions or dependancy error. Thus flatten your architecture as much as possible and actively reduce single points of failure. If there is error then the architecture should make it easy to find. Unfortunately this is a problem of encapsulation and heavy abstraction, that it make finding the real error harder to find. Maybe I am biased, but with Sahana P2 finding and resolving issues was quite trivial (again as we did not deviate much from vanilla PHP)

Permit an easily readable and directly modifiable database

A lot of people say that you should not modify the database directly, but actually during the nature of data post disaster is such you have to do a lot of data clean up and if for some reason you have data corruption, you do need to be able to look and modify your database. The database scheme thus should also be accessible. This was one reason that some references is Sahana always used 3 letter abbreviations (rather than a number), to make it easy to look at one table and know what classes that entity was associated with. We wanted to strive to make it easy to understand the database by looking at it, rather than creating a new abstraction layer that needed to be learned and updated. Also when the functionality does not exist, there can be a lot of customization done through the database. One example is the bulk data imports and clean up of data that needs to be performed. Thus keep the database as stateless as possible and easy to understand and keep the database schema simple so that it can easily be tweaked if needed.

Well there are other things, but I think these are the key ones. However though these were the principles I encouraged as of late the recent improvements to Phase II have in certain areas not kept with them, so I think it really is an opportune moment for a new framework. In hindsight here are some of the things I would like to see in the new framework, beyond what I mentioned above. And for the record we did know a lot about the MVC pattern and OO before we started phase II, but choose rather not to use it to keep the framework lightweight. MVC and OO was rather left as a choice the module writer could make, however possibly to avoid confusion what I could have done was create an MVC layer that could have been used optionally.

This to add in a next gen DM framework

Optional Lightweight MVC based framework (with low learning curve).
Automated database abstractions (model OO) – reverse compiled from the data schema
Automatic CRUD/REST functionality (views), but can be overridden and does not trap you. Beware of CRUD as it is as old as MS Access. It will get you 80% of the way really fast, but the remaining 20% will make it hell and require you to break it, so you need to know when to apply it.
Easy database migrations for schema changes (support for multiple schemas for single entity).
Inherent support for importing and working spreadsheets
Inherent support for datatables and exporting data in common.
Inherent support for data versioning, credentialing and logging
True Disaster or Sub-Disaster Event multi-tenancy