Taking some time off

This year would be the first time that I am not attending LinuxAsia, although given the theme of Interoperability – it would have been an interesting one to go to. In keeping with my tradition I did not attend foss.in and will also not attend Gnunify (this one would have been nice to attend) since I am in Kolkata during that period. The only one I would love to go to is Freedel. Let’s see when the organisers get that one going. Horizon organised by the Durgapur LUG is also being given a skip.
The past few years I have been doing the rounds of conferences and small meets and I see the same faces talking about the same thing and nothing much being done. Thus, in order to shrug off my cynicism, I am taking a self imposed break to do some stuff – file a few bugs, complete a few hack tasks I had set for myself and in general try level best not to fall out of the habit of coding. Over the past few months the sluggish and intermittent attempts to get back to coding has exposed how much out of practice I have become – and that’s frightening. Not all that I do would be available immediately, but I hope that with the plans that I have for Randomink, we can get some good things going.

Why I think BOSS is not a good business model…

I have blogged about¬† BOSS earlier. The issue is that C-DAC does not need a BOSS like fork (for lack of a better word let’s use fork) to monetise the development efforts that have gone into L10n. In fact a line item product oriented model like BOSS might just not cut much ice even though it has all the basic advantages of becoming the Operating System of choice for government departments. What could be more of robust a revenue stream is offering services (consultancy, deployment, support) around the corpus of work that C-DAC has under the aegis of L10n. However, to kickstart that effort, the primary objective would be to ensure that the components are widely used. To do this it would be required to do the follwoing:

  • open up the body of work under appropriate OSI compliant licenses (GPL, LGPL whatever fits the bill and is approved)
  • begin putting up the components through appropriate packaging mechanisms into upstream distributions viz, Fedora, Ubuntu, Debian, Mandriva etc

This would allow more and more users to be consumers of the products and thus enable putting in place a feedback loop. However, L10n components would lend themselves well to custom service consultancy eg creating assistive technology embedded OS or a digital library that integrates text to speech as a plugin/extension. These are the areas where C-DAC is best placed to leverage its experience and expertise. Extensive usage and distribution through upstream distributions also ensure that a completely integrated build process is put in place for accepting and implementing feature requests, patches among other things. The easiest way of doing this is putting in place the following:

  • Repository of packages for various OS versions which can be accessed by the users (yum, apt-get etc)
  • An issue tracker
  • A mailing list
  • A wiki

Working with upstream projects (GNOME, KDE, OpenOffice.org, Mozilla etc) and distributions mean that there is a healthy exchange of information and technical know-how sharing that would establish C-DAC as a true player in the FOSS space.

The core issue with doing an OS and then wrapping a business model around it is that of ensuring compatibility of kernel ABI and software API. Add to that the logistics of application certification and hardware certification and one has a full fledged task at hand. To top it all one needs to take the onus of supporting the binary/source collation for a fixed span of years. All this would be the task of a nicely populated engineering team. Services around existing competence provide a far more easier learning curve and productisation route.

Looking inwards on the Long Tail

I talked about the Long Tail of FOSS deployments. What also gets added on as metadata is the context of support. A significant percentage of the Long Tail deployments is fulfilled by small vendors. These are not the grey market assemblers, but the smaller branded assemblers who custom create the PC configuration based on what can provide maximum push to the bottomline. These smaller vendors have a good amount of interaction with institutes and businesses within their local area eg. colleges and universities who want to purchase small numbers of desktops/workstations, SOHO setups etc. Sometimes they also provide services viz setting up webservers or print servers etc. These agencies form a niche audience for distribution of pre-loaded Linux boxes and support on services. Thus, what has long been the bottleneck – having the depth of support teams to provide desktop support to deployments might be taken care of.

The next step of course is creating recipes for deployment. This would enable the standardised builds to be rolled out and which would also take the headache of supporting a myriad number of base systems off the support teams. Of course this also means that a small number of folks could do consulting on all these domains. Looked at in isolation these don’t seem like lucrative revenue streams, however looked at in the perspective of numbers – they are nice enough to get involved.These are but bits of the steps that need to be taken to ensure that we overcome the challenges that hinder FOSS from being popular. These include:

    • providing a complete experience for the home user
    • providing services and solutions for the SOHO units
    • reaching out to the maximum possible number of users through locally present vendors
    • creating FOSS skillsets and awareness where none exists

    However, these are all examples of consuming FOSS. Contribution to FOSS is the next stage of evolution

The ‘Long Tail’ of FOSS Deployments

I had been thinking about writing on this for some days now Рthe Long Tail  of FOSS deployments.

While talking about FOSS deployment numbers it is an inherent tendency to pick and choose the high profile or the largest number of units deployed. This I propose misses the significance. For what does not get counted and cherry picked is the large install base of home users, one-off edge servers (file-print-network-firewall-mail etc) and the niche deployments like servers for Media categorisation systems (note how deftly one avoids falling into the Digital Library trap). Most of these long-tail-deployments would have been done by small but not exactly mom-and-pop teams who do a good enough consulting business but are not too big to be part of the success stories. I would argue that much can be done if the established players in the FOSS scene and especially the principal vendors reached out to these consultancies through a defined programme.

There’s a feeling that most of these are fly-by-night operators who tend to cut costs. I would say they are not. Knowing which side of your bread is buttered and where the butter comes from, these teams pay more than adequate amount of attention towards creating stacks of solutions that base themselves on well tested OS internals. They don’t have classically defined release engineering and other teams so they try their utmost to ensure that their solutions are based on stable platform and the component versions are thoroughly tested and re-tested and benchmarked. Most of them use a customised base OS. Custom OS does not mean the total revamp of the OS internals. However, it does mean the core OS libraries which are required plus the application/solution. Examples of these might just be an Education Stack, a Stack for Medical Stations/Hospitals, Digital Library Stack, GIS Stack, Point-of-Sale Stack etc

Given that all of us are trying to figure out how to reach out to more and more potential consumers of FOSS based services, the Long Tail of deployments would be an extended means of outreach. What needs to be put in place is taking the members of the Long Tail to the next level of contributing in terms of bug reports and quality engineering

Another random idea

It might be nice to have a feature whereby I can drag and drop an email address into an IM area (say the Buddy list area of Gaim) and start off a conversation or at least be possible to initiate conversation if the target is not online at the moment. Does it exist yet ? Don’t know for sure.

If only IRC/Freenode mandated registration of nicks with email addresses, then it would also be possible to right click on a nick and fire up the MUA Composer window for the mail.

The stuff we sometimes get to read

Sometime back I had posted a letter to the editor of LinuxForYou.It seems that they never learn or never listen. Take the case of the article titled “OSS Adoption in India: The NRCFOSS Way!” To quote from the article:

BOSS (Bharat Operating Systems Solutions) is a Linux distribution that has been developed by CDAC for enhancing the use of FOSS in India. This product has been developed specifically targetting the Indian environment. BOSS has a pleasing desktop apart from a feature that supports Indian languages. Subsequent versions of the product are expected to support the education domain as well. However, the main goal has been to localise BOSS into all 22 official Indian languages so that IT can reach even the masses that are not conversant with English

The last time I checked nearly everybody was doing L10n work upstream either via distributions viz Fedora, Debian or Ubuntu or via projects like GNOME, KDE, OpenOffice.org, Mozilla. So the question that can be asked is: Why another distribution ? And more specifically why not contribute directly upstream ? Spending money and throwing resources to re-invent the wheel is something that we are generally getting good at while doing the necessary PR spin to make it appear like a virtuous job. Just to put a fine point on things, BOSS is actually a souped up Debian looking inanely like a Fedora – make whatever you want to make out of it.

But the issue is not with BOSS or the effort. I am kind of fuming at LFY. Where’s their sense of their importance in the scheme of things gone ? Instead of blandly and blithely reporting all this – where are the questions ? When will LFY realise that blind hagiography will end up doing more harm for FOSS in India rather than amateur reporting style ?

Random idea of the day

To begin with I changed the look of the blog to the Classic one – slightly boring but till the time one comes up with a new look this should do fine.

The random idea of the day was this – how about creating a list of words and thus their spelling forms that translators use in the .po files ? Now given the state of Unicode Normalization that might be difficult although one can look at UTF-8 Processing Library in an attempt to resolve it. Once the list of individual words (one instance of each since there would be repeat usage of the word) is created that can be used to collate and standard spelling guide and thus one area would be standardised.

Is it possible to do it ? I don’t know yet. Ramakrishna says there are too many hurdles he faced during his Entrans days – but would be sure good to have.