http://logic10.tumblr.com/

#IndiaBlocks - We don't need SOPA / PIPA for censorship

Thu, 06 Sep 2012 11:26:00 +0530

India doesn’t need SOPA / PIPA for censorship. It can block anything on web. There is supposedly a draconian IT Rules 2011 amendment, but even that is not required / followed for blocking anything the government does not like. They don’t need SOPA / PIPA / even follow their own legislated IT Rules to block stuff on web. Yes, we are cool like that.

For the first time, Wikipedia article gets blocked through a DoT order. http://en.wikipedia.org/wiki/Persecution_of_Muslims_in_Burma is not accessible if you are in India. Reading 2012_Assam_violence and Internet_Censorship_in_India will give some background.

Both the Supreme Court and the Delhi High Court have rejected petitions to unblock the content. Internet Freedom is progressing in a worrying direction

Translating and Editing SVG

Tue, 07 Aug 2012 10:12:28 +0530

Picture is worth 1000 words. Illustrations are key in conveying the idea in a simple visual manner. But editing them puts in a entry barrier of knowing to use the tools to edit them even if it’s a spelling correction / minor update. Translating them into other languages was even more a pain. It won’t be the case anymore once the User:Jarry1250’s TranslateSvg extension gets deployed after final touches. It is currently available at http://translatesvg.wmflabs.org/wiki/Main_Page for beta testing. It nicely sits on top of the Translate extension, so one gets all the nice features like translate suggestions from the translation memory among others.

Another feature that will be very useful(not yet available) in improving editors ability to use these illustrations effectively will be Online Cropping / Editing of SVG. User:Planemad did an amazing map of Indian National Highways. But for individual highways one has to crop the particular segment and recreate though doing this manually takes a lot of time and effort. A lot of geo-visualisation illustrations can be easily made if there is a tool which is as simple as TranslateSvg that can make these.

Mobile Input methods

Fri, 03 Aug 2012 12:03:29 +0530

@Yuvipanda and me were discussing about available Tamil input schemes for Android. Until ICS Tamil had no support in android and pre-ICS phones had very limited number of languages supported. Jelly bean added some more languages, but input methods, complete language support even for those limited languages have a long way to go. Even though android doesn’t provide these tools out of the box, developers passionate about language have also written apps like TamilVisai which is the popular opensource input method editor available for Tamil.

Although these apps let people type, some of these are sub optimal for a seemless experience. This app for instance lets users input into its own preview box and then to the text area in the interface. This will defeat auto-complete for example. But this app was developed couple of years back when there was no rendering support, so the preview box made sense back then.

With Mobile page views of Wikipedia growing at a rapid rate across languages, it is essential to have good input method tools for all languages to increase them further because the current page views are mostly through clicking link and not really searching. This will also help in enabling mobile contributions.

While there can better layouts for touch and it need not always be based on QWERTY unlike desktops, but to people who come from desktop, having these layout will not hurt. Extension:Narayam supports over 30 different keyboard layouts and creating an app which can use the rules for the layouts will make mobile input a lot easier on Android and of course more layouts can be added.

Localisation Tools @ Wikimedia

Thu, 02 Aug 2012 09:27:00 +0530

As part of my role at Wikimedia Foundation, I shall be posting about things related to localisation / internationalisation efforts at Wikimedia Foundation Localization team and things generally related to Wikipedia / i18n.

For those who are unaware of the way it works, here is a short intro.

Wikimedia projects exist in 280+ langauges.
The software Mediawiki is localized at translatewiki.net which itself is a wiki powered by Translate extension
Similar to the large army of volunteers who contribute content to Wiki projects, there is also a small army (~5000+) of volunteer translators in all these 280+ languages.

The localisation team develops / maintains extensions / additional piece of software which help users of these language versions of the Wikimedia projects read / contribute to them by providing tools which assist them.

Extension:WebFonts delivers fonts from the server through the browser to viewers and does not mandate user’s system should have the fonts installed to view the pages.
Extension:Narayam provides an input method solution for languages which do not have their own physical keyboards (primarily used for non latin scripts).
Extension:LocalisationUpdate, Extension:CLDR, Extension:TranslationNotifications, Extension:Babel are some of the other behind the scenes infrastructure which assist localisation.
Fixing bugs to support complex scripts / Right to left languages in the all of wikimedia software
Universal_Language_Selector - a new tool which assists users in switching interface language and use language tools is currently being built.

I will keep posting more about these in detail in the days to come.

SHRI/SRI ? -- ஸ்ரீ = ஶ் + ரீ != ஸ் + ரீ - Unicode

Sat, 30 Jun 2012 19:02:00 +0530

ஶ் - An introduction

For those of you who are not aware of the existance of this character, it is a grantha character SHA normally used only in sanskrit texts written in Tamil. Unlike ஸ, ஷ, க்ஷ, ஜ and ஹ. ஶ made a late entry into Unicode. ஶ however was very much alive in print form particularly in Hindu religious texts in Tamil for a long time now.

Code point, Glyph :

Let us first understand how Tamil characters are treated in unicode. All Uyirmei characters are considered as “complex glyph“ and are stored with consequent individual elements code point. For instance க் gets stored as க, followed by ்(pulli/Virama). When consecutive individual elements come in the text, the font chooses to display a single complex glyph. This is also the reason why using the standard character length of Indic text will be longer than what it is! Read Gerard’s blogpost twitter-in-chinese-or-in-tamil

ஸ்ரீ = ஸ் + ரீ to ஶ் + ரீ:

Until Unicode 4.1, ஶ did not have a codepoint in Unicode. So ஸ்ரீ was considered as a complex glyph of the form ஸ் and ரீ. Later ஸ்ரீ was considered to be ஶ் + ரீ. Linguistically if one is “strict”, it shall be right to call ஸ்ரீ as ஶ் + ரீ as that was the intended pronunciation. One can even see how the glyph’s ஶ் and ஸ்ரீ are slightly related in shape. But if one accepts that language evolves (read previous post on Sri’s cultural aspect) and the fact that ஶ never made into mainstream except for religious texts, ஸ்ரீ = ஸ் + ரீ can be accepted. Kaa.Sethu wrote other usecases non-conjunct form of ஸ்ரீ. Seeing that the definition change could well be a good thing.

Dual encoding / re-coding ஸ்ரீ:

I am not sure why Unicode did not think of potential recoding text when they changed the definition of ஸ்ரீ = ஸ் + ரீ to ஶ் + ரீ. May be they thought Tamil Unicode was in its infancy and not much text would be there. The fact that am writing this blogpost after 7 years of Unicode 4.1 is saddening. Most of the popular Unicode fonts are pre 4.1 and do not have glyph for ஶ். So they still use the old definition of ஸ்ரீ = ஸ் + ரீ. The input method developers also ignored the Unicode standard may be they did not have the “vision” and were shortsighted by the font’s display.

But there are people who follow the rules of the game and upgrade themselves to latest standard even if they stand out from the crowd. Among the major technology players, Apple has implemented Unicode strictly and has updated. This is the reason why reading texts in and out of Mac/iOS a pain. ஸ்ரீ typed on most non-Mac input tools uses ஸ் + ரீ and hence Apple font does not display the complex glyph and choses to display them individually as ஸ்‌ரீ. Other hand ஶ்ரீ typed on Mac will be shown as seperate glyphs outside Apple world which uses fonts with older standards. Where is the interoperability of Unicode? LOST! Think of searching these texts? We are back to pre-unicode days! Some fonts like Lohit-Tamil show the same glyph for both the forms, but that is a deviation from Unicode standard, but a step forward in usability.

Solutions?

1. Ask everyone to upgrade to latest Unicode standard. Easier said than done.

2. Make a proposal to Unicode to announce ஸ்ரீ and ஶ்ரீ canonically equivalent. This however will be a cultural issue with people following grammar strictly / treating language as a static entity opposing and modernists supporting it based on mass usage.

PS: For TanitTamizh folks, this will be a non issue, but fortunately / unfortunately Tamil masses still have to live this problem until a solution is found.

#OpenDataHYD Story

Sun, 24 Jun 2012 14:48:13 +0530

#OpenDataHYD Story:

Moments from #OpenDataHYD on 23 June 2011

SHRI/SRI ? -- ஸ்ரீ = ஶ் + ரீ != ஸ் + ரீ

Sun, 17 Jun 2012 23:14:11 +0530

Recently, reading Unicode standard for Tamil gave me some insights, both cultural and technological. I will first put my understanding / views on the cultural aspect and then come to technical part of it in a seperate post.

ஸ்ரீ is a ligature used to represent ஶ் + ரீ, one of the commonly used forms of ஶ், which is a consonent of modern Tamil primarily used to write sanskrit words. The equivalent devanagiri alphabet is श्री pronounced as shri. If one has traveled across Tamil Nadu, they can say many people will have problems pronouncing ஸ்ரீ. I have heard people calling me “See”kanth, “Stri”kanth. The most common pronunciation I have heard however has been “Sri”kanth which is how I write my name in English. This however changed when I came out of Tamil land and having Hindi speaking friends who used to call me “Shri”kanth. I felt they have stronger MTI just like those who called me “Stri”kanth.

Over a period of time, Tamil people have started pronouncing ஸ்ரீ as Sri instead of Shri. This is a deviation from “old-school vedic folks”(for the lack of better word to stereotype) who treat essence of word lies in its pronunciation. However, extrapolating Tamil’s pronunciation rules (which is only defined for Tamil characters btw), one can adopt a stand to pronounce it the way one wants, i.e either Sri or Shri.

For personal reasons, I have chosen to henceforth write my name in Tamil as ஸ்‌ரீகாந்த் as against ஸ்ரீகாந்த் as I believe written form is meant to serve as pronunciation guide and ஸ்‌ரீகாந்த் is close to how my name is pronounced.

PS 1: If you are unable to view ஶ் (SHA) on your screen, you can see the character U+0BB6 in the unicode table. It just means the font you use still doesn’t have a glyph for U+0BB6.

PS 2: In most cases you will not be able to type ஸ்‌ரீ on your computer (unless you use a Mac) and in the next post I shall write about why it is so, why apple products display ஸ்ரீ as ஸ்‌ரீ in some cases and how we should try to fix it part of the technical aspect of SHRI SRI

Reading Tamil texts in old script.

Fri, 15 Jun 2012 00:35:00 +0530

Tamil Script had undergone a reform on select characters to bring uniformity and simplify the writing style in 1960’s. The table below shows the old way of writing on the left and the current way on the right. I have mild memories of seeing bus boards having னை in சென்னை(Chennai) in its old form with elephant’s trunk during early 90’s.

Now its possible to read texts in their old form digitally, thanks to the newly released Lohit-Tamil-Classical font. Announcemnt Download link

Having classical fonts will not only help giving the user the feel of reading historical texts the way they were written, but will also be probably making digitisation of old texts simpler with OCR solutions.

Writing short texts

Thu, 14 Jun 2012 22:32:17 +0530

I used to blog sometime back in college. I wouldn’t say twitter killed it. Its mostly my impatience to sit and write long texts. So I now attempt to restart sharing content in “my own” place (as against a facebook wall post :)) writing extended short texts about myself, Wikipedia, Tamil, Technology, Public Transit, Chennai, India and other random things.