Agree with Srikanth on WebFonts. We at the Tamil Wiki projects have already tried the webfonts out and found the Lohit Tamil - the font available via webFonts extension has serious readability/rendering issues in most operating systems (most except Redhat Linux). It is a big step backwards in usability for those who have better system fonts (which in Tamil is like 85-90% of the reading public). The i18n team is not giving the option of making the system font default. It is either webfonts with Lohit (or other faulty fonts) as default or no webfonts at all. [1][2] So in case of Tamil, the community will not ask for the webfont extension, until we have a bugfree freely licensed font. Our advice to the other communities is to thoroughly test this in all operating systems and then decide if the font webfont extension is offering as default is worth forcing everyone to switch. In fact please discuss anything that the i18n team is offering thoroughly and test it before going for a rollout. Because once the rollout is complete, your control is lost and you will be entirely at the mercy of the i18n team about any changes [1] My request to the i18n team is to discuss with the stakeholder communities, well before you start developing new extensions. Talk with us Give us a feature road map and develop software to suit our requirements and needs. And please dont go for one-size-fits-all solutions. As of now, the inverse is happening. We are being given products and enhancements without our consultation. Any request to roll back/change is met with refusal [3]. As demonstrated in the case of Lohit-Tamil and Webfonts, it took a face to face meeting at Mumbai Hackathon and comparing the font rendering in different machines, before an i18n member would admit there was an issue. Before that i18n team was casually dismissing our concerns without listening to what we were saying.[2] In Tamil Wiki webfont case we were lucky that we had a Hackathon happening shortly and Srikanth was there to raise the issue in person with Santhosh and the Lohit developer (Redhat). If this is the case with Tamil where we were fortunate enough to have a developer amongst ourselves, untested rollouts of extensions are going to cause more serious issues in relatively undermanned wikis. Though the development is done in good faith, it cant get live without user testing / quality acceptance. I hear there are more such extensions in the pipeline from the i18n team. My sincere request to you guys is this - talk to your end users (us) before you plan the features and develop them. Requirements gathering from users is supposed to be the first step of any software development. Instead of waiting for feedback until you are ready for deployment and then arguing about the feedback, getting early feedback will everyone save time.[1] The WMF features team did the same type of mistakes with Feedback tool, image filter, Wikilove and the AC Trial. They appear to have learned from the community reaction and have hired Oliver Keyes to engage with the community on such issues. [4] You can atleast talk to us and find out what we actually require before forcing us to accept what you got or/and trying to convince us that it is good for us. [1] https://bugzilla.wikimedia.org/show_bug.cgi?id=31936 [2] http://ultimategerardm.blogspot.com/2011/11/freely-licensed-fonts-are-ugly-… [3] https://bugzilla.wikimedia.org/show_bug.cgi?id=32257 [4] http://www.gossamer-threads.com/lists/wiki/wikitech/255518 On Thu, Nov 24, 2011 at 6:34 PM, Srikanth Lakshmanan <srik.lak(a)gmail.com>wrote;wrote:

On Thu, Nov 24, 2011 at 20:53, Erik Moeller <erik(a)wikimedia.org> wrote: Exactly. The agreed answer was to improve Lohit-Tamil and fix issues before using it on Webfonts instead of getting to develop/use preferences which Tamil community originally suggested. Fixing issues is good in the sense, mediawiki / wikipedians dont just consume opensource, they contribute back, but realistically might take little longer. I am willing to help, just that I need to learn and do things :) As per Santhosh, "Spying the system" is not possible currently and hence we cant know that there are fonts on the system, leave alone checking if they are superior. I was wondering if this requirement can go into HTML5 if it makes sense / feasible, but thats a long way out. (NB - as Siebrand noted, the i18n folks are traveling right now, so it We shall wait :) -- Regards Srikanth.L