Hi,
Due to my ongoing wikihiatus, I haven't followed much of discussions in this list.
But, I'm interested in this issue because I was following and supporting the original
request filed by User:SanthoshGuru asking for such a support in 2005.[1] While I welcome
the recent energy in building and launching solutions for Indic languages by WMF. However,
like Srikanth and Bala have mentioned, we should ensure that we don't break the
experience of thousands of readers visiting the wikis everyday. The situation is very
different now than in 2005. So, let us take care to test every new launch tested out in
translatewiki before rolling out. We can perhaps create a rollout routine to include a
thumbs-up from some community volunteers.
Regards,
Sundar
[1] -
https://bugzilla.wikimedia.org/show_bug.cgi?id=2361
"That language is an instrument of human reason, and not merely a medium for the
expression of thought, is a truth generally admitted."
- George Boole, quoted in Iverson's Turing Award Lecture
________________________________
From: Srikanth Lakshmanan <srik.lak(a)gmail.com>
To: Discussion list on Indian language projects of Wikimedia.
<wikimediaindia-l(a)lists.wikimedia.org>
Sent: Tuesday, December 13, 2011 9:17 PM
Subject: [Wikimediaindia-l] Webfonts deployment on Indic Wikiprojects
Hi all,
Pardon for a long mail, just unavoidable.
We have seen the WebFonts roll out[0] last night to most Indic wikis except Malayalam and
Tamil. We in Tamil Community felt WebFonts extension is just not ready for us. We would
like to share on why we in Tamil community didnt chose webfonts and also what could work
better in future for Indic communities during technology adoption.
1. Unavailability of "Quality Free fonts" :-
During the development, our request for not setting the default font as a lower quality
font was rejected citing "it will defeat the purpose of extension".[1] The
available fonts had issues and deploying the extension with those low quality fonts to
everyone would not only defeat the purpose of extension, but also gives Tamil Wikipedia a
bad image when people just cant read the fonts even though they had better fonts in
system. (Just like how i18n team says people who see boxes will just simply close the
window, we say giving these poor fonts will also lead to same thing and not help the
cause, instead will also hurt those who have better fonts.).
2. Quality of User Experience :-
We are a smaller wiki, we have a smaller reader base, but still we are ranked 7th most
visited website in Tamil according to alexa[2]. Just like how no-nonsense / no-mediocrity
is tolerated in any code that enters WMF cluster, any change which will affect the
site's look and feel, user experience will have to be of highest standards and must be
accepted by community. WebFonts were just not ready to enter Wikipedia, since it changes
the UX for all the readers to help a potentially lower number of users who dont have Tamil
fonts than the current existing reader base.
What we feel was wrong in WebFonts deployment
We have also been seeing the wikis where they have been rolled out and reporting issues.
We ourselves are reporting issues inspite of not taking WebFonts, with the hope software
just gets better and some day we can deploy them. Dont get us wrong, we are not against
technology, we just need it in better form and are not in any urgency. (After all we at Ta
wiki initiated an RFC and asked Webfonts even before the announcement was made). We would
also like to mention some points which we feel i18n team could have done better for a
smoother launch.
1. Font Testing :-
The point of language support team is that the people who are aware of language give
feedback to make any software better supported for the language. We are not sure if
Font-Testing was ever done at all for those languages where the WebFonts were deployed.
The hinting issue which was a concern and made us raise against deployment in Tamil is
also present in Hindi,Sanskrit,Telugu(atleast till we saw) and gave the same worst
readability. The i18n team did font assessment[3], testing only 1 word to test the font.
Can any font be tested with just rendering of 4 characters / 1 word? For Tamil,we did a
test in little more comprehensive way(We would not say its complete)[4]. This should have
been a *must* to see rendering / font issues with chosen default font especially since the
fonts are being set default to every single user to the site. Sadly community was involved
the least, a note was posted in Village pumps and we dont think community involved itself
in any testing and poor quality was eventually pushed without proper testing.
2. Real world testing :-
Though cross browser testing was done, there was a severe lack of real world testing and
as a result we are seeing a host of issues being discovered post launch. Average PC in
India might have 1 GB RAM, Firefox 3 / 4, worse IE 5.5 / IE6 on 100 Kbps semi-broadband
connection. We cannot tell them move to latest or ignore them. More care should have been
taken especially since the webfonts is bound to set a default font.
With only a few hours of testing serious issues have been found - in IE 8 where webfonts
might be rolled back [5], IE 7.0 (where webfonts dont work) [6] , ubuntu + Firefox (fixed
now) [7] and Win+Firefox 5.0, [8]. We are still testing for other browsers and usecases
and dont know how many issues we will discover. In short - This code is not ready to go
live, especially when it is being made default compulsorily for everyone. There are
serious performance issues for typical Indian internet connections as well.[9]
3. Communication and Community Engagement :-
Most of the above things could have solved earlier if there was more communication and
community engagement. We asked for more information, engagement on this very list. There
was no reply to the mail on increasing community engagement for i18n projects[10]. Most
communities know WebFonts is coming on Dec 12, didnt know what was coming, any further
details. Worse, Even Indic Consultant was not having clear information. Why this lack of
transparency? Community is more than willing to help, if only they are informed. Even
though we did not take up WebFonts, we have spent time to help making it better.
And the end users in the wikis dont know where to report and follow up issues. (Not
everyone is aware of and familiar with the Bugzilla process). The request we raised to
have a visible bug reporting link has not been acted upon [11]. There might be a lot of
issues going unreported, because people dont know whom to report to. When a change of this
scale is being done, Community admins must be advised to run Sitenotice campaigns to
inform the users about the change with some solid newbie oriented documentation. Infact
this must be done for RFC itself, so as to make an informed decision. We did the same for
RFC in Tamil[12].
Proposal for Future i18n / any special deployments to Indic wikiprojects :-
The WebFonts deployment is a classic example of making deployment without enough community
engagement. Can this done in any of English / German / Russian wikipedia which have a
strong community? The fact that Tamil / Malayalam resisted was because the community had
concerns over the solution. Till an hour before deployment yesterday, we did not know if
Ta wiki projects will get webfonts despite our objections. We had been pursuing Siebrand
and Gerard across forums - facebook, twitter, meta talk pages, village pumps, gmail chat
etc looking for answers. But till the deployment happened, we had no clue what we would be
getting. This method of deploying in silence *must* stop ASAP. Any deployment to any Indic
wiki must go through the community (language support teams) informed of the change with
Indic Consultant kept in loop. We suggest Shiju Alex to work on a policy and put it in
place regarding this. If there is problem identifying community members to
help, we
are sure Shiju will help connecting.
Irrespective of that happening Tamil Wiki Projects will follow this process.
1. Test any deployment on translatewiki
2. File Bugs and verify in translatewiki till it reaches acceptable level.
3. Language support team member will make a RFC page explaining the merits / demerits of
the technology in simple terms with use of screenshots / external links
4. Reverify / Ask for deployment in largely-inactive Wikiprojects like WikiQuote /
WikiBooks *post community concensus*
5. Test again / File Bugs
6. Reverify / Ask for deployment Wikiprojects next in line in terms of activity (
Wiktionary / Wikinews/ Wikisource)
7. Test again / File Bugs
8. Only after ironing out all issues, any deployment will be allowed in Tamil Wikipedia.
We had burnt our fingers during Narayam deployment already once and community was so
resistive of Narayam itself and was asking to go back to older javascript solution. After
that we followed the above process for bringing back Narayam on all Tamil Wikiprojects.
We suggest the other communities adopt something similar. We sincerely hope that the
community engagement is improved, not just before deployment, even from start of
development.
[0]
http://blog.wikimedia.org/2011/12/12/going-live-december-12-2011/
[1]
https://bugzilla.wikimedia.org/30506
[2]
http://www.alexa.com/topsites/category/Top/World/Tamil
[3]
http://translatewiki.net/wiki/WebFonts_assessment
[4]
http://translatewiki.net/wiki/User:Sodabottle/test1
[5]
https://bugzilla.wikimedia.org/32775
[6]
https://bugzilla.wikimedia.org/33024
[7]
https://bugzilla.wikimedia.org/33025
[8]
https://bugzilla.wikimedia.org/33018
[9]
https://bugzilla.wikimedia.org/33027
[10]
http://lists.wikimedia.org/pipermail/wikimediaindia-l/2011-November/005153.…
[11]
https://bugzilla.wikimedia.org/32942
[12]
http://tawp.in/r/2r1f
--
Regards
Bala Jeyaraman & Srikanth.L
_______________________________________________
Wikimediaindia-l mailing list
Wikimediaindia-l(a)lists.wikimedia.org
To unsubscribe from the list / change mailing preferences visit
https://lists.wikimedia.org/mailman/listinfo/wikimediaindia-l