On the sidelines of the GNOME.Asia 2011 at Bangalore, an article in The Hindu “This one’s no gnome” was published. When Srinivasa Ragavan tweeted about it, I mentioned about the ill-formed comment
The paragraph of interest is here:
Excited to be in India, he concedes that community interest here is still on the lower side. Setting this straight is particularly important when it comes to GNOME localisation. “Localisation is a huge challenge here, mainly because there are so many languages, and also because of the way the alphabet font is linked. This is where Free software is critical, because smaller the user base, lesser the chances proprietary firms will take this up.†We count on enthusiastic developers, who are proud of their language and want to preserve it in the digital realm, Mr. Cameron adds. In fact, he points out there are “big business opportunitiesâ€, domestic and international, for those who commit themselves to such projects.
Brian mentions about community interest in localization being less. On the face of it, this appears to be a conclusion derived from looking at number of participants in the language communities. The way I’ve learnt to look at localization is to try and see the communities and the efforts which sustain. Localization is a steady and incremental process. The communities which take the time and make the effort to reach and maintain their place in and around the “supported” percentage for the individual Indic languages are the ones doing great work.
With each release of GNOME there is an obvious increase in the number of strings up for translation. I don’t have the exact statistics to support this fact but I guess a trend-line generated from the total number of strings available in each release super-imposed over the trend-line for the ‘supported percentage’ figure would see an increase with each release. I’m sure someone actually has this data. This basically means that within a short period of ~4 months (factor in the code freeze and string freeze etc), which may or may not also overlap with other project releases, the localization teams end up completing their existing release work, review pending modules and polish up translation consistency ensuring that with each release of the desktop there is a step towards making it more user-friendly.
That’s why localization is a big challenge. Not because of the number of languages and certainly not because of the “way the alphabet font is linked”. For what it is worth, the latter bit is more in the realms of internationalization and there are efforts at multiple levels to ensure that the remaining few outstanding issues get fixed.
This brings us to a small gripe I’ve had about a lot of Free and Open Source Software projects who take their L10n communities and participants for granted. I’ve rarely seen a project board reach out and talk with the participants to figure out what can be done to make things better. For example, would a community doing L10n do more awesome work if they were funded for a 2 day translation sprint ? Do the language communities have specific requirements on the translation and content infrastructure or, the translation workflow ? Have these issues been brought up in Technical Board meetings ? GNOME isn’t the only one which repeatedly fails to do this transparently, but it is among the highly visible FOSS projects which seems to assume that it is an obligation of its volunteer contributors to keep the statistics ticking.
What suggestions do you have? Send them to release-team@gnome.org or discuss on gnome-i18n please.
Thanks for dropping by Olav. So which list should get the suggestion ? The lists you mention are have a bit of orthogonal objectives.
And, if you notice, the greater part of the blog was about my impression of why Brian’s comment is ill-formed and why I think he arrives at a somewhat mistaken conclusion.
So, do you want me to suggest what Brian should be speaking or, do you want me to suggest on what the GNOME Board should be doing about L10n ?
The “Supported language” definition should definitely be discussed on gnome-i18n mailing list and you are not the first to bring this up. Now is a good time, hence feel more than encouraged to start the debate. 🙂
On a side-note, the “ever-growing” number of strings to translated is totally dependant on definitions, as by the redefinition of modulesets the number of strings to translate for “core” GNOME 3.0 is lower than for GNOME 2.32, plus to mention first attempts by Friedel and Claude to help with priorisation (“reduced translation files” on l10n.gnome.org that exclude the schema strings).
The suggestions you have about l10n towards the release-team. You can either discuss on gnome-i18n and then forward the suggestions to the release-team, or discuss directly with us. Not everything is practically possible of course. Only 6 months and perfection is a goal, but we have to take into account to what’s possible in practice.
I don’t know what suggestions you have, so I don’t get why you wish to suggest that to the board. Seems more like gnome-i18n material and then release-team to make the final decision (we coordinate releases; suggestion seems to be about i18n and release process).
Andre, thanks for taking the time to read and respond. When you mention the work being done by Friedel and Claude, do you have a set of data points about how much their proposal of reduced translation files actually reduces the number by ? I recall reading a post from Friedel talking about how to redefine and arrive at essentials (I think it was off p.g.o) but I don’t follow the gnome-i18n mailing list closely enough to track where the proposal exactly is.
I’ll see what can be done to read up on the existing set of proposals.
And again, you’d notice that the thrust of the post is my impression of where Brian seems to have missed the point. I am yet to touch the i18n/l10n aspect. That is another post for another day.
I have no link by hand, but the gnome-i18n@ archives on mail.gnome.org should be helpful, plus all the data (counts of strings for normal and reduced sets) is available on l10n.gnome.org so it shouldn’t be too hard to find out if you want.