Hold Open The Glass Door...Please

Its that time of the year again when frantic efforts are underway for yet another Fedora release. Version 12, code named Constantine. Due to a turn of events, this time I got to be part of quite a few intricate processes - involving schedules, process changes and some amount of coordination. All for the Fedora Localization Project of course. As per the Release Schedule, the translation deadline is today - 10th September 2009. However, this time the FLSCo and FLP have been coordinating with Fedora Engineering to help us review the translations directly on the User Interface, by building test packages before the final freeze. Thankfully, FESCo gave us a nod on this and probably right after the translation deadline ends we would be ready to start talking to the package maintainers to get started on the package builds.

Although working on each of these challenges have been mind-numbing at times and mostly late at night after long hours at the day-job, its been immensely satisfying and fun. Probably, I got to experience a very microscopic fraction of the many complications that are very much part of an OS release and I can only try hard to imagine the entirety of the challenges faced and wrestled to the ground each day by people like John Poelstra. There are lots of people all over the world working very hard and for them Fedora is very personal.

Long time since this page saw some activity. *sigh*. This could have been a post of many things, like:

How stress induced fatigue (my dad's words not mine) caused me to sleep for nearly 36 hours at a stretch

The updates to the Gnome Mango system done by Olav Vitters and account system documentation done by Christian Rose has made things so easy for us translators

The mad rush for KDE 4.1 Translations

The LC Python workshop conducted by ramkrsna, at our office in Pune. Rahul Sundaram followed up with a talk on contributing opportunities in Fedora

Our new car

The huge power and water shortage that happened in Pune and messed up our daily schedules

The much-delayed fun trip to Mumbai and about the time spent with Barkha and her family, the ride on the deccan queen, boat ride to elephanta, visit to mahesh lunch home, getting soaked in the rain at Juhu beach, riding back to Pune in an ambassador taxi amidst pouring rain

My views on why overt channel admins (the pronounced green medals, not the access lists) on irc channels in some open-source projects creates unwanted hierarchical levels.

Mozilla 3.0.2 translation sprint. Am waiting for a few bug responses at the moment, but hopefully that should not stop the inclusion of bn-IN this time.

But then let me talk about something thats really much more important. The other day Ani showed me the search feature on the KDE Translation Project website, that allows searching of a term/string in translated content. The setup in this case gets the content from a selected directory of the svn, runs a query for the search string and presents the output results (string and its translated version) with direct link to the source documents. A database is also involved somewhere in between the process.

So a few of us were talking about having a similar tool that would allow us to search strings on user-defined content locations and present the strings with the search items, corresponding translated content and pointers to the source document. And so evolved Translation-Filter, by Kushal. A nifty little tool, that does just what we need. Its still being worked upon, but at the moment what you can do with it is:

Define a custom location with multiple .po files

Provide a string to search in the defined location

Get an output with the original english string containing the search item,corresponding translated string and the source file name

Provide a list of strings to search via a plain text file

Save search results as .html pages

Use the tool from the command line and a basic GUI dialog box

The project is a part of Fedora already and Kushal has packaged it.

At this moment the benefits look huge. Primarily it will allow us to ensure consistency of bn-IN translated content across projects (the ones using .po files at the very least). Perhaps (as Sayam thinks) very soon we can make a web-based version of it too. So right now... kushal++ :D

Tags: fedora-l10n, gnome-l10n, kde-l10n, l10n-tools

A few days back, I had blogged about the FLSCo Elections. The results are now out and we have the first FLP Steering Committee. Congratulations to a great team and wishing them the very best.

The First FLSCo:

Dimitris Glezos

Diego Búrigo Zacarão

Bart Couvreur

Noriko Mizumoto

Marek Mahut

Fabian Affolter

Piotr Drag

Tags: fedora-l10n

The first Steering Committe of the Fedora Localization Project is on the way to be formed. The nomination period started on the 7th of April 2008 and closes on 13th April 2008. The elections would be held between 14th April - 20th April 2008.

The Committee would be comprised of 7 members with one member being designated as the Chair and Project Leader.

The top four vote-receivers serve on the FLSCo for 12 months and the next three elected candidates would serve in the Committee for 6 months

The Candidates have to be part of the Fedora Localization Project (i.e. member of the cvsl10n group)

The members of the cvsl10n group would be eligible to vote

More details about the Elections can be found here.

If you are an active member of the Fedora Localization Project and would like to be involved in creating new ideas and implementing long-term strategy for the project, please put in your candidacy at the Nominations Page.

:-)

Tags: fedora-l10n

The bug, reported earlier, has been fixed over the weekend. Thanks danwalsh and Stickster.

:-)

This bug might cause a few guffaws, but then imho the messages would look more apealing to me once corrected. Nothing personal, danwalsh. :-)

Claude Paroz reported on the gnome-i18n mailing list today about the msgfmt errors that were showing up on the damned lies page against some files in a few languages. Bengali_India also has a few red medallions and the problem was baffling me since msgfmt checks on the files locally were not showing any errors. It was only after submitting them to the Gnome svn that the msgfmt error messages were showing up. After discussions on the gnome-i18n mailing list and irc conversations, the following seems the correct solution for plural form related errors for languages that do not have any plural forms.

Open .po file in a text editor (like gedit) and insert the following line in the header section:

"Plural-Forms: nplurals=1; plural=0;\n"

should look like this:


Once you encounter a string with a plural form there are two ways to go about it.

1. If both the strings contain a variable (like %d) indicating a number, then the translation can be done as per the language requirements for handling single and plural numbers. e.g.

msgid "%d file has been modified"
msgstr_plural "%d files have been modified"
msgstr [0] ""
msgstr [1] ""

In this case, after translation the string should look like the following in a text editor:

msgid "%d file has been modified"
msgid_plural "%d files have been modified"
msgstr [0] "%d-টি ফাইল পরিবর্তন করা হয়েছ "

Note: The %d has been retained in the msgstr.

2. The second way is the more important of the two and requires caution. If the singular version string does not have a variable, but the plural string contains a variable, then the translation of only the plural string has to be done. e.g.

msgid "one file has been modified"
msgstr_plural "%d files have been modified"
msgstr [0] ""
msgstr [1] ""

In this case, after translation the string should look like the following in a text editor:

msgid "one file has been modified"
msgstr_plural "%d files have been modified"
msgstr [0] "%d-টি ফাইল পরিবর্তন করা হয়েছ "

Note: The %d has to be present in the msgstr and will take the relevant number of "files modified" when the program encounters such a situation.

Also note, in both cases above the msgstr [1] "" has to be manually removed from the strings.

As a final check, please do ensure that the following command does not throw any errors:

msgfmt -vc -o /dev/null (filename.po)

For more information about plural-forms this document is extremely helpful.

Thanks to Sayam and his dad for digging out an old book and sending it across in reply to the RFC call for bug 247233.

As per an extract from this document, the symbol looks best to be positioned as below:



The above reference is from the book titled Bangiya Shabdakosh by Haricharan Bandopadhyay published by the Shahitya Akademi. Right now Sayam is gathering a copy of the complete three volume set in Kolkata.

A bug related to the positioning of the ৺ (Isshar sign - U+09FA) was fixed by Rahul today after quite a bit of research by Amitakhya and me. The character was being positioned at-level with the other alphabets:



As far as everyone could remember, the dot within the sign would have to be at level with the matra - the horizontal bar on top of the alpbhabet, looking like this:



According to the Chalantika written by Rajshekhar Basu, this sign is used as a prefix for the following:

1. Name of a deceased person (শ্রী শ্রীমতি শ্রীযুক্ত etc. are used for living people)
2. Name of a deity
3. Name of a holy place

dd told me earlier that, since it was used mainly by printers there had been no standardization. That compounded the problem. It would be great to have an RFC on this matter. As of now the current positioning looks perfect in conjunction to all the shraddha chiti (শ্রাদ্ধ চিঠি) that I have seen over time.

Anyways, moving from the morbidness surrounding this entire symbol, its good to see so much activity happening with the Lohit fonts thats very much part of the Fedora Indic locales. It holds a special place in my heart because... I named them (shamless plug) 3 years back! :-D