Eana Eltu: Translator, Dictionary, API and putxìng.

[Toruk Makto]

The US central timezone is GMT-6 hours in standard time and GMT-5 hours in daylight savings time (which is coming up this weekend, BTW). Your test showed 6 hours, not 8, so it looks like that worked.

[`Eylan Ayfalulukanä]

Of course, another fix is to have your update cron job run daily at a time when it won't matter what the DST offset is (assuming your computer clock runs on UTC, which is standard practice for me). (You may have already said this, and I didn't catch it  )

[Tuiq]

There's a time of the day where the DST does not matter?

What?

[`Eylan Ayfalulukanä]

If you want to do a daily update, and do it as soon after 'midnight' as possible, and you are dealing with a periodic DST offset, you make your update time, say, 1 hour and 5 minutes after 'midnight', standard time. Then, during the standard time part of the year (assuming you are not switching time as well), you update 1:05 after 'midnight'. During DST, you update :05 after 'midnight'. So you can still update once a day, and ahve a list that no more than about an hour old, regardless of the status if DST.

All I know is DST creates a LOT of extra work for me, including two 'overnighters' a year for computer clock updates (proper time, to the 30th of a second, is very critical in a broadcast plant.)

[Tsyalatun te Eyktan Txuratu'itan]

The US central timezone is GMT-6 hours in standard time and GMT-5 hours in daylight savings time (which is coming up this weekend, BTW). Your test showed 6 hours, not 8, so it looks like that worked.

Irayo ma Markì.

While I'm posting here, it seems that the NaviData.sql file is no longer updating on eanaeltu - the last update was 1st March - did something break?

I know Seze uses this file to source the data for his learning apps, and people have been saying that hasn't been updated for a while now either.

[Tuiq]

SpeakNavi.pm is acting up. The definition for --nay ( {\bf (unpro.)} creates a new noun that is related to the original by being a step down in some relevant heirarchy - size, rank, accomplishment, etc.) is incomplete. Argument seven, which is the sdesc, is missing and therefore, SpeakNavi.pm dies.

[Toruk Makto]

This is now corrected. Thanks y'all!

[Tirea Aean]

Could someone with access to the Perl scripts please make it so that Russian and French are included in NaviData.sql dump?

[Tuiq]

That requires SQL access, not Perl access. It's a database thing. I'd unlock Russian if I could right now (but my access is currently borked?), but French is at zero percent and therefore certainly not going to be included.

The rule was, I believe, 90%-or-more.

[Tirea Aean]

Oh. For some reason I thought it was Perl scripts that used MySQL to do stuff.

Anyways, irayo.

EDIT: French is at zero? O_O But their dictionary, http://files.learnnavi.org/docs/NaviFrancais.pdf seems up to date.
DOUBLE EDIT: OH. They must not be using EE for that. That dictionary is very different than the rest. I wonder why..

[Tuiq]

Technically, it's Perl using MySQL data to create SQL. However, I've moved the "which language is part of the SQL/jMemorize" long ago into a separate branch in the database, because it makes for handier editing.

Also, yes, that's not one of ours. There was one guy working on a French translation once, or at least asked, but it's either all outdated now or never happened.

[Tìtstewan]

The reason why there is no French version of it, is that Xelloss has created a dictionary and included it into the "Na'vi for Dummies".

[Taronyu Leleioae]

This might be helpful info...

http://files.learnnavi.org/docs/NaviFrancais.pdf

Looking at the french dictionary document, it has an edition date of Aug 2013.  But the format/font is different. 
Thus I think Xelloss is updating the french edition manually and posting the dictionary changes periodically to the media reference section.
Maybe he hasn't had access to be able to create a centralized/linked database file?

[Tìtstewan]

Exsactly this I wanted to mean...
And it is also in the Na'vi for Dummies page 117 following

[Tuiq]

This might be helpful info...

http://files.learnnavi.org/docs/NaviFrancais.pdf

Looking at the french dictionary document, it has an edition date of Aug 2013.  But the format/font is different. 
Thus I think Xelloss is updating the french edition manually and posting the dictionary changes periodically to the media reference section.
Maybe he hasn't had access to be able to create a centralized/linked database file?

The database file is populated by EE's data, which is basically generated by translating Taronyu's dictionary. Now there are reasons as to why somebody would not want to use EE as it has happened in the past (but I think we've improved a bit since then).

In May 2013, somebody wanted to/started translating the dictionary, but I believe we ran into some technical difficulties around that time. I think we've fixed it, but the person disappeared (?). In theory, French shouldn't give us as much trouble as other languages, but I believe there were some general server issues around that time.

[Taronyu Leleioae]

Tuiq,

Just for self understanding/clarification...  Which database is THE authoritative one?

Is Taronyu's dictionary creating EE's database file?  Or is the central EE database file creating Taronyu's dictionary? 
I've been actually told both in the past and I want to make sure we're explaining the correct direction in the new NiaN draft.

Thanks!

[Tuiq]

I suppose the probably best answer is "both". Taronyu's dictionary has become EE's database and vice versa. The original dictionary fed the database, which has since then been maintained and creates the new PDF. All the fancy output is also created from this raw output, therefore using more or less the dictionary itself.

It's probably some sort of third-case. The database, if you were to "print it out on a sheet", would be more or less a giant mess of LaTeX. EE then strips the LaTeX down, tries to make something useful out of it, and puts it into SQL/other formats.

[Tuiq]

As a more technical appendix, I've posted in that thread the PDF that EE was, more or less, based on back then. It was simply reading the LaTeX line-by-line, inserting the words into a (slightly more fitting) database structure. There, it was then modified by Taronyu (and the translators), where a Perl script merged the database into a TeX and compiled it again.

At the same time, I've written an IRC bot that would read said database and offer simple word lookups/grammar parsing. It was really just doing stuff like "The third argument of words of type 'affixN' is English, the first one is Na'vi". That's still how it is done - it's reading certain fields for certain types to build an internal, more meaningful database. This database is what generates the SQL and isn't saved anywhere; it's re-created and destroyed every time the stuff is generated (usually once a day I believe).

It's difficult to say, as technically I do suppose that EE's database is Taronyu's dictionary, which is made from EE's database, which was initialized with Taronyu's original dictionary. It's a real mess to be honest at this point and EE2 would surely make things easier by saying "There's a centralized database that creates the dictionary, the SQL and everything else."

[Taronyu Leleioae]

o_O

Interesting...  Thanks for taking the time to explain this!


I have one followup question.  Where does Markì inject the changes/new words?  Into EE database?

[Tuiq]

It's added to the EE database, yes. It's added in the same way that Taronyu would have added it to his LaTeX though - so more or less, you are simply adding one more line with a word in it.