multilingual/spec

From twext

(Redirected from multilingual)
Jump to: navigation, search
 

most recent spec is at i18n
input has the latest on xcroll.. xcroll needs some sweetening..


Contents

[edit] .twx

.twx is maybe we turn titles into somekinda database
so we define and find data in a ..numbered namespace..

..n..twx

..0..1..2..3..4..twx
..what..where..when..who..etc..twx
..title..langs..time..editors..tags..twx
..long_honkin_title..ENGLISH.espanol..200714.093422.1..waqas,duke..file_system,spot,meaningful_url,twexter,db..twx

above could easily print reports:
0: title: LONG HONKIN TITLE
1: langs: ENGLISH.espanol
2: clock: 200714.093422,1
3: edits: waqas, duke
4: mtags: file system, spot, meaningful url, twexter, db

search could focus on specific ..n.. to find tags or whateva..
<ignore>..bla..yawn..zzz..</ignore><BLINK>3</BLINK>..zzz..bla..

[edit] timestamp

timestamp per save provides single point of truth in memory.. 
so a .twx file needs not title (..0..) thus can maybe serve 
dichos and citage..
filemaker has a nice way to save.. it just saves.. possible while timestamping ?

[edit] mtags

tags (google labels) categorize info.. 
can simple markup make tags meta?
a;author
o;owner

so, if ..4.. = ..mtags.. then in ..4..:
..l;link,c;credit,a;author,l;loop,w;whateva.. simple arrays
 ? include language_ID: x;ENGLISH,xx;espanol ??

[edit] translatable

if
  1A=A3
  2B=B1
  3C=C2
then
  1=3
  2=1
  3=2

so could be rearranged right to left:
..who..tag..when..where..what..twexml

disorder, reordered:
..etc..what..where..who..when..twexml

..3..2..1..0..twx

[edit] feedback

roberto liked it.. waqas?


















image:XCROLLLOOKFEEL4.gif

scroll down for usr/login routine mockup, comments














image:TwexterMultilingualRough.gif

waqas: login doesn't need to be a tab

when usr logs in, tab converts to usrpref, including:
  • languages to include/exclude in #slop
  • personalized format preferences (default)
  • change password, edit user page, manage account etc
  • manage chunxas for specific language pairs
  • manage mtports for specific language pairs
the point is, usr may have lots to manage and might like easy prefs access

waqas: editor doesn't have to be a tab

it's standard in mediawiki and we're aiming to wixi.. still, simple is good, so if above "window shade" edit can work w/ all twext pages (or even sections) that might be nice

above is rough sequence for visitor who creates account

  1. visitor gets a basic intro text twext
  2. visitor can play with controls but can't save files?
  3. visitor creates account
  4. login tab changes to $usrname tab
  5. $usr controls account by clicking $usrname tab
  6. selects languages to load in #slop menu
  7. can #addlang (soon)
  8. $usr controls styles maybe w/ a preview?
  9. $usr controls account passwd, create homepage, etc
  10. edit has new input only xcroll option
  11. saves edit
  12. creates new file
  13. find or "search" for existing files (in #slop pair}


MULTILINGUAL TWEXTER SPEC
multilingual twext helps us save, manage, find twext texts..

[edit] SAVE (and search) titles

basic twexter lets users try simple interface.. to get twext multilingual and findable, users need to SAVE files, probably with TITLES and hopefully with TAGS..

[edit] XUL

evert says XUL is way to go.. i'm still trying to figure all this stuff out..

[edit] XML

ideal format will be flexible and searchable.. this may be already be happening with zeen's deliver of basic twexter :) if so, then we can add features to our xml output..

[edit] XSLT

"flexible" meaning easy to let other programs read data in saved file, as a single point of truth with variable output: then convert to various formats:

sounds like this happens w/ XSLT.. zeen say ok=)

[edit] searchable?

you tell me.. sounds like XML lets us define new doc type, then perform searches to find only our new type of doc.. even filter specific searches within our new doc type.. if so, our new xml doc type might implement some features:

[edit] language identification

give user option to identify languages of TEXT and twext parts of saved file.. for example, within a document,

  • TEXT may be kiswahili
  • twxt may be portugues

if someone is in brazil learning kiswahili, they'll want to find twext docs with big kiswahili TEXT supported by little twxt in portugues..

the design should anticipate trilingual and multilingual content within a single document.. at a root level, we might try to avoid being trapped by categories

[edit] tags

let user simply add tags and maybe somekinda shortcut smart tag.. tags are good because data isn't trapped by categories.. categories are good because they help sort data.. tags are a great example of being able to categorize data without trapping it in some freaking folder somewhere.. we should find an easy, flexible way to include and modify tag info in our xml docs..

  • t; idea, half-baked, suggestion tag
  • a; last name, first name author tag
  • o; yadda yadda, inc owner (publisher) tag
  • x; translator name name of translator tag

idea is to make it easy to tag a twext file w/ info easy to sort.. ie if line starts with "x;" then doc may include translator info repeat: i'm new to xml so this may sound real stupid..


[edit] title

3rd in line.. titles are useful, but tags might be more useful.. especially in the context of delyric citage.. another type of title-less twext object may be dichos (simple quotes).. maybe titles could just be tags? please don't hurt me

[edit] twext folders?

twext folders is an old idea to store source files in folders.. it's stoopid which may be a good thing™.. and clunky which may be a bad thing™..

twext folders is probably way outdated.. replaced by xml.. the point was and is do not trap data in programs, databases, prioprietary or free.. the beauty of html and txt is they're platform independent.. truly free

representations of twext folders might be useful for kids using wixi..

[edit] SLOP select language options menus

Select Language Option Menus might be DHTML trick.. basically, drop down menus loaded with user_pref langauges.. i've seen some work very well, others not.. so maybe two ways to go:

  1. slick DHTML (precursor to xnav)
  2. old school html (ugly but widely useful)

SLOP lets user Select Language OPtions.

image:Slop1.png

slop uses dynamic menus: twexter responds to changes in either menu instantly, with no need for any additional input button.

the interface language is controlled by the twext_slop_menu on the left.

so if user changes twext_slop_menu from english to español, then interface language changes from english to español

image:Slop2.png

on the right, the TEXT_slop_menu controls the TEXT language; this menu directs search queries, chunktext inputs or format fetches to internal xml format

so if user changes TEXT_slop_menu from ESPAÑOL to ENGLISH, queries, fetches and inputs will interact with xml

image:Slop3.png

the TEXT_slop_menu does not control the interface language, but make one exception: the word for "TEXT" (left of TEXT_slop_menu) should be in language selected by TEXT_slop_menu. confusing? either eliminate "twext" and "TEXT" labels (rely on lowercase in twext_slop and UPPERCASE in TEXT_slop) or just go DHTML:

image:Slop4.png

solutions must be robust, bug-free, strong on most browsers and work with unicode.

if in any SLOP MENU user selects "+"

image:Slop5.png

temporarily save user input, so user can be sent back to page without losing input then send user to prefs:

[edit] XPREF twext user preferences

to multilingualize, we need to identify languages.. adding languages to twext will hopefully be very easy, so many languages, even dialects, slangs, hybrids may be added.. too many..

users are likely to want to exclude many languages from their twext interface.. users may also want to control personal preferences for format, style and output of twext text between any two languages..

http://twext.cc/dev-old/xo/control.html shows ancient attempt at user preferences interface.. soon regurgitating here..

[edit] FIND

hopefully "internal xml" output will produce flat files that, if stored in urls, can easily be searched with your favorite search engine.. hopefully, we can add tags to such "internal xml" output to filter searches by language, author, translator, generic tags, etc..

is this possible with xml output we're defining?

waqas?

question re: naming convention

  • "internal_xml_output"

twext is meaning lots of things..

  • for end users, a "twext" file should be something easy for them to find in the languages they want..
  • for developer, "twext" is meaning
    • chunk translation input
    • xml output
    • xslt conversion to
    • many end file formats

what should we call "twext internal xml ouput"? xxml? xmlx? twexml? twextxml?

waqas?

[edit] ADDLANG add languages to system

before adding langs to twext, user should be able to

  • PREView
  • SAVE
  • PREF
  • FIND
  • SLOP

addlang should cost minimal user work, and avoid being trapped by categories.. when preview, save, prefs, find, slop work, then we can add languages to twexter:

image:TwexterAddMaya.gif


all above features can work with xlyric script which formats lyrics only.. above stuff working in a complete system will turn songs into language teachers and work between many language pairs..


[edit] XTEXT twext line break format script

after above complete, then the kicker: twext line breaks will work with normal texts.. in all kindsa different window sizes..

feel free to make xtext script earlier.. if i gotta pay for it, i wanna see addlang as soon as possible :)


[edit] iDNS

idns they say is coming, i hope they're right, 'cuz i am a nut for meaningful urls.. it'd be nice to save titles you could just type into the url.. might work with somekinda a search function or conflict resolver..

idns is probably a deeper layer for a separate bid xc.cx

 

reconstructing
http://twext.cc/dev/dev2006.html

1. BASIC TWEXTER SPEC defines an interface people might use to get text twext.. it may have useful features:

  • INPUT
    • stoopid, not hard to understand for a user
    • xcroll easily edit/compare text/translation
    • input/preview combined in single interface
  • OUTPUT
    • solve html table prob
    • format adjustable
    • XML ready > CSS

2. MULTILINGUAL TWEXTER:

  • save and find basic twexter files, identified by attributes like language_pair, tags, title, author, owner, translator, license, etc..
  • prefs let us limit our focus to languages we're learning and lets us customize our twext experience
  • slop makes it easy for us to navigate languages
  • addlang lets people easily add languages, dialects, slangs to get twext

3. AUTOMAGIC TWEXTER:

  • xtext fast, flexible twext line break service
  • chunxa automatically chunks TEXT to get twext
  • mtport connects machine translation to chunxa
  • xcat connect chunxa + mtport + basic interface
  • xurl automagically twext translates urls

comment: automagic twexter is likely to suck at first but will hopefully learn to suck less.. xcat might help machines learn from human corrections.. at this stage, twexter may be useful to OmegaWiki

 

Personal tools