Internationalization / i18n support

[pencilcode]

Internationalization / i18n support

[lhaussknecht]

+1

[allan-simon]

+1

[davidbau]

A question.

When teaching in an environment like Scratch, you translate all the text inside the blocks to the native readable language.

When teaching in Java, you wouldn't translate Java's library functions like Array, String, .length, etc. These would be left in English (right?)

What is the right approach to code when in Pencil Code, where we want to put literal code inside the blocks?

[luebken]

Ok. Maybe I misunderstood the goal: What differs Pencil Code from Scratch?

e.g. I would translate the following

pen green, 3
rt 360, 10
lt 45, 30

to German

stift grün, 3 
ld 360, 10 // links drehen 
rd 45, 30 // rechts drehen

Edit: Oh sorry now I get you. You are right some things wouldn't be translated like library functions or keywords like do or for.

[davidbau]

Right. Here is a question for teachers of non-english-speaking students. We can break things down into different strings that might or might not be internationalized. Which ones does it make sense to translate?

(1) Help text such as "sets the pen color" hover text or other tutorial text.
(2) Labels on buttons or other UI elements such as the "Save" button.
(3) Turtle library function names such as "rt" for 'right-turn' or "pen" or "button" or "label"
(4) Color names that come from standards such as "saddlebrown" or "red" from the standards http://www.w3.org/TR/css3-color/#svg-color
(5) Event names (and corresponding event-listening functions) that come from standards such as "click" or "keyup" or "keydown".
(6) Other less-seen names that come from standards such as "font-family", "color", "background-color" css attributes. "" (HTML created by the label function), "", "" (HTML created by the button function), "" (HTML created by the table function), and CSS selectors for "label img table button". etc.
(7) Library function names from base libraries such as jQuery "hide" "show" "finish" "remove" "animate".
(8) Javascript standard library names such as "length", "Array", "push", "pop", etc.
(9) Coffeescript and Javascript keywords such as "for", "switch", "if", "break", "return".

Certainly #1 and #2 should be translated. Almost certainly #8 and #9 should not be translated. The others are a matter of judgement. The disadvantage of translating computer symbols like function names and color constants is that programs then become incompatible between locales.

One goal of Pencil Code is to propel and equip students to use mainstream programming languages, which is why many "professional" idioms such as jQuery, CSS, HTML5, javascript are exposed. Which ones above would a non-english-speaking teacher want to have translated for students?

[allan-simon]

I would say as much as possible, 1) and 2) being essential, 3) being quite important, the rational being that if someone really wants to expose himself to "real world programming" then he can always switch to the english version.

My experience teaching computer science to university students in China is that the fact that programmng languages/documentation are in english is a HUGE struggle for them, something that even as a French I didn't notice before because for latin-language-speaker "click" is "clic" , animate is "animer" , so we didn't need to learn all these domain specific words.

For "incompatible between" locale I think blockly for example solved this by getting an intermediate representation which is locale-less

[davidbau]

The dilemma (and promise) of Pencil Code is that there is no intermediate representation.

When you toggle from blocks to text mode, your program is exactly the same. So if we have a Chinese block for "right turn" shown as 向右, then when toggling to Javascript or Coffeescript, it should read, in the Coffeescript program text as

向右 90

In other words, with the current design, in order to have a Chinese-readable right-turn block, there would be a function called 向右 in the underlying library.

So I guess I see two ways to solve internationalization:
(1) Translate the turtle library. We'd have a version for each language, and programs written for one language would not be able to run in a different language.
(2) Change the block editor so that text shown in the blocks is not necessarily the same as text when you switch to text-code mode.

Maybe what you are imagining is (2)?

[luebken]

The dilemma (and promise) of Pencil Code is that there is no intermediate representation.

That's what I especially like about Pencil Code. Teach first programming steps in Pencil Code and directly transfering the knowledge to "classical" programming.

So what about starting with translating help texts and basic resources? And see in which contexts this works.

Then we could have the larger discussion about how to properly teach programming to non-english speaking children.

[allan-simon]

@luebken +1 , translating that should not be blocked by "what other things will be translated"

@davidbau yes i was imagining something like that for 2) sorry if i didn't expressed myself clearly. as for 1), for my selfish point of view of teaching an homogenous class or one student at a time, i don't think there's a need for software to be compatible interlanguage. after of course if we want to permit people to share code in some kind of "global village" then it is indeed not wanted.

[xinan]

Instead of translating the whole library, how about adding a layer in between the editor and the interpreter to map those non-English code to their English version?
For example, for the Chinese version, in the editor it appears:

向右 90

But when you hit Run, it actually gets processed to

rt 90

before it gets executed.

Then for each language we just need to have a mapping table. There will be some work to do with error handling though.