Idea

[DWand]

An idea of the manager

[DWand]

firehist wrote:

@PascalPrecht It's going well! :) I can push on it, I just have no access to "settings" of this project (for example to edit the title + description GItHub). That's not really a problem.

BTW, I think about an other application "extra angular-translate" to manage the translations dictionnary as well as now. I use angularJS in my job (http://teads.tv and http://manager.teads.tv) and we have to internationalize the app with an external translator team.
The problem is that only devs can easily update these files.

What do you think about provide a rapid "translations manager" interface with persistent translation and manage delete/add/update throught grunt-angular-translate ? (I think about use deployd to store and angularJS to provide a rapid standalone manager per project).

[DWand]

PascalPrecht wrote:

@firehist That sounds good to me! Maybe you wanna talk to @DWand he also mentioned that he wanted to build at least an interface for managing translations.

[DWand]

DWand wrote:

@firehist, @PascalPrecht, yea, I want to build a GUI app to manage translations. I had even started, but was interrupted by work and academy. There are almost nothing done at the moment.

[DWand]

knalli wrote:

JSON based files, or what are you planning?

[DWand]

firehist wrote:

@DWand @PascalPrecht MMMh it looks like we'll rocks this together :) pretty cool.

First, we can create a repository to start discussions about global goal of this application.
Then we can create one issue per feature to speak more in details each one.
What do you think about it ?

First of all which name to handle this useful project ?
translate-manager
i18n-manager
...

BTW, two questions.
One is about angular-translate, there is any "dictionary" / "context" dev planned ? (I mean for example in module "settings" calla {{ "my_key" | translate:"settings" }} to avoid any replace of same word in different context)
Second is about interpolate system. It should be very nice if we can provide an interpolate service to provide some translation management, plug with our translate manager. For example display translation keys instead of language translation and show a popup with link to edit specific language. What do you think about it?

@knalli I think it should be good to use deployd or similar tools to provide a persistent data and allow us to do more funny things faster. You should be able to import/export JSON files, PoEdit files.

// Import "language JSON files"
// en_US.json
{
  "MY_TRANSLATION_KEY": "MY_LANGUAGE_TRANSLATION",
  "MY_TRANSLATION_KEY_01": "MY_LANGUAGE_TRANSLATION_01"
}

// Import "keys JSON files"
// i18n_keys.json
{
  "count": 2,
  "updated": "Jan 21 2014 13:36:47 GMT+0100 (CET)",
  "keys": ["MY_TRANSLATION_KEY", "MY_TRANSLATION_KEY_01"] 
}

etc.

[DWand]

tspaeth wrote:

👍

In the last enterprise project, we had the translations as java properties - files. That transformation to JSON btw is quite easy. Was just some little sources on the Java side (I couldn't find an easy mapper for properties2json in the JS ecosystem...).
Just to mention: Integration of this might also be a nice feature :-)

[DWand]

DWand wrote:

Hm... It seems that I thought in completely different direction :)

I thought to use sqlite3 as a persistent translations holder to be able to use it independent from the language used in development.

Also I thought to use node-webkit as an environment to be able to use web technologies in development, since all people here are familiar with HTML, CSS and Javascript. I have already used node-webkit once to develop app for my course work (but I used it to connect to SQL Server, not sqlite3). This solution seems to be completely cross-platform, but also has some disadvantages.

UPD:
I can also help a little with Java, C# or C++. But I'm not sure if I'm good here.

[DWand]

knalli wrote:

Regarding properties: Well, you have to support some encodings as well as the key: value syntax and the key=value syntax. Additionally, you can put the values in quotes. Said that, it it quite more easier in Java than in JavaScript.

So, if we are speaking in a Java context.. a simple maven plugin or a servlet filter should be fine?

[DWand]

DWand wrote:

Ping @firehist @PascalPrecht @DanielaValero

[DWand]

tspaeth wrote:

@DWand Yes, the solution is different, but I was just thinking from the practical perspective out of a dozen of projects. Depending on who translates, when and who checks - a direct translation table mostly will not work as even translation texts might be quality assured by the business dept. before deployment. Therefore mostly the conversion to the app deployment has been a technical task, but the creation of the translation texts was a task of probably external people having limited or no access to the app infrastructure.

Therefore something like the po-file-gettext-editors come out of that. Or even excel-tables to be parsed :-). Usually the KISS principle. But if someone manages a google calc style HTML/AJS editor for moving and editing in "translation tables" (whatever format they are and from whatever source they come from) - could be a killer app though :-)

[DWand]

DanielaValero wrote:

I am arriving late to the thread. After reading your comments, there are two points which are not clear to me,

1. This translations manager is aimed at translators who are not developers, right?

2. it should generate a json file(s) to work with angular-translate, right?

I think that before speaking about the architecture, we should define clearly what we want or need. (which is not clear to me after reading the evolution of the conversation)

What is the problem/necessity we want to solve? I think is how can we integrate into the angular-translate workflow those translators, who are used to working with poEdit or linguan, generating .po, .xml or .xls.

Or maybe I misunderstood and you want to build another translator manager, which generates output for different projects written in different architectures. In such case, then ignore me :)

[DWand]

tspaeth wrote:

@DanielaValero exactly this questions I see also - great post though!
So let's stay tuned what the inventor wanted to target the app to/for/at. ;-)

[DWand]

DWand wrote:

Hm, don't event thought about tables. I was think about something like this:

As for application, I thought to create app for:

1. Translators can manage translations
2. Developer can generate json files to insert to app
3. Provide some stuff like check for data consistency (all keys to be filled, keys-sets are equal for every language and so on)
4. Maybe import other formats (xls, csv, po, whatever) and export to angular-translate's json

[DWand]

tspaeth wrote:

@DWand From the usability perspective, this includes a lot of clicking, typing, submitting for the translation of e.g. 50 variable texts - or not?! Technically, this looks good, but I would say the translation speed with this concept is slower compared e.g. to direct JSON or Java-properties style aka plain text ;)

[DWand]

DWand wrote:

@tspaeth, which is why I agree that "Programmer should never create UI" 😄
The main idea is to visualize that some keys have not translations or some languages has untranslated keys (different color, for example, or special icon, etc.) This is both impossible to see in plain json and uncomfortable to analyze in terminal output.

This is what I thought one or two month ago being alone. Now we can combine all our thoughts and create much more useful and better app.

The only thing I'm sure now is that relational database is the best thing to keep translations because it ensures data consistency out of the box. And also it could be moved wherever people want. Moving DB to some server enables team to work with translation together at the same time from different locations. If I understood right @firehist mentioned something about it.