[fpc-pascal] Lazarus, FreePascal, and translated compiler messages

[Stéphane Aulery]

Hello Sven,

Le vendredi 08 mai 2015 à 02:00:06, Sven Barth a écrit :
> Am 08.05.2015 11:39 schrieb "Stéphane Aulery" <saulery at free.fr>:
> > With Frédéric Da Vitoria, we began to review the status of the translation
> > of the FPC. We concluded that it is very difficult to maintain and not
> very
> > useful because it is not implemented with ressourcestrings. We have not
> given
> > up to translate, but one wonders if it will not fit first FPC for better
> > support in the long term.
> >
> > I need your opinion on this issue, given that the interweaving with the
> > TeX manual is also very problematic.
> 
> There are IMHO three reasons it is done the way it is:
> - it has grown that way (probably from a time before FPC supported
> resourcestrings)
> - the IDs, the verbosities and the messages are located in one single
> location avoiding a drift of these
> - it is easy to generate the documentation as it's part of the same file
> 
> It might provide obstacles for translators, but it eases the work for the
> compiler developers and IMHO that is the main reason to keep it the way it
> is.

It's not the only problem:

- It is not possible to perform FPC in accordance with the language
  passed by the environment variables;

- This part of the manual is virtually untranslatable because msgdif
  tool does not synchronize the TeX part;

- Translations are never up to date and do not compile actually.

My goal is to improve the state of the compiler and the service to users
without complicating the work of developers and translators.

On this point I think the state of the compiler is not good for
developers, translators and users.

The developer must edit one file at a frightful codification
complicating hir life instead of a clean and structured code (including
errors of descriptions). Look what it did in the Microsoft environment.
It's not just the error messages but also their descriptions, ID and
level, which are accessible during code execution.

This is not available but tucked only in a manual, which is almost
untranslatable in the current state. In fact none of the translations
is updated because it is not maintainable. I even sent a patch to fix
ALMOST ALL languages because it does not even compile! So, who uses them?

Two reasonable solutions are possible to me:

- Finding a compromise that suits everyone, but it takes some work to do;

- If no one is interested, competent, available, and given that the intended
  audience is supposed to have enough English, we can remove these translations
  and mechanism complicating the compilation and clearly notify users that this
  part cannot be translated in the current state.

The compilation of fpdoc of ressourcestrings is also broken. So I ask
myself honestly, although I would have preferred that all this is
restored and translated, if the second solution is not by far the best
yet.

Regards,

-- 
Stéphane Aulery