Unicode xgettext problems with _() over several code lines Topic is solved

[hgiritzer]

Is there any recommendable practice to spread long _() sentences over several code lines without getting troubles with the Unicode compiler and/or xgettext?

MSVC in Unicode builds complains about strings like:

Code: Select all

    _("text1" \
      "text2")

saying "mismatched wide string concatenation".
Putting them as:

Code: Select all

    _("text1")
    _("text2")

make them appear as two separed strings to xgettext
(but should be one as they belong together)...
doing:

Code: Select all

    _("text1\
       text2")

would solve both problems above but the resulting string would be

Code: Select all

"text1         text2"

instead of

Code: Select all

"text1 text2"

.

Maybe the only simple way to solve this might be:

Code: Select all

_("text1 text2")

even if the string goes > 80 columns...

[chris]

Hi,

did you try

Code: Select all

_("text1"
"text2");

?

Chris

[leio]

I did that in my project. VS2005 was bitchy about it in unicode builds.
Currently I ended up doing

Code: Select all

_("text1"
wxT("text2")
);

But I fear that xgettext won't parse that as I'd wish it to. My project isn't yet set up for i18n.

[frm]

No way.

I've tried both your suggested way

_("text1"
"text2");

but this still gives the C2308 error:mismatched wide string concatenation

And

_("text1"
wxT("text2")
);

works but is not recognized at all by xgettext.

I'm really unsure how to handle this...

[hgiritzer]

How do others work with Unicode and xgettext (poEdit)?

Do you write your _("xxx") items always into one line of code, regardless of how long this code line then might become?

Is there no practicable way to split long _("xxx") items onto two or more lines of code to keep the code somewhat readable?

[cecilio]

Hummm! In my project I have the following code:

wxString s = _("text 1"
"text 2"
"text n");

I thought it was a good way to solve the problem you mention and I didn't get compiler errors. So I checked my code and I realized that currently I am not generating a Unicode build.

I remember having generated Unicode builds with the code like this:

wxString s = _( _T("text 1")
_T("text 2")
_T("text n"));


With no problems. I will try to do some testing and tell you my results.

[hgiritzer]

Thank you for your efforts.
Please keep in mind, that we need a solution for Unicode build in combination with xgettext translation (poEdit).

It's easy to use only one of them, but problematic when using both together.

[cecilio]

As promised, here are my results:

To me the best solution is:

Code: Select all

wxString s = wxGetTranslation(
     _T("line 1")
     _T("line 2")
     _T("line n")
);

It is important to use 'wxGetTranslation() as macro _() DOES NOT WORK. You get error 'Symbol LL not defined'

[frm]

wxString s = wxGetTranslation(
_T("line 1")
_T("line 2")
_T("line n")
);

I've tried this but it's not recognized by xgettext... even using the -kwxGetTranslation option it doesn't 'see' the strings...

[hgiritzer]

Well, it seems to me that it is the best practice to write the whole sentence into a single long line of code.

The code might become a bit harder to read, but as we can see in this discussion there is no 100% working alternative for this.

[cecilio]

I've tried this but it's not recognized by xgettext... even using the -kwxGetTranslation option it doesn't 'see' the strings...

Sorry! This morning I didn't do tests with poEdit, I just tried find a solution to compile. I did not take poEdit into account

So, probably hgiritzer answer:

the best practice to write the whole sentence into a single long line of code.

is the best answer for now

I was doing more test this evening but I did not find any better answer! poEdit does recognizes wxGetTranslation (when instructed) but the problem seem to be in the inside _T() macro.

I do not have more ideas. If any one finds a solution, please, post it.

[cecilio]

This is my best attempt. It is better than a single long line:

_("This program is free software; you can redistribute it and/or modify it \
under the terms of the GNU General Public License as published by the Free \
Software Foundation; either version 2 of the License, or (at your option) \
any later version.");

The only point to take into account is that continuation lines must start at column 1.

[Cryogen]

Hi Guys,

I need a solution to this, too, and this doesn't work for me. I have:

const wxString s = _("some text");

and this

const wxString s2 = _("some much, much, much \
, much, much, much, much, much \
, much, much, much longer text");

in a header file. A translation exists, the catalogue is loaded and other strings are translated but these refuse to budge. Any ideas?

Thanks.

[doublemax]

This probably doesn't work because constant and static variables are initialized before the actual code execution starts, so translation files are not loaded yet.

[cecilio]

Hi,

quote]This probably doesn't work because constant and static variables are initialized before the actual code execution starts, so translation files are not loaded yet.[/quote]

Yes, it doesn't work.

My approach to solve this is to initialize all const strings in the object constructor. Something as:

Code: Select all

myobj::myobj()
{
    InitLanguageDependentStrings();
}

void myobj::InitLanguageDependentStrings()
{
    static fInitialized = false;
    if (!fInitialized)
    {
        fInitialized = true;
        m_string1 = _("String to translate");
    }
}

Cecilio