Jump to content

  • Log In with Google      Sign In   
  • Create Account

None English Language Issues


Old topic!
Guest, the last post of this topic is over 60 days old and at this point you may not reply in this topic. If you wish to continue this conversation start a new topic.

  • You cannot reply to this topic
8 replies to this topic

#1 Stainless   Members   -  Reputation: 941

Like
1Likes
Like

Posted 25 May 2014 - 03:52 AM

I've written a mod tool for a community of flight sim addicts, but they are having problems running it.

 

Loading an aircraft only works on English language versions of windows. The aircraft are in a very complicated format. There are multiple mesh files per aircraft, some are in a text file format and some in a binary format. Naturally I use a TextReader and a BInaryReader as required.

 

It's coded in C# as a winforms application with XNA embedded as a renderer.

 

The exception seems to be the "Invalid number sequence" exception you get when you read a multi-byte character by accident when you want to read a byte.

 

I know this can happen on a call to PeekChar, so I deliberately avoided using that.

 

Does anyone know any other functions that can produce the same error?

 

The results are looking good for me but the whole project is a failure if I cannot get it working for everyone else.

 

2222460_orig.png



Sponsor:

#2 Hodgman   Moderators   -  Reputation: 30387

Like
6Likes
Like

Posted 25 May 2014 - 04:20 AM

I don't know if this is your actual problem, but I had a similar issue recently when trying to give my tools to an Italian. My issue was that when doing things like float.Parse, you have to pass it CultureInfo.InvariantCulture.
My data-files contained English-style numbers, and without passing that culture object, the Italian PC parsed them wrong.

#3 Stainless   Members   -  Reputation: 941

Like
0Likes
Like

Posted 25 May 2014 - 07:15 AM

On a float ?   Wow never thought of that. Guess I've got a lot of find and replaces to do. sad.png

 

Thanks Hodgman



#4 Matias Goldberg   Crossbones+   -  Reputation: 3399

Like
2Likes
Like

Posted 25 May 2014 - 08:37 AM

Yeah, to explain the problem, in some cultures "1.000,50" is read as one thousand with fifty cents; in another culture it has to be written as "1,000.50" (the comma and/or the point are different)



#5 unbird   Crossbones+   -  Reputation: 4977

Like
2Likes
Like

Posted 25 May 2014 - 09:13 AM

FxCop has globalization specific warnings. E.g. it will cough at int.Parse(someString).

#6 Stainless   Members   -  Reputation: 941

Like
1Likes
Like

Posted 25 May 2014 - 04:06 PM

Changing all my Float.Parse worked. (Thank you very much for that hint)

 

What I don't understand is why 1.0 cannot be processed regardless of language.

 

If one language uses 1,0 that's fine, but it shouldn't be a replacement for a 1.0, rather a new case to consider..

 

Supporting both is sensible, throwing an exception is just silly. Would you expect to write float x = 1,0; in a c++ file?

 

Without talking to you guys I would have never even considered Float.Parse as an issue. 

 

Guess I have one more thing to worry about smile.png



#7 Matias Goldberg   Crossbones+   -  Reputation: 3399

Like
1Likes
Like

Posted 25 May 2014 - 09:46 PM

This is called "locale". I don't know how C# exactly approaches the problem, but C/C++ has a locale simply named "C" which is used for C syntax parsing (i.e. 1.5 means 1 and a half; 1,5 is invalid)

Supporting both is sensible, throwing an exception is just silly. Would you expect to write float x = 1,0; in a c++ file?

All C/C++ files are parsed using the C locale; which means I expect only float x = 1.0f; to work
 

If one language uses 1,0 that's fine, but it shouldn't be a replacement for a 1.0, rather a new case to consider.

If you're parsing a text file written by a regular user (or generated content from the OS that also uses the system locale) "1.000,5" can be read as "one thousand and fifty cents", or as "one and zero zero zero invalid five".
"1.000" can be read as "one thousand" or as "one".

Just go to Control Panel->Regional Settings; and change the country you claim to be. Then open the calculator, open Excel (if they were already opened, restart them to make effect) see it for yourself.

In your particular case, you want the locale to be consistent, because you're reading data generated by you. Float.parse defaults to parsing data suplied by the user or by other apps following the locale.

#8 DvDmanDT   Members   -  Reputation: 893

Like
1Likes
Like

Posted 26 May 2014 - 03:48 AM

In Swedish, numbers are supposed to be written with a space between all thousand groups and a comma as decimal separator. I have a comma button on my numpad, not a dot one. Other countries use an apostrophe or similar for their thousands while English uses comma for thousands and dots for their decimals. I think it's great that the default is to use locale specific formatting, but it would have been much better if these format differences didn't exist. I wish they had standardised them instead.



#9 Stainless   Members   -  Reputation: 941

Like
1Likes
Like

Posted 26 May 2014 - 12:32 PM

I actually had a laugh about this today, now that the app is working of course, wasn't laughing before.

 

When Microsoft started pushing XML, I was totally against it. I could not accept the massive increase in file size that storing a text version of a number instead of the binary version created.

 

As usual, I was ignored and we all use XML all the time now. 

 

Now I find out that storing a number in a text format can produce massive internationalisation problems smile.png

 

Maybe it's just me, but I find that highly amusing.






Old topic!
Guest, the last post of this topic is over 60 days old and at this point you may not reply in this topic. If you wish to continue this conversation start a new topic.



PARTNERS