Default charset for Eclipse projects

[ptalbot]

What default charset do you use for your workspaces?

Right now all the projects of Servoy 5.2 uses a setting for 'Text file encoding' that is 'Inherited from Container' - in my case it is UTF-8, but I've seen some properties that were not correctly interpreted, like the signature in the Splash window where the copyright symbold appears as a square character if I launch it in debug mode, when launched out of Eclipse, it appears correctly.
This is why I think that your workspace settings are different and I wonder if that could not cause problems elsewhere.

Any ideas?

[rgansevles]

Patrick,

I have replaced all copyright characters with their unicode notation so that all source is in ascii.

Thanks for noting,

Rob

[ptalbot]

Thanks Rob, but actually I was more worried about other places where this could break, so I wanted to know which charset to use.

[jcompagner]

most of the time the type of file defines how it does store the contents,
like java propertie files should be: ISO 8859-1, xml files define it in there headers tag.
java files are not specified to one encoding, they take the system encoding. This is in my eyes a mistake of sun, they should have fixed it to UTF8
So for java files it is better to use plain ascii...

[ptalbot]

OK then. Thanks!
But I think I will use a special instance of Eclipse for Servoy sources if it is to use ASCII, because if I set it as Eclipse default all my other projects (currently using UTF-8) will be corrupted when I edit the files.

[jbader]

java propertie files should be: ISO 8859-1

I actually was going around in circles with this recently, as I was not aware that .properties files were required to be latin-1. I had sent my solution's i18n to a translation service, and got back a UTF-8 encoded file. When I brought this into Servoy, the first time I opened the i18n editor and saved/updated/added, my high-ascii (accented latin in my case) was changed to unicode literals. This was probably a result of Servoy implementing some version of native2ascii? The trouble I ran into though was that b/c the file was not latin-1, the native2ascii conversion of the high ascii chars to unicode didn't work properly, and I ended up with a completely defunkt .properties file. What I ended up having to do was run native2ascii at the command line on the utf-8 file and pass in a flag that told it that the file was actually utf-8. When I did this the high ascii was escaped properly and then when I brought it into servoy all was well.

I think that in order to prevent other's from learning all of this the hard way as I did, perhaps Servoy can run a charset detect on the .properties file and pass the gotten encoding as a flag to its native2ascii implementation? This will make it somewhat dummy proof.

Just an FYI/sharing my experience

[ptalbot]

Yes, properties should be ISO-8859-1

But Jeff do you mean you still don't know how to escape your utf files by hand?
I've perfected along the years the use of \u00E9 and \u00E8 in French

For native2ascii, I don't thing Servoy is doing anything. Eclipse does.

[jbader]

But Jeff do you mean you still don't know how to escape your utf files by hand?

Well I could do it I suppose by hand if I had a table of escape sequences for all the high-ascii in Spanish and Brazilian Portuguese (which I guess I do on the Mac), but the problem for me was that I *didn't do it originally* and then had a couple thousand values to deal with in a utf-8 file after the fact. The problem here was really my ignorance; not knowing the .properties had to be ISO-8859-1, doesn't have a whole lot to do with Servoy itself.

[jbader]

But Patrick I'm not convinced that Servoy had envisioned it's developers writing the escape sequences directly into the i18n editor in the first place. I say that because there is some implicitness that happens if you pay attention...

The editor itself appears to be UTF-8. You can see this because high-ascii (whcin in the .properties file is escaped) is displayed as its character in the list of i18n values and in the editing boxes themselves within the i18n editor window. You can type high ascii characters directly into the i18n editor. As soon as you save/add/update, Servoy (or eclipse?) escapes the high ascii chars and writes the proper escape sequences to the .properties file (assuming your .properties file is ISO...).

So really (at least in my experience in 5.1.x) you don't have to think about the escape sequences, or type them into the i18n editor.

[ptalbot]

But Patrick I'm not convinced that Servoy had envisioned it's developers writing the escape sequences directly into the i18n editor in the first place. I say that because there is some implicitness that happens if you pay attention...

The editor itself appears to be UTF-8. You can see this because high-ascii (whcin in the .properties file is escaped) is displayed as its character in the list of i18n values and in the editing boxes themselves within the i18n editor window. You can type high ascii characters directly into the i18n editor. As soon as you save/add/update, Servoy (or eclipse?) escapes the high ascii chars and writes the proper escape sequences to the .properties file (assuming your .properties file is ISO...).

So really (at least in my experience in 5.1.x) you don't have to think about the escape sequences, or type them into the i18n editor.

I was only joking