Re: Request for review on German localizations for icedtea-web

[Jacob Wisor]

Hello Jiri,

I have tested your patch. Here are some of my notes regarding it.

"Jiri Vanek"<jvanek at redhat.com> wrote:
> On 02/22/2013 03:04 AM, Jacob Wisor wrote:
> > Hello there!
> > 
> >>> Jacob, Alexander - in your emails you have come to conclusion, that there
> >>> are errors - both lexical
> >>> and semantic in original English properties.  If you will have spare time,
> >>> do you think - as
> >>> separate patch, you can fix them?

As I have seen, you have adjusted the relevant message about absolute paths. It is still kind of bothering me because it suggests that there are only two operating systems on this planet. ;) I would rather reformulate it so that it just gives a hint that the specified string is not an absolute path, regardless of its possible form on any given OS. I understand that you want to provide a helpful message to the user but in the world of Java we are probably left with nothing else than just saying that it is not an absolute path, or having specific messages for specific OSs (what I would not advise to do).

Btw, please note that Linux and Windows are names so that the have to begin with a capital letter in English. Further more Windows is a registered trademark and as such there has to be some notion of it in the product, documentation, or has to have an adjacent trademark symbol when mentioned. AFAIK this especially applies to US law but to many other jurisdictions as well. And, for the sake of completeness it should probably be stated as "Microsoft Windows". So as far as that is concerned, I would rather leave out mentioning explicit OS's names.

There are even more trademark issues in the Messages.properties file, but I am going to comment on them in my proposals later.

> >> *** OK, I can try to have look, but generally sometimes the source is quite
> >> confusing and the translator can only guess, what is the correct meaning.
> >> Rather the developers themselves should see our comments about unclear
> >> parts and rephrase them so that the meaning is clear...
> > 
> > I can post some proposals for the messages in question. It should not be really that much, but a few of them are utterly confusing indeed.
> >   
> > Btw, as I already wrote to Alexandr, I was able to compile at least NetX on Windows today, so I can see the German and Polish translations in the context of the running application. This helps a lot.
> > Having done that, I have also found a minor bug (I did not check whether it has been already reported) in regards to one message on Windows. I have already commented on it. It is the message that refers to absolute paths that must begin with a /. Though absolute paths on Windows do not begin with a / (it is a forbidden character anyway), icedtea-web assumes the (semi-) correct default value. The default value is an absolute Windows path beginning with a drive letter or device. Icedtea-Web still complains about it that it does not begin with a / and prints an error message about the value. It should be easy to fix.
> > 
> > Anyway, I will post an updated localization patch ASAP.
> > 
> > Regards,
> > Jacob
> > 
> 
> Wou!
> So there will be also Polish trnaslation?
> /me amazed :)
> 
> And I'm happy netx is working on windows !o)
> 
> I'm attaching patch to window paths. If you will be able to test, it would be nice (As there is no
> win machine around :( ) I will send [rfc] for it as separate email later.

So as already mentioned, I have tested it. It compiles and runs. You can have a look at the output log it produces. There are still some cases where the validation does not take effect.
Btw, after reviewing your code I would rather suggest to use java.io.File.isAbsolute(). Please do not try using your own regex pattern to test that, since absolute paths on Windows NT can be even more complex beginning with "\\?\" or "\\.\". There are some further rules to that, so it is probably safest to rely on the work that has already been done by Sun or Oracle. It is even more advisable to use java.io.File.isAbsolute() because every platform specific JRE implementation has their own implementation of it and thus knows best how to check for absoluteness. This way, you do not have to even check for a specific OS before, as you do in boolean net.sourceforge.jnlp.config.BasicValueValidators.canBeWindows(java.lang.String).

You do not need to bother about Windows stuff for now - unless you really want to. I am going to build IcedTea-Web for and on Windows anyway. There are even more flaws that should be fixed for Windows. Especially determining the user's and application's data files and folders should be fixed for Windows. So I am going take a look into that.

Thank you for the patch and cheers! :)
Jacob
-------------- next part --------------
A non-text attachment was scrubbed...
Name: netx.log
Type: application/octet-stream
Size: 5908 bytes
Desc: not available
Url : http://mail.openjdk.java.net/pipermail/distro-pkg-dev/attachments/20130301/9d36c1ac/ISO-2022-JPBbmV0eC5sb2c.log