[opencms-dev] Problems with labels in workplace_ru.properties
Paul-Inge Flakstad
flakstad at npolar.no
Wed Sep 30 21:23:03 CEST 2009
Hi Christian
It is indeed horrible, just imagine how much easier it would have been to use utf-8 (or at least be able to specify the encoding)..! No idea what Sun has been thinking (or not thinking at all) all the time they haven't addressed this issue. Yes, the native2ascii thing is fine, but damn what a messy "solution" that is...
I agree that the format of the .properties files have a nice simplicity to them, and I don't see anything wrong with them other than the encoding issue. But of course, XML is a strong contender if there's ever going to be a change.
However, that would require a lot more re-writing than just switching over to UTF-8 while keeping the format, I would assume (since you'd have to change both the format and the encoding)? As far as I can tell from what (little) I've read on the subject, UTF-8 is largely "backwards compatible" with ISO-8859-1, so in my mind it seems like a pretty straight-forward swap for Alkacon - if they ever wanted to. Eventually, I guess it's up to Sun if they want to change the specification. I just thought it would be cool if Alkacon stepped up and set an example, OpenCms being a multilanguage CMS and all.
I was considering whether or not to extend/modify the OpenCms core in order to read .properties-values as UTF-8 (probably something like you've done already), but for now, my method is sufficient, so I think I'll put that idea on ice for now.
Thanks for your reply, comforting to know I'm not the only one who's been struggling with the properties encoding.
All the best,
Paul
________________________________
From: opencms-dev-bounces at opencms.org [mailto:opencms-dev-bounces at opencms.org] On Behalf Of Christian Steinert
Sent: 30. september 2009 18:55
To: The OpenCms mailing list
Subject: Re: [opencms-dev] Problems with labels in workplace_ru.properties
Paul-Inge Flakstad wrote:
I was a bit mistaken in my last post. By specification, all .properties files are Latin-1 encoded, and when loading from (or saving to) a stream, ISO-8859-1 character encoding is used. All characters that cannot be represented in this encoding must be Unicode escaped.
This is probably old news for the more experienced developers, but to me, it's news. I'm glad to finally have learned the cause of my "mysterios" encoding problem, but bewildered adn confounded by the facts... Why why WHY???
At some point I hat also stumbled over this one - horrible, isn't it? In the end I just wrote my own property file loader method that pushes the property data string through the native2ascii implementation from GNU Classpath and then loads the resulting property bundle.
I don't know why Sun never fixed the property file spec by adding an optional encoding signature, but somehow they never did.
Further reading: http://www.thoughtsabout.net/blog/archives/000044.html
Guys at Alkacon: Any possibility that, in a future release, you would consider disregarding the specification for .properties files, and use UTF-8 instead? In my opinion, OpenCms would become more user-friendly if you did - at least when dealing with multilanguage sites that should support use of characters not specified in the Latin-1 set.
Maybe it would be best to switch to XML-based resource bundles at some point. I like the simplicity of the old-fashioned property format, but since their encoding behavior is definded like that, it's maybe best to leave them aside rather than bending the spec oficially.
Best Regards
Christian
Cheers,
Paul
-----Original Message-----
From: opencms-dev-bounces at opencms.org<mailto:opencms-dev-bounces at opencms.org>
[mailto:opencms-dev-bounces at opencms.org] On Behalf Of
Paul-Inge Flakstad
Sent: 30. september 2009 15:42
To: The OpenCms mailing list
Subject: Re: [opencms-dev] Problems with labels in
workplace_ru.properties
Self-replying :)
Given that my assumptions are correct:
The workplace_xx.properties files are read during the
workplace initialization, using the default encoding of the
JVM, which typically depends upon the locale and charset of
the underlying operating system.
In my case, the workplace_ru.properties file is read as
ISO-8859-1, and as a result, no strings fetched using
CmsJspActionElement#label(String) make any sense - it's all gibberish.
The solution is something along the lines of this:
public String labelUnicode(String key) {
String jvmDefaultCharsetName =
Charset.defaultCharset().displayName();
try {
return new
String(this.label(key).getBytes(jvmDefaultCharsetName), "UTF-8");
} catch (java.io.UnsupportedEncodingException e) {
return new String("[Default label: " +
this.label(key) + "]");
}
}
This seems to be working just perfectly. I don't have to
think about character encoding since utf-8 is the default
encoding all over OpenCms (I can just leave the
"content-encoding" property blank), and I can even mix
special characters from different languages all in one
.properties file.
I would even propose to add a method, like the one suggested
above, to CmsJspActionElement. (I'm pretty sure I attempted
every single possibility within OpenCms to get the correct
strings from my workplace_ru.properties returned, using the
"standard" label(String), but never got anything but strange
symbols. If someone for some reason _needs_ to have their JVM
default encoding set to ISO-8859-1, while at the same time
supporting a multilanguage OpenCms system, there seems to be
no method native to OpenCms that enables getting correct
russian (for example) strings from the workplace_xx.properties files.)
Cheers,
Paul
PS: I know I should propably set the JVM default encoding
manually to UTF-8 instead, but I'm unsure of any possible
side-effects. So until then, this is a pretty decent workaround.
-----Original Message-----
From: opencms-dev-bounces at opencms.org<mailto:opencms-dev-bounces at opencms.org>
[mailto:opencms-dev-bounces at opencms.org] On Behalf Of
Paul-Inge Flakstad
Sent: 29. september 2009 22:57
To: The OpenCms mailing list
Subject: [opencms-dev] Problems with labels in
workplace_ru.properties
Hi all
In one of our multilanguage sites, Russian and English
content is mixed. Everything's working as expected (since
we're using UTF-8 encoding), but all the labels read from
workplace_ru.properties, using
CmsJspActionElement#label(String), is just gibberish...
As a workaround, I created my own label(String, Locale)
method that does nothing more than simply read the value
straight out from the workplace_ru.properties file. When
using this method to access the labels, everything is OK, but
the .properties file is "parsed" upon each invocation, so
it's not desirable to keep using it.
I've tried this:
Set the HTML charset to utf-8.
Set the JSP pageEncoding to utf-8.
Set the OpenCms <defaultcontentencoding> to utf-8.
I also checked the HTTP response header, it also says utf-8.
Also, I been experimenting with different constellations of
encodings (including Cyrillic iso-8859-5), but to no avail.
Can anyone please provide some insight?
(Just so there's no mistake, I'm reading the labels from the
.properties file to use them as text on a web-page, not in
the OpenCms workplace. Things like "Photo:", "Published by "
and alike.)
Cheers,
Paul
_______________________________________________
This mail is sent to you from the opencms-dev mailing list
To change your list options, or to unsubscribe from the list,
please visit
http://lists.opencms.org/mailman/listinfo/opencms-dev
_______________________________________________
This mail is sent to you from the opencms-dev mailing list
To change your list options, or to unsubscribe from the list,
please visit
http://lists.opencms.org/mailman/listinfo/opencms-dev
_______________________________________________
This mail is sent to you from the opencms-dev mailing list
To change your list options, or to unsubscribe from the list, please visit
http://lists.opencms.org/mailman/listinfo/opencms-dev
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://webmail.opencms.org/pipermail/opencms-dev/attachments/20090930/1a0898c2/attachment.htm>
More information about the opencms-dev
mailing list