This Bugzilla instance is a read-only archive of historic NetBeans bug reports. To report a bug in NetBeans please follow the project's instructions for reporting issues.
ide log is saved in utf-8 on nb trunk, and multibyte chars in the logs are garbaged on ide and terminal. If this is specification of the ide log, please let us know.
the ide.log is not meant to handle multibyte chars, it's pure 7-bit ascii only
Date as Log Session in ide.log is garbaged in ja locale. After converting the log from utf-8 to euc-jp, the date specified by multibyte chars are able to be read properly on sol9. On redhat and windows, also need conversion from utf-8 to appropriate encoding to read the date of logs.
Created attachment 13763 [details] garbaged multibyte chars of Log Session
Created attachment 13764 [details] date correctly displayed after conversion
INFORMATIONAL *********** Exception occurred ************ at Fri Feb 27 19:09:16 JST 2004 java.net.ConnectException: XXXXXX <= this message which means connection was refused in ja has been saved properly in euc-jp on sol9.
Using Locale.ENGLISH to print dates of exception and log session, so the log contains just ascii-7 characters. That addresses the main problems with utf-8. If there are other messages in log file that cannot be read, they have to be solved separately, I am affraid, as the intention of log file is to be in utf-8 to be "readable" on any system. I think it is a bit late to fix this in release36, it has already entered high resistance, but if you really want the issue to go there, increase the priority. /cvs/core/src/org/netbeans/core/NbErrorManager.java,v <-- NbErrorManager.java new revision: 1.55; previous revision: 1.54 /cvs/core/src/org/netbeans/core/TopLogging.java,v <-- TopLogging.java new revision: 1.43; previous revision: 1.42
Date of log session in ide.log are now not localized on nb trunk 200303081900 as following. >Log Session: Tuesday, March 9, 2004 12:57:37 PM JST The log session date was used to be printed using system locale on nb351. Could you fix this message as it was used to be? Related issue: http://dtsw.eng/cgi-bin/bugtraq_showbug?bugid=4938662
Excuse me. Please ignore the bugtraq issue which I submit in previouse comment. Related Issue: http://dtsw.eng/cgi-bin/bugtraq_showbug?bugid=5006181
Let me disagree (ofcourse you can as well and reopen the bug). The log file is not meant for end users, but more for us, as a source of information when a bug is reported. As such it does not have much value to have localized messages in it, as we would not be able to read them anyway. That is why I think that logging in english is reasonable and that the previous behaviour, when localized texts appeared in the log, was a bug. I would like to thank you for pointing us to this bug and I would suggest to report another if you find that some other part of the system is printing localized texts into log file.
You are right, and I agree with that ide.log file is not only for end  users, but for developers mainly. The problem was, as you mentioned, the date as log session was printed in UTF-8, and there was a possibility to mix up with multibyte message which would though from other product using system locale. Now it is printed in ascii and garbaged characters as log session no longer exists. Therefore, I am closing this bug as verified, but based on following reasons, I still think the multibyte messages in ide.log is better to be printed in system locale, and need to be translated basically. Need to be localized because there are 4W reasons that determine the purpose of ide.log as not for developers only. - Who is the first person to trace bugs which needs to be reported? - end user. - Where end user will send the bug report to? - Localized user support site first in the most of case. As you see by selecting "Support and Docs" and "Contact the NetBeans IDE Team" from ide menu under "help", the URLs lead user to reach one of localized site using system locale. This URL might be changed, but this function uses to reach localized site now so far. - Why end user submit the bug? - Because end user find it during their operations in their each system locale. - What kind of bugs should be reported is user's decision, and the messages are better to be translated for their comprehension. Therefore, I would like to say basic message is better to be localized using system locale for end users. Need to use system locale for the encoding of multibyte messages in ide.log because the message which is thrown by other product use system locale as the tendency in ide.log, so there should be equality for the localized message encoding of netbeans and others. Otherwise the encoding type will be mixed with netbeans and others. I agree if there is a specification of ide.log that says every multibyte chars should be in UTF-8 or something else, but considering about other message which might be thrown by not only netbeans, the multibyte message should be in system locale for the encoding equality. However, issue appeared at the first time is no more occurred, so I will close this bug as verified. I didn't have a such a view at the first time that ide.log is for developers, and I agree with it now, but still please consider to use user's locale for date stamp and translate it by the reason of why I submitted this issue explained above. Any thoughts are greatly appreciated.
Following part of messages in ide.log comes from Bundle.properties and will be localized in UTF-8. The message appearance depends on the user's operation, and I presume there is more message coming from Bundle. Let me desagree now with your point of view that all message in ide.log should be in ascii because fixing all message in ide.log as NO18N message is functional regression. - NetBeans IDE Dev (Build 200403081900) - New modules: - Clients of obsoleted Open APIs are encouraged to remove this dependency. See http://www.netbeans.org/download/dev/javadoc/OpenAPIs/org/openide/doc-files/upgrade.html#3.5i-sep-I - Turning on modules: - Installing new module: Open APIs Support - Turning on modules: If there is a specification that encoding type of multibyte chars in ide.log should be UTF-8, please let me know, but all messages thrown from other product have used system locale for the encoding, and the specification cause other problem as bugtraq issue 5006181. - All mulbitibyte chars in ide.log should be in UTF-8 as the specification? - If that's so, what do you think about the encoding-type equality for other product and netbeans? All messages which is able to be localized should use system locale for the encoding, otherwise the encoding type will be mixed with other product, and using UTF-8 only seems functional regression for international end users.
Quote from the bugtrack: The main purpose of the log file is providing diagnostic information to support and Sun IDE developers. Log file should not be exposed to regular users, unless there is a critical problem and the user needs to report defect back to Sun. It is not expected that the user would understand the log file. I believe that the ide.log should not contain any non-ascii characters. So does everyone I could talk to. In my opinion none of the messages that is printed into log file should be localized. If we found such message, let's fix it to not be localized. I've been asked to provide specification. I do not know of any. If you show me specification that says that content of ide.log should be localized and in default system locale, I am ready to fix it. Until then I am closing this issue as wontfix.
I agree ide.log contains only ascii characters. In the development (Build 200403151900), the date stump is in English as follows: >Log Session: Friday, March 26, 2004 8:07:13 PM JST However in the 3.6 RC1 (Build 200403161736), the translated date stump is generated. Would you change the 3.6 RC1 to output English date stump in ide.log?
The fix has not been backported to 3.6. It still remains in current trunk.
verified-closed