Bug 579824

This bug needs to be added to the "known issues" section of the release notes.

A warning box must also be added to the install guide.


+++ This bug was initially created as a clone of Bug #579766 +++

Description of problem:
User can enter unicode entities in the forms on the Installer UI, but these are not handled correctly.  The resulting mangled text is what gets obfuscated, versus the unicode entities.

Version-Release number of selected component (if applicable):


How reproducible:
Every time

Steps to Reproduce:
1. Unpack and start server.
2. "$JBOSSON_HOME/bin/generate-db-password.sh 你好"; observe results
3. In the password field, enter the characters '你好' and initiate install
4. After installer commences -- and possibly failed -- view results in rhq-server.properties
  
Actual results:

[root@core-01 bin]# ./generate-db-password.sh 你好
Encoded password: 68f725778bb36d3b
[root@core-01 bin]# grep "rhq.server.database.password" rhq-server.properties 
rhq.server.database.password=55a670d2613a05d127d3d58e7aa8e158df8592078de921bc


Expected results:

Password should match

Additional info:
I think what is happening is that the unicode characters, upon submit, are getting translated into their html entities -- so "你好" becomes "你好", which is in turn obfuscated.

We need to fix the UI to allow unicode input (preferable but intensive) and/or document whether we'll allow unicode characters in passwords.  As it stands, I suspect Oracle and perhaps Postgresql do both support it.

--- Additional comment from ccrouch on 2010-04-06 10:33:17 EDT ---

Corey, can you check how we handle unicode passwords in 2.3.1? From your analysis if the installer really is converting these to html entities before storing them in rhq-server.properties, then unicode passwords will never have worked.

If thats the case, then I'm fine with making this an RFE and doc'ing this restriction for 2.4

--- Additional comment from cwelton on 2010-04-06 11:28:13 EDT ---

I don't suspect this is a regression - in that sense, I agree that it's not necessarily something we need to 'fix' this release... however the new obfuscation tool introduces disparity.  As users previously hadn't been provided a clean route to enter unicode passwords, it was not as much an issue.  generate-db-password.sh, however, does allow it, so we're need to manage this in some way or another.

I am fine with a doc note.  I can mark this FutureFeature and get a docs issue written up.