Observations on LocalizedDisplayScriptTest

[Kathey Marsden <km...@sbcglobal.net>]

Recently I have been looking at this test because I was debugging a jvm 
specific failure and I noticed the following.

1) The system property derby.ui.codeset does not have any effect because 
the script is run with runScript which takes its own encoding parameter. 
The file always gets read in US-ASCII even though the file format is 
Cp932. (It uses the constructor for ScriptTestCase that sets the 
encoding to US-ASCII.)
2) The test seems to be testing  LOCALIZEDDISPLAY ON which would only 
affect date and time output format and not have anything to do with the 
special characters in the test.

It seems like it might make the test easier to manage if we took out the 
ncharc column of the table and the special characters, took 
derby.ui.codeset out of the test and just had an ascii sql file and master.

Thoughts?

Kathey

Re: Observations on LocalizedDisplayScriptTest

[Kathey Marsden <km...@sbcglobal.net>]

Knut Anders Hatlen wrote:
>
> It might be OK to remove the non-ascii characters from that test, but I
> still think it's worth testing that derby.ui.codeset works. Do we have
> other tests for that currently?
>
>   
The following tests use derby.ui.codeset:

i18n/LocalizedAttributeScriptTest.java
i18n/LocalizedDisplayScriptTest.java
tools/SysinfoLocaleTest.java

I assume LocalizedAttributeScriptTest has the same problem as 
LocalizedDisplayScriptTest case, (but I haven't checked.) Since they use 
runScript the derby.ui.codeset is not used.  SysinfoLocaleTest tests the 
ouput of Sysinfo in UTF-8 format. It seems like it would be good to file 
a Jira to test derby.ui.codeset.  I think it confuses the issue to try 
to combine that testing with LocalizedDisplayScriptTest and we should 
use a codeset supported by all the jvms, not Cp932 which is only 
supported by IBM JVM's.   I must say also that I am a bit confused about 
the scope of derby.ui.codeset, especially when running scripts.  It 
might be good as part of the exercise of writing a new test, we beefed 
up the documentation some too.

Thanks

Kathey

Re: Observations on LocalizedDisplayScriptTest

[Knut Anders Hatlen <Kn...@Sun.COM>]

Kathey Marsden <km...@sbcglobal.net> writes:

> Recently I have been looking at this test because I was debugging a
> jvm specific failure and I noticed the following.
>
> 1) The system property derby.ui.codeset does not have any effect
> because the script is run with runScript which takes its own encoding
> parameter. The file always gets read in US-ASCII even though the file
> format is Cp932. (It uses the constructor for ScriptTestCase that sets
> the encoding to US-ASCII.)
> 2) The test seems to be testing  LOCALIZEDDISPLAY ON which would only
> affect date and time output format and not have anything to do with
> the special characters in the test.
>
> It seems like it might make the test easier to manage if we took out
> the ncharc column of the table and the special characters, took
> derby.ui.codeset out of the test and just had an ascii sql file and
> master.
>
> Thoughts?

Hi Kathey,

I have posted similar observations earlier in comments to DERBY-1726. I
reopened the issue now to get those comments addressed, and I also
pointed to this thread.

It might be OK to remove the non-ascii characters from that test, but I
still think it's worth testing that derby.ui.codeset works. Do we have
other tests for that currently?

-- 
Knut Anders