You are viewing a plain text version of this content. The canonical link for it is here.
Posted to derby-dev@db.apache.org by Mike Matrigali <mi...@sbcglobal.net> on 2007/05/31 00:03:04 UTC
norway specific failure in CollationTest2
I'll try to figure out what is going on with the CollationTest2 test.
I am not seeing it in my environment, and it does not seem to have
shown up in any of the publically posted nightly regression tests.
Can someone from Norway help, is it possible that 2 strings used in
the test, are to be treated as equal in Norway?:
"Wσgan" "Waagan"
The first string contains an unfamilar character (to my english
environment - I am not sure my mailer will even handle it correctly) as
it's second letter, and I am wondering if I should have used a unicode
escape for the data
instead. See the data string in the runDerby2670()routine of the
CollationTest2 test.
Is there a chance that in norway aa is supposed to be treated
as equal to the second letter in the first string listed? If so then
either order may be right, and the test is incorrectly testing for
one order or the other.
I tried to take the data from a donated test case and put it into
the junit test. It could be I messed something up on my side dealing
with a non english character in some step of the way (file handling,
code sets, editor, ...).
Rick Hillegas wrote:
> Julius Stroffek wrote:
>
>> Hi All,
>>
>> When I ran junit tests now for trunk I get 17 errors and 70 failures of
>> many different kinds.
>>
>> My junit console output with exceptions thrown is attached. Probably it
>> is something wrong with my test environment or so, but I have no idea
>> what happened. Are there any issues with upgrade tests due to a new
>> release? Probably this is not the case since these tests does not fail
>> in daily and tinderbox test runs.
>>
>> I always delete the contents of the current directory where I run tests
>> (which ends using as derby.system.home). I tried to checkout the source
>> again and start from a scratch but it does not helped.
>>
>> I run junit throught suites.All.
>>
>
> Hi Julo,
>
> I saw tons of upgrade errors in tests which I ran yesterday. Those
> errors disappeared after I synced my client to the mainline this
> morning. Now I am seeing the error which you are seeing in
> CollationTest2. Otherwise, the tests run cleanly for me.
>
> Regards,
> -Rick
>
Re: norway specific failure in CollationTest2
Posted by Øystein Grøvlen <Oy...@Sun.COM>.
Mike Matrigali wrote:
> Thanks, in my file what I am seeing is an 'o' with a tail on top-right
> Is that a valid norwegian character, and if so what is the
> proper unicode escape for it and where is it expected to sort with
> regard to aa.
>
That is not a Norwegian character. I am not an expert on this, but I
would think one reason for seeing a different character could be that
the character set (I am not sure that is the right word) of the program
you use to look at the file. If that is something else than Latin-1, I
guess you may see a different character.
--
Øystein
(first letter is O with slash)
Re: norway specific failure in CollationTest2
Posted by Knut Anders Hatlen <Kn...@Sun.COM>.
Mike Matrigali <mi...@sbcglobal.net> writes:
> Michelle Caisse wrote:
>> For a quick response, in Norwegian, a sequence of two 'a' characters
>> is equivalent to a single 'a'-with-a-circle-on-top character. Maybe
>> someone who both knows more about Norwegian and character sets can
>> confirm later.
>>
>> -- Michelle
> Thanks, in my file what I am seeing is an 'o' with a tail on top-right
> Is that a valid norwegian character, and if so what is the
> proper unicode escape for it and where is it expected to sort with
> regard to aa.
I guess the problem is that CollationTest2.java contains "å" unescaped
in some of the string literals. Therefore, which character the compiler
chooses to put into the string depends on the compile-time locale. Fixed
in revision 543077.
--
Knut Anders
Re: norway specific failure in CollationTest2
Posted by Michelle Caisse <Mi...@Sun.COM>.
I'm not familiar with that character. It's definitely not a "standard"
Norwegian character. Whether or not it ever occurs in Norwegian as a
borrowing from another language, I'm not sure. The Norwegian alphabet
includes ae digraph, o-with-a-diagonal-slash, and a-with-a-circle-on-top
which sort after a-z in the order given.
I don't know how these characters are represented in unicode. I also
don't know if the multi-character equivalents of these characters should
sort with them, because of the quirks Bernt mentioned in the spelling
equivalencies.
-- Michelle
Mike Matrigali wrote:
>
>
> Michelle Caisse wrote:
>> For a quick response, in Norwegian, a sequence of two 'a' characters
>> is equivalent to a single 'a'-with-a-circle-on-top character. Maybe
>> someone who both knows more about Norwegian and character sets can
>> confirm later.
>>
>> -- Michelle
> Thanks, in my file what I am seeing is an 'o' with a tail on top-right
> Is that a valid norwegian character, and if so what is the
> proper unicode escape for it and where is it expected to sort with
> regard to aa.
>
>>
>> Mike Matrigali wrote:
>>
>>> I'll try to figure out what is going on with the CollationTest2 test.
>>> I am not seeing it in my environment, and it does not seem to have
>>> shown up in any of the publically posted nightly regression tests.
>>>
>>> Can someone from Norway help, is it possible that 2 strings used in
>>> the test, are to be treated as equal in Norway?:
>>> "Wσgan" "Waagan"
>>>
>>> The first string contains an unfamilar character (to my english
>>> environment - I am not sure my mailer will even handle it correctly)
>>> as it's second letter, and I am wondering if I should have used a
>>> unicode escape for the data
>>> instead. See the data string in the runDerby2670()routine of the
>>> CollationTest2 test.
>>>
>>> Is there a chance that in norway aa is supposed to be treated
>>> as equal to the second letter in the first string listed? If so then
>>> either order may be right, and the test is incorrectly testing for
>>> one order or the other.
>>>
>>> I tried to take the data from a donated test case and put it into
>>> the junit test. It could be I messed something up on my side dealing
>>> with a non english character in some step of the way (file handling,
>>> code sets, editor, ...).
>>>
>>> Rick Hillegas wrote:
>>>
>>>> Julius Stroffek wrote:
>>>>
>>>>> Hi All,
>>>>>
>>>>> When I ran junit tests now for trunk I get 17 errors and 70
>>>>> failures of
>>>>> many different kinds.
>>>>>
>>>>> My junit console output with exceptions thrown is attached.
>>>>> Probably it
>>>>> is something wrong with my test environment or so, but I have no idea
>>>>> what happened. Are there any issues with upgrade tests due to a new
>>>>> release? Probably this is not the case since these tests does not
>>>>> fail
>>>>> in daily and tinderbox test runs.
>>>>>
>>>>> I always delete the contents of the current directory where I run
>>>>> tests
>>>>> (which ends using as derby.system.home). I tried to checkout the
>>>>> source
>>>>> again and start from a scratch but it does not helped.
>>>>>
>>>>> I run junit throught suites.All.
>>>>>
>>>>
>>>>
>>>> Hi Julo,
>>>>
>>>> I saw tons of upgrade errors in tests which I ran yesterday. Those
>>>> errors disappeared after I synced my client to the mainline this
>>>> morning. Now I am seeing the error which you are seeing in
>>>> CollationTest2. Otherwise, the tests run cleanly for me.
>>>>
>>>> Regards,
>>>> -Rick
>>>>
>>
>>
Re: norway specific failure in CollationTest2
Posted by "Bernt M. Johnsen" <Be...@Sun.COM>.
>>>>>>>>>>>> Mike Matrigali wrote (2007-05-30 15:47:48):
>
>
> Michelle Caisse wrote:
> >For a quick response, in Norwegian, a sequence of two 'a' characters is
> >equivalent to a single 'a'-with-a-circle-on-top character. Maybe someone
> >who both knows more about Norwegian and character sets can confirm later.
> >
> >-- Michelle
> Thanks, in my file what I am seeing is an 'o' with a tail on top-right
> Is that a valid norwegian character, and if so what is the
> proper unicode escape for it and where is it expected to sort with
> regard to aa.
"å" (sometimes aka "aa") is \u00e5 in lowercase and \u00c5 in
uppercase.
See e.g. http://unicode.org/charts/PDF/U0080.pdf
>
> >
> >Mike Matrigali wrote:
> >
> >>I'll try to figure out what is going on with the CollationTest2 test.
> >>I am not seeing it in my environment, and it does not seem to have
> >>shown up in any of the publically posted nightly regression tests.
> >>
> >>Can someone from Norway help, is it possible that 2 strings used in
> >>the test, are to be treated as equal in Norway?:
> >>"Wσgan" "Waagan"
> >>
> >>The first string contains an unfamilar character (to my english
> >>environment - I am not sure my mailer will even handle it correctly)
> >>as it's second letter, and I am wondering if I should have used a
> >>unicode escape for the data
> >>instead. See the data string in the runDerby2670()routine of the
> >>CollationTest2 test.
> >>
> >>Is there a chance that in norway aa is supposed to be treated
> >>as equal to the second letter in the first string listed? If so then
> >>either order may be right, and the test is incorrectly testing for
> >>one order or the other.
> >>
> >>I tried to take the data from a donated test case and put it into
> >>the junit test. It could be I messed something up on my side dealing
> >>with a non english character in some step of the way (file handling,
> >>code sets, editor, ...).
> >>
> >>Rick Hillegas wrote:
> >>
> >>>Julius Stroffek wrote:
> >>>
> >>>>Hi All,
> >>>>
> >>>>When I ran junit tests now for trunk I get 17 errors and 70 failures of
> >>>>many different kinds.
> >>>>
> >>>>My junit console output with exceptions thrown is attached. Probably it
> >>>>is something wrong with my test environment or so, but I have no idea
> >>>>what happened. Are there any issues with upgrade tests due to a new
> >>>>release? Probably this is not the case since these tests does not fail
> >>>>in daily and tinderbox test runs.
> >>>>
> >>>>I always delete the contents of the current directory where I run tests
> >>>>(which ends using as derby.system.home). I tried to checkout the source
> >>>>again and start from a scratch but it does not helped.
> >>>>
> >>>>I run junit throught suites.All.
> >>>>
> >>>
> >>>
> >>>Hi Julo,
> >>>
> >>>I saw tons of upgrade errors in tests which I ran yesterday. Those
> >>>errors disappeared after I synced my client to the mainline this
> >>>morning. Now I am seeing the error which you are seeing in
> >>>CollationTest2. Otherwise, the tests run cleanly for me.
> >>>
> >>>Regards,
> >>>-Rick
> >>>
> >
> >
--
Bernt Marius Johnsen, Database Technology Group,
Staff Engineer, Technical Lead Derby/Java DB
Sun Microsystems, Trondheim, Norway
Re: norway specific failure in CollationTest2
Posted by Mike Matrigali <mi...@sbcglobal.net>.
Michelle Caisse wrote:
> For a quick response, in Norwegian, a sequence of two 'a' characters is
> equivalent to a single 'a'-with-a-circle-on-top character. Maybe someone
> who both knows more about Norwegian and character sets can confirm later.
>
> -- Michelle
Thanks, in my file what I am seeing is an 'o' with a tail on top-right
Is that a valid norwegian character, and if so what is the
proper unicode escape for it and where is it expected to sort with
regard to aa.
>
> Mike Matrigali wrote:
>
>> I'll try to figure out what is going on with the CollationTest2 test.
>> I am not seeing it in my environment, and it does not seem to have
>> shown up in any of the publically posted nightly regression tests.
>>
>> Can someone from Norway help, is it possible that 2 strings used in
>> the test, are to be treated as equal in Norway?:
>> "Wσgan" "Waagan"
>>
>> The first string contains an unfamilar character (to my english
>> environment - I am not sure my mailer will even handle it correctly)
>> as it's second letter, and I am wondering if I should have used a
>> unicode escape for the data
>> instead. See the data string in the runDerby2670()routine of the
>> CollationTest2 test.
>>
>> Is there a chance that in norway aa is supposed to be treated
>> as equal to the second letter in the first string listed? If so then
>> either order may be right, and the test is incorrectly testing for
>> one order or the other.
>>
>> I tried to take the data from a donated test case and put it into
>> the junit test. It could be I messed something up on my side dealing
>> with a non english character in some step of the way (file handling,
>> code sets, editor, ...).
>>
>> Rick Hillegas wrote:
>>
>>> Julius Stroffek wrote:
>>>
>>>> Hi All,
>>>>
>>>> When I ran junit tests now for trunk I get 17 errors and 70 failures of
>>>> many different kinds.
>>>>
>>>> My junit console output with exceptions thrown is attached. Probably it
>>>> is something wrong with my test environment or so, but I have no idea
>>>> what happened. Are there any issues with upgrade tests due to a new
>>>> release? Probably this is not the case since these tests does not fail
>>>> in daily and tinderbox test runs.
>>>>
>>>> I always delete the contents of the current directory where I run tests
>>>> (which ends using as derby.system.home). I tried to checkout the source
>>>> again and start from a scratch but it does not helped.
>>>>
>>>> I run junit throught suites.All.
>>>>
>>>
>>>
>>> Hi Julo,
>>>
>>> I saw tons of upgrade errors in tests which I ran yesterday. Those
>>> errors disappeared after I synced my client to the mainline this
>>> morning. Now I am seeing the error which you are seeing in
>>> CollationTest2. Otherwise, the tests run cleanly for me.
>>>
>>> Regards,
>>> -Rick
>>>
>
>
Re: norway specific failure in CollationTest2
Posted by Michelle Caisse <Mi...@Sun.COM>.
For a quick response, in Norwegian, a sequence of two 'a' characters is
equivalent to a single 'a'-with-a-circle-on-top character. Maybe someone
who both knows more about Norwegian and character sets can confirm later.
-- Michelle
Mike Matrigali wrote:
> I'll try to figure out what is going on with the CollationTest2 test.
> I am not seeing it in my environment, and it does not seem to have
> shown up in any of the publically posted nightly regression tests.
>
> Can someone from Norway help, is it possible that 2 strings used in
> the test, are to be treated as equal in Norway?:
> "Wσgan" "Waagan"
>
> The first string contains an unfamilar character (to my english
> environment - I am not sure my mailer will even handle it correctly)
> as it's second letter, and I am wondering if I should have used a
> unicode escape for the data
> instead. See the data string in the runDerby2670()routine of the
> CollationTest2 test.
>
> Is there a chance that in norway aa is supposed to be treated
> as equal to the second letter in the first string listed? If so then
> either order may be right, and the test is incorrectly testing for
> one order or the other.
>
> I tried to take the data from a donated test case and put it into
> the junit test. It could be I messed something up on my side dealing
> with a non english character in some step of the way (file handling,
> code sets, editor, ...).
>
> Rick Hillegas wrote:
>
>> Julius Stroffek wrote:
>>
>>> Hi All,
>>>
>>> When I ran junit tests now for trunk I get 17 errors and 70 failures of
>>> many different kinds.
>>>
>>> My junit console output with exceptions thrown is attached. Probably it
>>> is something wrong with my test environment or so, but I have no idea
>>> what happened. Are there any issues with upgrade tests due to a new
>>> release? Probably this is not the case since these tests does not fail
>>> in daily and tinderbox test runs.
>>>
>>> I always delete the contents of the current directory where I run tests
>>> (which ends using as derby.system.home). I tried to checkout the source
>>> again and start from a scratch but it does not helped.
>>>
>>> I run junit throught suites.All.
>>>
>>
>> Hi Julo,
>>
>> I saw tons of upgrade errors in tests which I ran yesterday. Those
>> errors disappeared after I synced my client to the mainline this
>> morning. Now I am seeing the error which you are seeing in
>> CollationTest2. Otherwise, the tests run cleanly for me.
>>
>> Regards,
>> -Rick
>>
Re: norway specific failure in CollationTest2
Posted by "Bernt M. Johnsen" <Be...@Sun.COM>.
>>>>>>>>>>>> Mike Matrigali wrote (2007-05-30 15:03:04):
> Is there a chance that in norway aa is supposed to be treated
> as equal to the second letter in the first string listed? If so then
> either order may be right, and the test is incorrectly testing for
> one order or the other.
Well..... the character "å" replaced "aa" where "aa" was pronounced
"å" around 1925. The only place you will find "aa" today is in names
of persons and companies originating before that time. E.g. the
brewery "Aass" and Kristian's last name "Waagan". In those cases "å"
and "aa" should be treated as equivalent. "aa" is not used in any
other Noregian words (and not in names of places either). All foreign
uses of "aa" is not to be treated as "å" except for some Danish ones
(sigh....).
So in my little encyclopedia, The Swizz canton "Aargau" is found under
"A" just before "ab", while the Norwegian skier "Aamodt" is placed
under "Å" (which is our 29th letter) just before the Norwegian
municipality "Åmot". get it?
And how this tranlates into Derby's collation implementation? I have
no idea ;-)
("å" is in this mail is the same as HTML's å)
--
Bernt Marius Johnsen, Database Technology Group,
Staff Engineer, Technical Lead Derby/Java DB
Sun Microsystems, Trondheim, Norway