Thread: timestamptz test failed by postgresql 6.5.0

timestamptz test failed by postgresql 6.5.0

From
yangtao
Date:
Recently, I compile the Postgresql 9.6.0 on Linux by myself. However, I meet one problem about the timestamptz.sql test case.
The expected timestamptz.out is different from the result of timestamptz.out for ime zone America/Caracas and
America/Santiago

Part of expected timestamptz.out :
DROP TABLE TIMESTAMPTZ_TST;
-- test timestamptz constructors
set TimeZone to 'America/Santiago';
-- numeric timezone
SELECT make_timestamptz(1973, 07, 15, 08, 15, 55.33);
        make_timestamptz         
---------------------------------
 Sun Jul 15 08:15:55.33 1973 CLT
(1 row)

SELECT make_timestamptz(1973, 07, 15, 08, 15, 55.33, '+2');
        make_timestamptz         
---------------------------------
 Sun Jul 15 02:15:55.33 1973 CLT

Part of the result  timestamptz.out :
DROP TABLE TIMESTAMPTZ_TST;
-- test timestamptz constructors
set TimeZone to 'America/Santiago';
-- numeric timezone
SELECT make_timestamptz(1973, 07, 15, 08, 15, 55.33);
        make_timestamptz         
---------------------------------
 Sun Jul 15 08:15:55.33 1973 -04
(1 row)

SELECT make_timestamptz(1973, 07, 15, 08, 15, 55.33, '+2');
        make_timestamptz         
---------------------------------
 Sun Jul 15 02:15:55.33 1973 -04

I don't why the timezone in the output "CLT" becomes "-04" in the output. Could anyone please give me some idea to find the trouble cause?
Thanks!


Best Regards!
Tao


 

Re: timestamptz test failed by postgresql 6.5.0

From
Tom Lane
Date:
yangtao <yangtaoyours@126.com> writes:
> Recently, I compile the Postgresql 9.6.0 on Linux by myself. However, I meet one problem about the timestamptz.sql
testcase. 
> The expected timestamptz.out is different from the result of timestamptz.out for ime zone America/Caracas and
> America/Santiago

Hm, did you use configure --with-system-tzdata?  If so, this is not
terribly surprising; the 9.6.0 test cases rely on the behavior of the
IANA timezone data as it stood at the time, and those zones changed
since then.

Is there a reason you're trying to build 9.6.0 rather than 9.6.latest,
which includes a fix for this [1] as well as hundreds of
far-more-significant bugs?

            regards, tom lane

[1] https://git.postgresql.org/gitweb/?p=postgresql.git&a=commitdiff&h=d8ec6b9c8c265c2f29b1c0e0e4205895baaa326d



Re:Re: timestamptz test failed by postgresql 6.5.0

From
yangtao
Date:
Tom, Thanks!
Yes, I compile it by configure --with-system-tzdata, should I remove it?
What's more, because some other components dependency, I must use version 9.6.0.
I also have tried the 9.6.5 version, find the timestamptz test case is updated.



Best Regards!
Tao




At 2021-01-04 14:07:39, "Tom Lane" <tgl@sss.pgh.pa.us> wrote: >yangtao <yangtaoyours@126.com> writes: >> Recently, I compile the Postgresql 9.6.0 on Linux by myself. However, I meet one problem about the timestamptz.sql test case. >> The expected timestamptz.out is different from the result of timestamptz.out for ime zone America/Caracas and >> America/Santiago > >Hm, did you use configure --with-system-tzdata? If so, this is not >terribly surprising; the 9.6.0 test cases rely on the behavior of the >IANA timezone data as it stood at the time, and those zones changed >since then. > >Is there a reason you're trying to build 9.6.0 rather than 9.6.latest, >which includes a fix for this [1] as well as hundreds of >far-more-significant bugs? > > regards, tom lane > >[1] https://git.postgresql.org/gitweb/?p=postgresql.git&a=commitdiff&h=d8ec6b9c8c265c2f29b1c0e0e4205895baaa326d >


 

Re: timestamptz test failed by postgresql 6.5.0

From
Tom Lane
Date:
yangtao  <yangtaoyours@126.com> writes:
> Yes, I compile it by configure --with-system-tzdata, should I remove it?
> What's more, because some other components dependency, I must use version 9.6.0.

No.  For god's sake, fix whatever components you are using that think
they must have 9.6.0 rather than 9.6.later.  Read the release notes [1].
Insisting on 9.6.0 four years later is professional malpractice.

(I'll not even get into the question of whether you should be installing
some later release branch than 9.6.x; but I can't help noting that 9.6's
EOL is 11 months away.)

            regards, tom lane

[1] https://www.postgresql.org/docs/release/



Re:Re: timestamptz test failed by postgresql 6.5.0

From
yangtao
Date:
em~~~,  I read the release notes [1] right now.
I try my best fix it as you suggested, and persuade my partners to not use version 9.6.0. 
Thanks for your help!

Best Regards!
Tao






At 2021-01-04 14:36:30, "Tom Lane" <tgl@sss.pgh.pa.us> wrote: >yangtao <yangtaoyours@126.com> writes: >> Yes, I compile it by configure --with-system-tzdata, should I remove it? >> What's more, because some other components dependency, I must use version 9.6.0. > >No. For god's sake, fix whatever components you are using that think >they must have 9.6.0 rather than 9.6.later. Read the release notes [1]. >Insisting on 9.6.0 four years later is professional malpractice. > >(I'll not even get into the question of whether you should be installing >some later release branch than 9.6.x; but I can't help noting that 9.6's >EOL is 11 months away.) > > regards, tom lane > >[1] https://www.postgresql.org/docs/release/