[Ludiafuncs-hackers] Add integration test(ludia_funcs+pg_bigm )

Back to archive index

Fujii Masao masao****@gmail*****
2015年 9月 4日 (金) 18:43:50 JST


On Fri, Sep 4, 2015 at 2:56 PM,  <ishii****@nttda*****> wrote:
> Thank you for review.
> I fixed problem, and attached new patch.
> Could you review the patch?
>
>> On Fri, Sep 4, 2015 at 10:28 AM,  <ishii****@nttda*****> wrote:
>> > Hi,
>> >
>> > I attached regression test patch.
>> >
>> > This patch includes integration test(pg_bigm + ludia_funcs).
>> > This test succeed with both PG9.1 and PG9.4.
>> >
>> > The user can test by using following command.
>> >     make installcheck TEXTPORTER=** PGS2_DEBUG=1 PG_BIGM=1
>> >
>> > Could you review patch?
>> > If the patch has unnecessary tests, please comment.
>>
>> +ifdef PG_BIGM
>> +  ifdef TEXTPORTER
>> +    ifdef PGS2_DEBUG
>> +    REGRESS += pg_bigm
>> +    endif
>> +  endif
>> +endif
>>
>> Please imagine the case where we compiled ludia_funcs without PG_BIGM flag.
>> In this case, if we want to run the regression test with pg_bigm, we need
>> to recompile the module with PG_BIGM flag. OTOH, please imagine the case
>> where the module was compiled with PG_BIGM flag. In this case, if we decided
>> not to install pg_bigm extension, the regression test for ludia_funcs always
>> fails.
>> So we need to recompile the module without PG_BIGM flag. These seem messy.
>> So I'd like to add new regression test command like "make installcheck-bigm"
>> rather than adding new PG_BIGM flag.
>>
>> If the regression test with pg_bigm contains the tests which we can run
>> only when the module was compiled with PGS2_DEBUG or TEXTPORTER, I think
>> that those tests should be separated into different ones like
>> ludia_funcs.sql, pgs2-debug.sql and textporter.sql. That is, only when
>> ludia_funcs was compiled with PGS2_DEBUG flag, "make installcheck-bigm"
>> runs the tests using pgs2_debug option in addtion to the normal tests.
>> Thought?
>
> I modified Makefile.
> If PGS2_DEBUG is defined, the user can run pg_bigm test with "make installcheck-bigm".
> (I removed textporter test, so the user does not have to do "make TEXTPORTER=**")
>
>> +-- Test the case where textporter get text from variaus file type.
>> +CREATE TABLE textporter_sample (filehonbun text); INSERT INTO
>> +textporter_sample(filehonbun) VALUES
>> (pgs2textporter1('@abs_srcdir@/data/word2007_test.docx'));
>> +INSERT INTO textporter_sample(filehonbun) VALUES
>> (pgs2textporter1('@abs_srcdir@/data/excel2007_test.xlsx'));
>> +INSERT INTO textporter_sample(filehonbun) VALUES
>> (pgs2textporter1('@abs_srcdir@/data/powerpoint2007_test.pptx'));
>> +INSERT INTO textporter_sample(filehonbun) VALUES
>> (pgs2textporter1('@abs_srcdir@/data/pdf1.5_test.pdf'));
>> +INSERT INTO textporter_sample(filehonbun) VALUES
>> (pgs2textporter1('@abs_srcdir@/data/text_test.txt'));
>> +INSERT INTO textporter_sample(filehonbun) VALUES
>> (pgs2textporter1('@abs_srcdir@/data/csv_test.csv'));
>> +
>> +SELECT * FROM textporter_sample;
>>
>> Since these tests are not directly related to pg_bigm, I think that they
>> should be included in the regression tests for textporter.
>> Thought?
>
> I agree with you.
> I moved these test to the regression tests for textporter.

Thanks for the patches. Firstly I committed these patches.

Regards,

-- 
Fujii Masao




Ludiafuncs-hackers メーリングリストの案内
Back to archive index