Hi João,

João Henrique Freitas writes:
 > > (3) Failure!: LIBDBI fixture -> Test select cases -> 
 > > test_retrieve_zero_rows -> Value: '0x0' Error 1146: 'root' at 
 > > [test_dbi_misc.c] line [8]
 > > Failure!: LIBDBI fixture -> Test select cases -> test_retrieve_zero_rows 
 > > -> [-1] should not match [-1] at [test_dbi_misc.c] line [12]
 > > Failure!: LIBDBI fixture -> Test select cases -> test_retrieve_zero_rows 
 > > -> [-1] should match [0] at [test_dbi_misc.c] line [13]
 > > Running "libdbi tests"...
 > >
 > > I tend to believe that this is a libdbi bug. I've tried test_dbi.c
 > > which returns 26 rows instead of zero with at least the mysql and
 > > sqlite3 drivers. I've never noticed this problem before. This actually
 > > shows how valuable the cgreen framework is going to be for us! In any
 > > case, I'll investigate if the library indeed has a problem here.
 > >
 > > I'll check the other drivers as soon as time permits.
 > 

I've revisited this problem yesterday, and I concluded that your test
is too strict at this point, in contrast to my initial suspicion that
this is a libdbi bug.

I looked at the output of the old test program and got the following
results for numfields:

mysql: 26
pgsql: 26
sqlite: 0
sqlite3: 26
firebird: 22

26 and 22 are the proper number of columns if you look at the CREATE
TABLE statements. Firebird skips a few tests and hence requires less
columns. The only wrong number is returned by the sqlite driver. This
appears to be by design, and was obviously changed in the 3.x series.

Your current test in test_retrieve_zero_rows() asserts that numfields
equals 0 which is not going to work. There is no unique return value
to test against. If at all, we'd have to test against the actually
created number of columns or against 0 in case we're looking at
sqlite. Is there a cheap way to keep the number of columns tested
against in sync with the CREATE TABLE statements?

The only remaining issue that I remember is creating the database for
a couple of engines. Did you find some time to address this issue? I
think once these two issues are fixed, you should go ahead and merge
the new test code into HEAD.

regards,
Markus

-- 
Markus Hoenicka
markus.hoeni...@cats.de
(Spam-protected email: replace the quadrupeds with "mhoenicka")
http://www.mhoenicka.de

------------------------------------------------------------------------------
Crystal Reports - New Free Runtime and 30 Day Trial
Check out the new simplified licensign option that enables unlimited
royalty-free distribution of the report engine for externally facing 
server and web deployment.
http://p.sf.net/sfu/businessobjects
_______________________________________________
Libdbi-drivers-devel mailing list
Libdbi-drivers-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/libdbi-drivers-devel

Reply via email to