Re: env DBI_SQL_NANO=1 make test 2>&1 | tee make-test-sql-nano.log

On 23.05.13 13:07, H.Merijn Brand wrote:
> On Wed, 22 May 2013 19:47:13 +0200, Jens Rehsack <[email protected]>
> wrote:
>
>> If you want to redo - fix
>>
>> #   Failed test 'Record 3'
>> #   at t/80_rt.t line 205.
>>
>> before (prints out a lot - very lot - of ",")
>>
>> Cheers
>
> Not as bad as you make me believe :)

What? I always said, that most of DBD::CSV could run with Nano,
probably at higher speed.

> But what do we want here?

World domination?

What I want should be clear - wake you up for moving to DBI::Test
in near future.

The tests in DBD::CSV cover 4 areas:
1) DBD::CSV
2) SQL::Statement
3) DBI::DBD::SqlEngine / DBD::File
4) DBI / DBD API itself

So a reasonable goal should be, to know what is covered where and why,
can it be moved, etc.

> Skip those tests is NANO is used?
> SQL::Statement *is* a prereq, so basically, I don't care if
> someone wants to limit the usage by restricting to that.

Maybe it doesn't need to be. Finally it's a much better (even not
impressive) SQL engine - but for simple table scans for importing
results into Oracle or DB2, it's probably more than enough.

> IMHO the fix is easy as in "Do not test 80_rt at all when under NANO
> most of rt is edge cases (and some tests are dups from SQL::Statement
> to verify fixes on that side)

Yes, reasonable (I always please you to do that, because of circular
dependencies ...).

Cheers
-- 
Jens Rehsack