Sorry Ole - I simplified my use case too much. On running it with the full set of Parallel operands I intend to use, I ran into a combination that doesn't work, or at least gives a misleading error message. Adding ‘--pipe’ to the simplified example causes this:
larryp-MBP:~ larry$ printf '%s %s\n' {a..d} | parallel -C ' '
--sqlandworker csv:///%2ftmp%2ffoo.csv/bar --pipe echo /{1}/{2}/
DBI connect('','',...) failed: No such directory '/tmp/foo.csv at
/Users/larry/bin/parallel line 12252.
No such directory '/tmp/foo.csv at /Users/larry/bin/parallel line 12252.
larryp-MBP:~ larry$ printf '%s %s\n' {a..d} | parallel -C ' '
--sqlandworker sqlite:///%2ftmp%2ffoo.sqlite/bar --pipe echo /{1}/{2}/
DBD::SQLite::db prepare failed: no such table: bar [for Statement "SELECT *
FROM bar LIMIT 1;"] at /Users/larry/bin/parallel line 12451.
Thanks!
— Larry
On 3/21/20 12:05 AM, Larry Ploetz - lplo...@gmail.com wrote:
> Thanks! Everything seems to be working as (currently) expected.
>
> - Larry
>
> On 3/20/20 7:20 PM, Ole Tange - o...@tange.dk wrote:
>> On Tue, Mar 17, 2020 at 8:22 PM Larry Ploetz <lplo...@gmail.com> wrote:
>>
>>> Well, I tried some things with CSV and got different errors:
>>>
>>> larryp-MBP:parallel larry$ printf '%s %s\n' {a..d} | gtee >(cat >&2) |
>>> src/parallel -C' ' --sqla csv:///%2ftmp%2ffoo.csv/bar echo /{1}/{2}/
>>> a b
>>> c d
>>> parallel: Error:
>>> Execution ERROR: Cannot open /Users/larry/git/parallel/bar: No such file or
>>> directory (2) at
>>> /opt/local/lib/perl5/site_perl/5.30/darwin-thread-multi-2level/DBI/DBD/SqlEngine.pm
>>> line 1624.
>>> called from src/parallel at 12446.
>> I can reproduce your error when using the unsupported `--sqla`. But it
>> is not a supported option. This works for me:
>>
>> printf '%s %s\n' {a..d} | tee >(cat >&2) | parallel -C' '
>> --sqlandworker csv:///%2ftmp%2ffoo.csv/bar echo /{1}/{2}/
>>
>> I am not going to spend time on figuring out why --sqla does not work.
>> If others find the reasons, I am cool with that.
>>
>>> larryp-MBP:parallel larry$ touch /tmp/foo.csv
>>> larryp-MBP:parallel larry$ printf '%s %s\n' {a..d} | gtee >(cat >&2) |
>>> src/parallel -C' ' --sqla csv:///%2ftmp%2ffoo.csv/bar echo /{1}/{2}/
>>> a b
>>> c d
>>> parallel: Error:
>>> Execution ERROR: Cannot open /Users/larry/git/parallel/bar: No such file or
>>> directory (2) at
>>> /opt/local/lib/perl5/site_perl/5.30/darwin-thread-multi-2level/DBI/DBD/SqlEngine.pm
>>> line 1624.
>>> called from src/parallel at 12446.
>> The same here: --sqla does not work, but this works for me:
>>
>> printf '%s %s\n' {a..d} | gtee >(cat >&2) | parallel -C' '
>> --sqlandworker csv:///%2ftmp%2ffoo.csv/bar echo /{1}/{2}/
>>
>>> I'm assuming it's OK to use the full DBTABLEURL (variable from the example)
>>> value on the command line.
>>>
>>> However, with 1 variable this works,
>> No. It works because you are writing --sqlandworker instead of --sqla.
>>
>>> but completely disregards the directory:
>>> larryp-MBP:parallel larry$ parallel --sqlandworker
>>> csv:///%2FBOGUS%2FBOGUS2/mytable3 seq ::: {1..10}
>> That looks like a bug in DBI. Thanks.
>>
>> /Ole
>>
>>
pEpkey.asc
Description: application/pgp-keys
