Re: [O] [babel] Some variables with no default value don't provoke an error

[Sebastien Vauban]

Hi Eric,

Eric Schulte wrote:
> "Sebastien Vauban" <address@hidden> writes:
>> Eric Schulte wrote:
>>> "Sebastien Vauban" <address@hidden> writes:
>>>> Weirdly enough, in the following code block, I must add a default value for
>>>> vars `table', `column' and `type' but not for the var `nullability'.
>>>>
>>>> I've even been able to add fake vars `something' and `else' with no error
>>>> being reported (at ingestion time):
>>>>
>>>> #+srcname: add-column-in-table(table="", column="", something, type="", 
>>>> else, nullability)
>>>> #+begin_src sql
>>>> -- add column `$column' (if column does not exist yet)
>>>> IF NOT EXISTS (SELECT *
>>>>                FROM INFORMATION_SCHEMA.COLUMNS
>>>>                WHERE TABLE_NAME = '$table'
>>>>                AND COLUMN_NAME = '$column')
>>>> BEGIN
>>>>     ALTER TABLE $table
>>>>     ADD $column $type $nullability
>>>> END
>>>> #+end_src
>>>>
>>>> Note that, in the above state, the code block is ingested with no error, 
>>>> but,
>>>> if I remove the default value of var `table', it then generates back an
>>>> error...
>>>
>>> I've just pushed up a check for these functional-syntax variables which
>>> will ensure that each is given a default value.  Since this check takes
>>> place at the location of the code block it /does/ include the name of
>>> the code block in the error message.
>>
>> If you have a couple of minutes, can you clarify some points to be sure I can
>> understand:
>>
>> - What do you call functional-syntax vars?  The ones in the #+srcname, next 
>> to
>>   the block name, as opposed to the ones declared as :var arguments?
>
> yes, that's exactly it, I don't know what "functional-syntax" is a good
> or descriptive term, but it is used in the source code so I'm now
> repeating it.

OK.

>>   The fact, then, that we can get a clearer message in case of error, seems 
>> to
>>   me an incentive to use that type of declaration...
>
> I personally prefer the traditional ":var" style, I'll have to add
> similar error checking there...

Good to know.

>> - Why was `nullability' not detected to have no default value?  Why were
>>   `table', `column' and `type' well correctly detected?
>
> Meaning after you assigned values to the first three no error was thrown
> when the fourth (nullability) wasn't assigned a value?  Could you
> provide a minimal example?

Yes, you summed up exactly what I (can assure you that) observed yesterday.
Though, now that the message includes the src-name, it is somehow fixed. I
can't reproduce it anymore... Thanks.

Best regards,
  Seb

-- 
Sebastien Vauban