Re: [Orgmode] Re: [Babel] How to pass variables to SQL blocks?

[Eric Schulte]

Sébastien Vauban <address@hidden> writes:

> Hi Eric,
>
> "Eric Schulte" wrote:
>> Sébastien Vauban <address@hidden> writes:
>>> (I thought this feature would be implemented using "real" SQL vars, hence 
>>> my @
>>> prefix. But viewing the code shows the replacement is made outside of the 
>>> SQL
>>> snippet's knowledge.)
>>
>> I agree that using real SQL variables would be preferable.  If you want
>> to submit a patch to this effect I would be happy to review it and apply.
>
> I'll give a shot -- not in the really short term, though. Which example should
> I look at?  shell?
>

great, and please no rush.

shell actually has rather sophisticated variable passing, maybe sqlite
and gnuplot would provide examples more similar to sql.

>
>
>>> Maybe the following should be put on some TODO list, even if not urgent?
>>>
>>>>> * No warning if block does not exist
>>>>>
>>>>> Please note that inexistent references are ignored, without further
>>>>> notice. That's the case for such a block (where I forgot the number
>>>>> suffix):
>>>>>
>>>>> #+srcname: contains-inexistent-ref
>>>>> #+begin_src sql :noweb yes
>>>>> <<add-column-in-table>>
>>>>> #+end_src
>>
>> The lack of notice (or error) was originally intentional.  The logic
>> being that there may be languages in which <<something>> constructs are
>> part of the language syntax, and in those languages Org-mode should
>> ignore such constructs unless the "something" contained in the <<>>'s is
>> actually the name of a code block.
>>
>> It may make sense to revisit this design decision.
>
> IMHO, it should, yes. Because a typo is so easy, and the results can be very
> different then, from what's expected... OK, there is always a preview mode,
> but an alert is better (for me).
>
> Maybe adding an option for those languages, that would suppress such warnings?
>

That may be the best idea, once again a header argument may be the most
flexible solution.

>
> Re helper functions for such named code blocks, do you have an idea what I
> should do to have the IMenu index of code blocks working? See previous thread
> [[http://mid.gmane.org/808w09foxd.fsf%40missioncriticalit.com][Re: imenu
> support for babel blocks]].
>

I don't have any experience with IMenu support, however to map over all
code blocks in a file extracting information I would suggest using the
org-babel-map-src-blocks macro.

Cheers -- Eric

>
> Thanks for any hint!
>
> Best regards,
>   Seb