Hi. Thanks for replying.


Nicolas Goaziou <mail@nicolasgoaziou.fr> writes:

> Thanks for your patch. Some comments follow.
>
>> From 3b6581c647cb87f0d3e8cee94ce2fb1fb122d3fd Mon Sep 17 00:00:00 2001
>> From: Dima Kogan <dima@secretsauce.net>
>> Date: Tue, 30 Sep 2014 22:36:21 -0700
>> Subject: [PATCH] Field formulas can now add columns as needed
>>
>> The org-table-formula-make-new-cols customization controls whether and how this
>> is done
>
> Your commit message is missing information and "TINYCHANGE" at its end.
> See http://orgmode.org/worg/org-contribute.html#sec-5

OK. Added. By the way, I signed the copyright-assignment paperwork for
Emacs itself. Would I need to do this again for org-mode?



>> +(defcustom org-table-formula-make-new-cols nil
>
> What about `org-table-formula-create-columns'?

Sure, why not.



>> +  "Non-nil means that evaluation of a field formula can add new
>> +columns if an out-of-bounds field is being set."
>
> First line needs to be complete. E.g.,
>
>   "Non-nil means a field formula can create a new column."
>
>> +  :group 'org-table-calculation
>> +  :type '(choice
>> +	  (const :tag "Setting an out-of-bounds field generates an error (default)" nil)
>> +	  (const :tag "Setting an out-of-bounds field silently adds columns as needed" t)
>> +	  (const :tag "Setting an out-of-bounds field adds columns as needed, but issues a warning message" warn)
>> +	  (const :tag "When setting an out-of-bounds field, the user is
>> prompted" prompt)))
>
> You need to add :version and :package-version keywords.

OK. I'm not sure what those values should be, so please double-check.


>>  (defgroup org-table-import-export nil
>>    "Options concerning table import and export in Org-mode."
>>    :tag "Org Table Import Export"
>> @@ -3125,7 +3135,22 @@ known that the table will be realigned a little later anyway."
>>        (while (setq eq (pop eqlname1))
>>  	(message "Re-applying formula to field: %s" (car eq))
>>  	(org-goto-line (nth 1 eq))
>> -	(org-table-goto-column (nth 2 eq))
>> +	(let* ((column-target (nth 2 eq))
>> +	       (column-count (progn (end-of-line)
>> +				    (1- (org-table-current-column))))
>> +	       (create-new-column
>> +		(and (> column-target column-count)
>
> Is this check really necessary? Doesn't `org-table-goto-column' already
> figures it out before creating a new column?

This is necessary if we want to be able to warn the user or to prompt
them ONLY if a new column has to be made. I.e. if we're not looking past
the bounds of the table the user should never be pestered. If the
customization variable is only t/nil then this can be dramatically
simplified, as you have observed.

By the way, is (org-display-warning) the preferred way to produce a
warning? It does into a new buffer, not into *Messages*.

Tweaked patch attached.