Re: [O] limitations of hard-coded field separator removed

[Bastien]

Hi Andreas,

Andreas Röhler <address@hidden> writes:

> attached a org-table-import.patch

Thanks for this patch.

I test it against latest master and got this error:

,----
| Debugger entered--Lisp error: (wrong-type-argument stringp nil)
|   re-search-forward(nil #<marker at 57 in tst.org> t)
|   (while (re-search-forward separator end t) (replace-match " | " t t))
|   (let* ((beg (min beg0 end0)) (end (copy-marker (max beg0 end0))) (separator 
(cond ((and separator (stringp separator)) separator) ((eq 4 
(prefix-numeric-value separator)) (read-from-minibuffer "Separator char: "))))) 
(goto-char beg) (unless separator (setq separator (org-guess-separator))) 
(goto-char beg) (while (re-search-forward separator end t) (replace-match " | " 
t t)) (goto-char beg) (org-table-align))
|   org-table-convert-region(14 57 nil)
|   call-interactively(org-table-convert-region t nil)
|   execute-extended-command(nil)
|   call-interactively(execute-extended-command nil nil)
`----

Beside this problem...

> removes limitations of hard-coded separator char(s).
> `org-guess-separator' accepts and detects all chars as field separators.
> It works based on the assumption, that char looked for appears in equal
> number at each row.

This looks clever, but is it safe?  E.g. if cells of the table are
surrounded by double-quotes, you'll have the same number of quotes in
each row, but the double-quote char by itself is _not_ the separator.

So I'm not sure about this heuristics -- want to ear/test more.

> Also a default value
> `org-table-import-default-separator' should make
> guessing faster in some cases.

That an easier way and I would favor it: keep the current behavior, 
but add an option for a list of separators that the user can set.  
Org would then test this list sequentially.  

What do you think?

-- 
 Bastien