I have took in consideration all of your points, is it better now?
The current patch doesn't overwrite the present behavior of
org-set-timer it only add the possibility to use hh:mm:ss format.


2015-04-24 18:49 GMT+02:00 Kyle Meyer <kyle@kyleam.com>:
> Brice Waegenire <brice.wge@gmail.com> wrote:
>> Hello,
>>
>> I've hacked a patch that use hh:mm:ss format instead of minutes for
>> org-timer-set-timer. I'm really not sure I did it in the right way,
>> any sugestions are welcome.
> [...]
>
> Thanks.
>
> I think it's nice to be able to specify seconds, but now you have to
> type 'N:00' (or at least 'N:0') instead of 'N' to get N minutes.  Should
> a plain number default to minutes?  I don't use org-timer very much, so
> I don't have a strong preference.
This seems like a better behavior
>
>
>> --- a/lisp/org-timer.el
>> +++ b/lisp/org-timer.el
>> @@ -429,17 +429,14 @@ using three `C-u' prefix arguments."
>>        (minutes (or (and (not (equal opt '(64)))
>>                          effort-minutes
>>                          (number-to-string effort-minutes))
>> -                  (and (numberp opt) (number-to-string opt))
>> -                  (and (listp opt) (not (null opt))
>> -                       (number-to-string org-timer-default-timer))
>
> By removing the listp check, you no longer get the C-u behavior
> described in the docstring.
I've re-added the C-u functionality, I didn't understood the whole
meaning of those lines.

>
>> +                  (and (stringp opt) (prin1 opt))
>
> Why not `(and (stringp opt) opt)'?
Because I don't really know how to program, but I was already thinking
that this prin1 function wasn't the right way do to this.

>
>>                    (read-from-minibuffer
>> -                   "How many minutes left? "
>> +                   "How many time left? "
>
> s/many/much/.  Also, it'd be nice to specify the format in the prompt.
>
>>                     (if (not (eq org-timer-default-timer 0))
>> -                       (number-to-string org-timer-default-timer))))))
>> +                       (prin1 org-timer-default-timer))))))
>
> The defcustom for org-timer-default-timer still says it should be a
> number.  If set to a number other than 0, this will fail.  Perhaps
> org-timer-default-timer should be updated to be a string in the hh:mm:ss
> format.
>
> --
> Kyle

2015-04-24 18:49 GMT+02:00 Kyle Meyer <kyle@kyleam.com>:
> Brice Waegenire <brice.wge@gmail.com> wrote:
>> Hello,
>>
>> I've hacked a patch that use hh:mm:ss format instead of minutes for
>> org-timer-set-timer. I'm really not sure I did it in the right way,
>> any sugestions are welcome.
> [...]
>
> Thanks.
>
> I think it's nice to be able to specify seconds, but now you have to
> type 'N:00' (or at least 'N:0') instead of 'N' to get N minutes.  Should
> a plain number default to minutes?  I don't use org-timer very much, so
> I don't have a strong preference.
>
>
>> --- a/lisp/org-timer.el
>> +++ b/lisp/org-timer.el
>> @@ -429,17 +429,14 @@ using three `C-u' prefix arguments."
>>        (minutes (or (and (not (equal opt '(64)))
>>                          effort-minutes
>>                          (number-to-string effort-minutes))
>> -                  (and (numberp opt) (number-to-string opt))
>> -                  (and (listp opt) (not (null opt))
>> -                       (number-to-string org-timer-default-timer))
>
> By removing the listp check, you no longer get the C-u behavior
> described in the docstring.
>
>> +                  (and (stringp opt) (prin1 opt))
>
> Why not `(and (stringp opt) opt)'?
>
>>                    (read-from-minibuffer
>> -                   "How many minutes left? "
>> +                   "How many time left? "
>
> s/many/much/.  Also, it'd be nice to specify the format in the prompt.
>
>>                     (if (not (eq org-timer-default-timer 0))
>> -                       (number-to-string org-timer-default-timer))))))
>> +                       (prin1 org-timer-default-timer))))))
>
> The defcustom for org-timer-default-timer still says it should be a
> number.  If set to a number other than 0, this will fail.  Perhaps
> org-timer-default-timer should be updated to be a string in the hh:mm:ss
> format.
>
> --
> Kyle