[IPython-dev] IPEP 15: Notebook autosave

Dmitry Chichkov dchichkov@gmail....
Wed Apr 10 22:09:02 CDT 2013


I've tried integrating notebook with Git, but that proved troublesome
and unusable. And so far a ridiculous solution to add notebook folder
to dropbox and just let it sync automatically and keep the history
have been working the best for me.

Dmitry


On Wed, Apr 10, 2013 at 4:53 PM, MinRK <benjaminrk@gmail.com> wrote:
>
>
>
> On Sat, Apr 6, 2013 at 7:02 PM, Aaron Meurer <asmeurer@gmail.com> wrote:
>>
>> I think it's a great idea. The traditional save model really makes no
>> sense because the "default" behavior (what happens when you exit without
>> saving) is not what the user wants to happen 99% of the time. I have a
>> question and a suggestion.
>>
>> What happens when the browser window is closed? Does a save occur? A
>> "checkpoint"? Is it possible to reliably do any action?
>
>
> I will make it do a regular autosave (I'm not sure there is a guarantee of
> success for calls in beforeunload though).
>
>>
>>
>> My suggestion is to add undo support. This is generally how the "poor
>> man's VCS" that you describe is usually implemented. Traditionally undo
>> history is cleared when the program restarts (though this isn't a
>> requirement), but this is fine because it just serves the need of "oops I
>> didn't want to do that," of which I think the "I don't want to keep any of
>> the changes since I last opened the file" of reverting to the last
>> checkpoint is a special case of. "Reverting to checkpoint" would then just
>> mean performing undo until the checkpoint. Having further actual checkpoints
>> is unnecessary. The undo points would serve as them, and as long as
>> performing multiple undos/redos is easy, that should be enough.
>
>
> We have minimal undo - on a per-cell basis you can undo edits, and you can
> undo cell deletions.  We do not have a real, rich undoable history of all
> changes to the notebook. I'm not sure how feasible that would be. Maybe when
> we work out the transactional save machinery, we can store a history of
> transactions, which should make pretty complete undo possible.  But I don't
> think it's in the cards in the immediate future.
>
>>
>>
>> Also a comment. Regardless of whether you do undo or revert, you should
>> think about how to word it/present it to the user so that it is clear that
>> only the notebook document is being changed. The kernel state necessarily
>> stays the same.
>
>
> Thanks, this is indeed important.
>
> -MinRK
>
>>
>>
>> Aaron Meurer
>>
>> On Apr 6, 2013, at 7:42 PM, MinRK <benjaminrk@gmail.com> wrote:
>>
>>
>>
>>
>> On Sat, Apr 6, 2013 at 6:35 PM, Aaron Meurer <asmeurer@gmail.com> wrote:
>>>
>>> The second bullet point seems incomplete.
>>
>>
>> yup, fixed.
>>
>>>
>>>
>>> Aaron Meurer
>>>
>>> On Apr 6, 2013, at 7:13 PM, MinRK <benjaminrk@gmail.com> wrote:
>>>
>>> I've sketched a proposal for autosave in the Notebook.  Have a look,
>>> discuss, etc.  It's a much more webapp-centric approach than emacs / MS
>>> Word-style autosaved backups.
>>>
>>> -Min RK
>>>
>>> _______________________________________________
>>> IPython-dev mailing list
>>> IPython-dev@scipy.org
>>> http://mail.scipy.org/mailman/listinfo/ipython-dev
>>>
>>>
>>> _______________________________________________
>>> IPython-dev mailing list
>>> IPython-dev@scipy.org
>>> http://mail.scipy.org/mailman/listinfo/ipython-dev
>>>
>>
>> _______________________________________________
>> IPython-dev mailing list
>> IPython-dev@scipy.org
>> http://mail.scipy.org/mailman/listinfo/ipython-dev
>>
>>
>> _______________________________________________
>> IPython-dev mailing list
>> IPython-dev@scipy.org
>> http://mail.scipy.org/mailman/listinfo/ipython-dev
>>
>
>
> _______________________________________________
> IPython-dev mailing list
> IPython-dev@scipy.org
> http://mail.scipy.org/mailman/listinfo/ipython-dev
>


More information about the IPython-dev mailing list