[Libreoffice-ux-advise] new range name dialog proposal

[Markus Mohrhard]

Hello Christoph,

2011/11/2 Christoph Noack <christoph at dogmatux.com>:
> Hi Markus, all!
>
> I had hoped for some more time today, but before I'm never able to
> answer, some quick thoughts.
>
> Am Montag, den 31.10.2011, 00:06 +0100 schrieb Markus Mohrhard:
>> Hello Astron, Christoph, Kohei,
>>
>> sorry that it took me a bit to respond but I tried to finish my
>> remaining tasks for 3.5 before focusing on this task.
>>
>> Let me just respond to the comments you made in all your posts:
>
> Thanks in advance! :-)
>
>> I think that performance is an important factor even if it only
>> matters for 20% of our users. We are working hard to make Calc faster
>> and easier to use with complex documents. In my opinion it is a
>> central point because the person using calc nearly every day are the
>> people who have much more big documents. And then range names are an
>> advanced feature which helps a lot in handling big documents.
>> Therefore I think we should be extremely careful introducing a nice
>> dialog which on the other hand makes it nearly unusable for the people
>> who really need and use this feature.
>
> Well, as I said earlier ... to me it is not about the time for doing
> recalculations. But the time that gets spend by working with a modal
> dialog which itself needs some child modal dialogs that requires further
> interaction by the user.
>
> So, personally, I still can't estimate the impact of the recalculations
> vs. the improvements due to better interaction design. Still, what I'd
> like to know, why MSO doesn't have such problems? Do they generally
> perform better with regard to ranges and calculations (being meant to be
> a serious and neutral question)?

In my opinion the loss throught the delay we may get for large
documents is much more important than the one or two clicks you can
save through a modeless dialog. It might only affect 20% of our
documents and users but these are normally the ones where using range
names is important. If I only have 50 values in a document I don't
need a lot of range names and the win of the modeless dialog would be
high but for someone having several thousand entries and some dozen
range names might only win two clicks for every range name change but
loose some seconds while calc needs to recalculate.
I know that this might not be satisfying from an UX point of view but
in my opinion in calc the performance is at least as important as a
nice UI. If we make it harder to use this "advanced" feature for
bigger documents our target group will not benefit from a nicer
dialog.

>
>
>> I'm still not totally convinced that a modeless dialog is the best way
>> to go, so I list here my pros and cons for this:
>>
>> pro modeless dialog:
>>
>> - can be used like the navigator(opened all the time if needed)
>>
>> - modifications in the spreadsheet are possible while working with the
>> dialog( is that really an advantage? makes handling of the dialog much
>> more difficult if we for example delete a sheet and now several of the
>> range names disappear or are modified)
>
> I don't understand the "handling of the dialog ... more difficult". I
> assume that's the developer's point-of-view, not so much the user's
> point-of-view, right? Because, users would benefit from knowing what's
> going on ...

Maybe. I see a lot of pain that comes with this. The dialog suddenly
needs to know about a lot of changes in the document, e.g. moving
cells, deleting sheets, renaming sheets, creating range names with the
name box, creating range names through uno, ...

>
>> con modeless dialog:
>>
>> - marking a range in the spreadsheet can no longer automatically be
>> used for the expression line ( nice feature in my opinion )
>
> Mmh, I had hoped it would still be available when e.g. editing the
> expression ... click in the field, select the cell range. To me, the
> "mode change to edit" is much saner communicated than today.

Sure as long as you are in the expression line this is still possible
but for me it was always an advatage that I don't need to select it
btu can easily click somewhere in the document and it is immediately
used in the range/expression line.

>
>> - a lot of undo steps in our undo history, with a modeless dialog we
>> can still use one undo entry for all range name changes that were done
>
> Not sure whether this is neutral or even an advantage ... since we
> "design for errors", so people can go step-by-step
>
>> - we can recalculate the sheets when we close the dialog
>>
>>
>> So my idea would be to use something in the direction of Astron's
>> suggestion but modal. Then we can think about how we can make ( if we
>> still want) this dialog modeless without loosing performance for the
>> 3-6 release. We would have a nice dialog for the 3-5 release without
>> any regressions and will have enough time to think about how we can
>> address the disadvantages of a modeless dialog. And we should not
>> forget that the feature freeze is 5 weeks and we should have finished
>> our work on this except for some minor bug fixes.
>>
>> I know that this is not a perfect solution but in my opinion a new
>> dialog which introduces performance problems is neither perfect.
>>
>> What do you think about this idea?
>
> Well, finally its up to you being the developer who needs to take care
> about the implementation ...
>
> On the other hand, we've spend quite some time in ideation and concept
> improvements I personally don't want to say goodbye to (too quickly) :-)

I don't want to throw away the work. I jus proposed that we make this
change a two step change. The first one, for 3-5, more an UI change
that is still a modal dialog but with the new nice dialog layout and
then with enough time a change for the 3-6 release that, as long as it
makes sense, make it modeless. I think this has two big advantages, we
have enough time to discuss about the modal/modeless change and can
even try different ideas without problems and we don't need to
integrate too many changes into calc core in the last month before hte
feature freeze.

> The "perceived interaction quality" still matters to me, and is an
> important factor besides the "computational speed" - although I have
> much less experience with the latter one. My assumption is still, that
> all the improvements (like the non-modal dialog that lets users work
> with the document without opening/closing the name dialog)
> over-compensate the potential time loss (for people who do e.g. three
> changes at once).

And I think the problem is that we can at the moment only make
assumptions but we need to do it right with the first try. We don't
have time to correct a wrong design decission before the 3-5 feature
freeze. Threfore I would prefer if we would go the safe way and start
earlier before the 3-6 feature freeze for the second part.

If you want we can discuss that in more detail on libreoffice-dev ;)

Markus