tpartner schrieb:
As I recall, this was discussed many months ago and the team agreed that it was acceptable to add new CSS stylings in upgrading from 1.91 to 1.92; and that was not considered a break on backwards compatibility since templates for even active surveys could be easily changed.

I added this note to the upgrade hints page.

What should be considered mandatory is tricky part here.
this is where it gets tricky with this kind of combined question types. Maybe it should be added to an upgrading section on the wiki that explains the difference between 1.92 and up and the way it was before. Also your post on css styles is something that should be easy to find I think. Do we have wiki upgrade hints?

I did and do agree that adding CSS is acceptable but I was under the impression that it was for presentation purposes only, not required for new survey behaviour.

Here's an update to my code that resets the mult-numeric inputs to 0 if they are left empty.

Without insight in the reasons why it behaves like it does in the current version, I would suggest that the way it worked in previous versions (no zero needed in empty fields) is more intuitive and what you would expect when configuring such a question (and how other systems I worked with behaved).

The suggested work around, with extra code in the template and setting minimum questions to zero does not work well. Once numbers are entered users must click twice to proceed to the next question which is practically unusable (the user experiences is no longer intuitive).

Also I am a bit concerned about the templates not being backwards compatible, as we have invested in creating our own.

Tony created a workable work around, adding zero's to the fields that are empty, this works well for me now.

Thanks all for the support.

aukevann schrieb:
The core question is, "for questions that have min_sum_value or equals_sum_value, is 'no response' acceptable".

When I initially implemented this, 'no response' was not acceptable in those cases, but the team voted the other way, so I changed it. I can change it back, but not without first getting approval from the team (as that might mess up other operational surveys and change the behavior of a stable release.
aukevann schrieb:
It is possible that we've been blowing this out of proportion a bit. The CSS extensions were meant purely to provide additional functionality, not to change functionality. They were meant to be backwards compatible. However, we have run into some cases like yours where not all tips were behaving as expected with hide_tip, which is why I suggested adding in those lines of CSS code.

I don't have a good sense of exactly what is working differently in 1.92 vs 1.91 with a built-in 1.91 template. Perhaps you could open a bug report about the templates clarifying the odd behavior you are seeing. If there is an actual bug, we could fix that in the core and authors wouldn't need to upgrade their templates.

Rather than debugging the work around, I would suggest for future development to either allow "no response" with the min_sum_value or equals_sum_value question or to make this an option. Would make it clear for new administrators how to set it up.

This happens in many software when a new version comes out. Custom templates often need some adaptation. Of course this is a pain and in an ideal world I would love to see the custom templates not breaking, but I think this is something we have to live with for the sake of new features and functionalities.

In general I would love to see a diferent template system all over, because the current one lacks a little bit of flexibility, but in general it is simple and works, so I can live with it.

aukevann schrieb:
Since this broke functionality you were depending upon, we can add a new advanced question option to 1.92+ to fix that "bug".

The proposed solution is to keep the old behavior on upgrade (so if you re-upgrade, or pretend to re-upgrade by setting the DBversion in your database to 154 and letting the upgrade process run again), but to have the default for new questions be the new behavior.

This is fixed in 1.92+ build 120606.

Existing surveys will use the old behavior (requiring an answer if equals_sum_value or min_sum_value are set). There is a new advanced question setting called "value_range_allows_missing" which lets authors choose the new (default) or old behavior.