· Aug 25, 2016

Caché 2016.2 and 2016.3 Field Test Programs

This announcement is to inform clients about changes in our 2016.2 and 2016.3 field test programs.  

The 2016.2 field test has been extended in order to address important syntax changes related to JSON. Simply stated, the JSON syntax introduced in 2016.1 had several inconsistencies that we did not want to perpetuate.   This is now addressed and the updated 2016.2 field test will be published today.  Concurrently, the syntax changes and guidance for developers will be posted on the Developer Community shortly. JSON is critically important to our Caché roadmap.

Doc DB has been removed from the field test due to the syntax changes. We have significant refactoring to do in our Document Data modeling code and we did not want to further delay the release of 2016.2 to accommodate that development. This feature will return at a later date.

The extension of the 2016.2 field test conflicts with 2016.3 field test schedule and having two releases in field test simultaneously would result in a delay of a general release for both versions.  Since 2016.2 has had significantly more client interest and is closer to a general release than 2016.3, we are cancelling the 2016.3 field test and withdrawing our plans to release it.  This will expedite both the general release of 2016.2 and the upcoming field test of 2017.1 later this year.  We regret any confusion that this may cause and we expect that it will lead to our releasing higher-quality versions to you faster.
If you have questions or concerns regarding this announcement, please feel free to post them on the Developer Community or contact your Account Manager or Sales Engineer. You are also welcome to contact the Worldwide Response Center (

Bill McCormick
Director of Product Management

Discussion (11)4
Log in or sign up to continue

I opposed strongly against the naming conventions used in the $-systemMethods so I should be happy.

But facing such a backwards incompatibility in 2016.2 makes me very sad. We've planned several major new features in our application all using REST  and JSON to be released on 2016.1 very soon. Now we have to review a lot of code and try to decide what to do with them in order to be compatible with future releases. I don't like the suggested macro solution, but it might be the only viable way. We might even switch back to our own JSON implementation which we have used for years.

The most important other issue I opposed to is the interface of the Dynamic array, it is not compatible with the existing COS %Collection interface. Is there any change that will be 'fixed' too ?

This really boiled down to the band aid removal technique. The existence of a switch implies people are not under any pressure to make changes to their code. It gives them a reason to wait. The longer we left this in the longer we had to support it and the more questions about why something might work in one syntax but not the other. I hate causing partner's pain but in this case the sooner the better.