Sequence type

Hi Oussama,

I am the product manager for Sequences, and thank you for your request.

I will check on whether there are other uses of the last updated timestamp behaviour that we currently have, that we might break if we adjust this behaviour.  But your requirement to have the ability to know when new data points are added to a Sequence, in a period of time, is well received and understood.

I will reply soon once I’ve dug into the question a bit more.

Kind Regards, Glen

Hi again Oussama,

Unfortunately, due to that field already having a defined behaviour (it is updated on metadata changes), to fulfil your specific request would be a breaking change that we cannot fulfil as other customers may be using this field for other purposes.

But, I am keen to understand the use case behind the request, and I will work with the team to explore what solution possibilities we have to fulfil your requirement.

If you can provide as much detail as possible about the use case, this will be very helpful.

Kind Regards, Glen

p.s. if you could let me know the language you’re programming in against the API too, I’ll check if there’s any functionality in our SDKs that might also be helpful here.

Dear Glen,

Thank you for your response and willingness to explore solutions to our challenge. To provide more context on our use case, here is a detailed description:

We are using CDF to synchronize and manage data from drilling activities, which are not time series data. This data is stored in sequences within CDF. After initial storage and transformation in CDF, other transformations are triggered on a scheduled basis. These transformation processes are designed to query not the entirety of sequences, but specifically those that have been recently updated.

Our challenge is that adding new data points to a sequence does not necessarily update the sequence's metadata timestamp. We attempted to resolve this by adding or modifying certain metadata (like the last index and the number of data points), but this has proven unreliable. Especially if changes occur within the data points of a sequence, we cannot force an update to the sequence's metadata.

We are seeking a solution that would reliably and efficiently detect changes made to a sequence, whether it's adding new data points or modifying existing ones. One potential solution we are considering is implementing a data versioning system. Each change would create a new version of the data point, allowing us to track modifications without solely relying on the metadata's update timestamp.

Thank you for your attention to this matter. We look forward to your ideas and suggestions.

Kind regards,

Oussama

We use python SDK!

Hi Oussama, I’ll discuss with the development team and see what we might recommend.

I like the sound of your versioning approach!

Glen

Hi Oussama, 

Would an approach where we implement a subscriptions API, similar to that of Time Series satisfy this requirement?

Kind Regards, Glen