Activity Types can have multiple attribute structures?

[Anonymous]

We have been able find out what are the attributes for most of the activity types in Marketo, by generating activities and querying from .Net with the getLeadChanges method.

We know that for each type there is a different set of fields, but now we have found out that for a given type or types (2 so far) there could be more than one set of attributes. I suppose this depends on how or from where the activity was generated.... right?

We have found this is true for MergeLeads and SendEmail... 

We generated a couple of Lead Merges, queried the API, and got back this set of attributes:

MergeIDsMergeFieldsMasterUpdatedMergedinSalesMergeSourceLeadID

It happened that days after that, after generating more merges for testing purposes, for some we get this same set of attributes, and for other we get this:

MergeIDMergeFieldsActivitiesMerged
AnonymousSource
CampaignFlowsMerged
WebTokensMerged
LeadID

Because of the architecture of our application, we need to pre-create the metadata for every activityType to treat it as a data entity, or a table... So, having this scenario, and also not knowing if one day someone will create an activity in a way we have not covered, is a bit scary. So, my question would be... Is there any way to have a list of all possible attributes or sets of attributes forall the activity types? Thanks JP

[Anonymous]

Unfortunately these can change over time. Our architecture has these fields in a freeform blob in JSON. You might try the decribeMObject call on Activities, and I'd also suggest you build your code to discard unneeded columns and just focus on capturing the pieces you know you need.

[Anonymous]

In relation to the question above, how come that the MergeIDs (Note the plural form) returns 'Too Complex' as value instead of the (secondary) LeadID that was merged?

Could you please explain (or send sample API script) how the IDs of the merged records can be extracted via SOAP API?

[Anonymous]

Maybe we should start not by asking how the merged IDs can be extracted, but if they can be extracted. So far, I've found the data available in the API to be rather underwhelming.