In https://github.com/CartoDB/camshaft/pull/355/, was added support for Merge chained analyses: _cdb_analysis_left_source/_cdb_analysis_right_source was being used as query alias. This is ok if there is only one Merge analysis, but if you add two analyses of this type to the same layer, a name conflict appears in the SQL query. To solve that, a buster was added at the end of every alias, just a numeric code generated using the current timestamp.
This timestamp worked to allow chained analyses, but this causes the query to change every time the map is instantiated, generating cache issues, as described here. So, we need each analysis with unique alias names, but this code shouldn't change every time the map is rendered.
Changes
Use analysis node ID as buster instead of current time.
Resources
Context
Merge
chained analyses:_cdb_analysis_left_source
/_cdb_analysis_right_source
was being used as query alias. This is ok if there is only oneMerge
analysis, but if you add two analyses of this type to the same layer, a name conflict appears in the SQL query. To solve that, abuster
was added at the end of every alias, just a numeric code generated using the current timestamp.Changes
buster
instead of current time.