The current "Generic IPF Interface Specification" documents specify delta times in selection policies as non-negative. However in many practical cases it would be helpful to allow negative values to separate successive products in the timeline from each other, when the product times overlap. (The case in point were two Sentinel-5P L1B files, whose times overlapped, because the underlying orbit times from two successive Orbital Event Files overlapped.) Such a change would not break the backwards compatibility of the language or the functionality of existing selection rules, but it would be a deviation from the above mentioned specification.
While the code change itself is minor (there is an explicit check for negative numbers in the parser, which needs to be removed), the required change for the documentation and for the unit tests is substantial. Due to this and the deviation from the specification a formal decision is required before implementing this change.
The current "Generic IPF Interface Specification" documents specify delta times in selection policies as non-negative. However in many practical cases it would be helpful to allow negative values to separate successive products in the timeline from each other, when the product times overlap. (The case in point were two Sentinel-5P L1B files, whose times overlapped, because the underlying orbit times from two successive Orbital Event Files overlapped.) Such a change would not break the backwards compatibility of the language or the functionality of existing selection rules, but it would be a deviation from the above mentioned specification.
While the code change itself is minor (there is an explicit check for negative numbers in the parser, which needs to be removed), the required change for the documentation and for the unit tests is substantial. Due to this and the deviation from the specification a formal decision is required before implementing this change.