It would depend somewhat on the difference in the exposures. If you're talking about a 2X or greater difference then I'd stack them separately. However, if they are pretty close in exposure (say 1 minute versus 50 seconds) then you may want to stack them all together. You might also want to stack each separately and then combine them with a scale value to match each total integration time. So, if you had one hour total with one sequence and only 30 minutes on the other you could combine then in a ratio of two to one (two parts for the one hour sequence and one part for the 30 minute sequence).
Blends like the above are easy to do with PixInsight using that software's PixelMath. Also, many (most?) integration or stacking routines scale each sub based upon some form of weighting, which can be something like signal to noise or even exposure time. Thus, knowing how the integration combines the subs will help in determining whether you can stack the sequences all together or separately.
If you are going for some kind of high dynamic range result (HDR) then it's probably better to stack the sequences separately and then blend the results (to maintain detail in both the highlights and shadows).
Edited by james7ca, 17 May 2019 - 11:34 PM.