How should I stack subs with different exposure lengths?

Precisely what David said above.  APP handles it all for you!

You could try DSS. It has multiple tabs or 'groups' that you can use to stack images of different exposure length.

I have a similar question, but from a different perspective, namely is there any benefit to stacking images of varying exposure durations? I understand that it can effectively increase the dynamic range when you're imaging a target that might for example have a saturated core but finer nebulosity at the other edge (M42 and NGC 2070 being two common examples).

I'm also using APP and as a matter of course try to keep saturated pixels to a minimum by keeping exposures short, but there are times when I prefer to go longer to reduce the number of subs and of course swamp noise.

In DSS you can stack each frame 'time' group in different tab.  The Main DSS tab for example the 60'  frames with its Darks, Flats and Bias, and the Group 1 tab the 90' frames with its Dark, Flats and Bias.  Then DSS will calibrate the frames of each time and after stack them together and produce the Autosave stacked image.

If the ISO is the same for both frames groups, then you do not need put other Bias in the Group 1, only in the Main tab because DSS will use the Master Bias generate of the Main tab for do the calibration in the Group 1 too.  Same for the Flat.

The same procedure is used for stack different nights imaging in DSS.

In this case, as the difference in time is not much, you can try to see the result stacking all together, but you need once you put all the light frames in the tab, choose for example all the 60' frames, right click and in properties change the time from 60' to 90'.

In Siril as was suggested, you can calibrated each frame time group separate  getting the db_pp_light frames.  Once you have the files calibrated of both groups time frames, you can put all them together and proceed with the registration phase r_pp_light and after the stack phase.