Sum of variables not parsed correctly in multiple estimation syntax

[leostimpfle]

The multiple estimation syntax fails if combinations of variables are provided (f1+f2 in the example). This also fails in the formula parser refactor #1118, so will provide a fix there.

Example

import pyfixest as pf

data = pf.get_data(N=1_000, seed=0, model="Feols")
fml = "Y ~ X1 | sw0(f1, f2, f1 + f2)"
pf.feols(fml=fml, data=data)

My use case is running a specifications with all possible combinations of fixed effects (e.g., no fixed effects, unit fixed effects, time fixed effects, and unit and time fixed effects). fixest implements a special "multiverse stepwise" syntax for this, see #1136.

[s3alfisc]

Yes, this I did not support on purpose to reduce complexity of the implementation (which, as you know, is a little bit untidy until we merge your PR). So per se I would not say it's a bug. Though if you / other users would expect this to work, then likely it is? 😄 In principle it should be possible to support this? Do you happen to know if this works in fixest?

[leostimpfle]

Yes, it works in fixest. It shouldn't be too hard to support this. Only the parsing needs to be fixed