Would it be possible by any chance to add an argument intercept=TRUE or FALSE in hqreg to also support models without intercept, as is possible in glmnet?

I was wondering if it would also be possible to allow argument value preprocess="none" so that no variable rescaling would be performed in hqreg?

Was wondering if support for the case where the covariate matrix X is sparse is planned any time in the future?

I would like to use hqreg but apply positivity constraints on the fitted coefficients (box constraints would also be nice, but for my application positivity constraints would be enough). In glmnet this is possible by setting lower.limits to 0 (and there's also an extra argument upper.limits for box constraints).

I was just wondering if this feature might be planned in some future update of the package?
[I understand it is possible to add positivity constraints in Newton's method, see
https://scicomp.stackexchange.com/questions/23499/best-way-to-add-a-positivity-constraint-to-newtons-method]

Hi,
would it be possible to add a weight argument for each training observation in order modify the loss function in particular for the quantile method: