The origin of Galactic halo stars and the contribution of globular clusters (GC) to this stellar population have long been (and still are) debated. The discovery of multiple stellar populations with peculiar chemical properties in GCs both in the Milky Way and in Local Group galaxies recently brought a renewal on these questions. Indeed most of the scenarios that compete to reproduce the present-day GC characteristics call for fast expulsion of both gas and low-mass stars from these clusters in their early infancy. In this framework, the initial masses of GCs could have been 8 to 25 times higher than their present-day stellar mass, and they could have contributed to 5 to 20 % of the low-mass stars in the Galactic halo. Here we revisit these conclusions, which are in tension with observations of dwarf galaxies and of young massive star clusters in the Local Group. We come back in particular on the paradigm of gas expulsion from massive star clusters, and propose an alternative interpretation of the GC abundance properties. We conclude by proposing a major revision of the current concepts regarding the role massive star clusters play in the assembly of galactic haloes.