Christian Perrier writes: > However, later in the bug log, and in his reassignment message, > Matthias says "even if a shell's postrm is a C > program, it needs to deregister the shell. You basically argue, > because remove-shell has a deficiency (the implementation as a shell > script), every shell should add a workaround for that deficiency? > I don't get it." > > I understand this as "moving the code to the prerm would be a > workaround". > > However, a quick look at some other shell packages postrm scripts > (namely dash and pdksh) showed me that they usually call remove-shell > in their prerm scripts.
well, zsh does remove it in the postrm. > So, at first look there does not seem to be an established policy to > call remove-shell in postrm. > > So, is there actually some reason, not mentioned in this bug log, for > the remove-shell code to be in postrm ? If the prerm fails, you have to make sure, that add-shell is called again in the appropriate place. Yes, I consider this as a workaround, if I'm not able to call remove-shell from the postrm.
