one codebase, simultaneous ensembles

[schmonz]

When two ensembles are naively using mob on the same repo, they’ll trample on each other’s work without meaning to. It looks like (I haven't actually tried this) if everyone sets MOB_WIP_BRANCH_QUALIFIER correctly, each ensemble session will have its own distinct working branch. This is a tradeoff, of course: in exchange for the benefit that ensemble handoffs remain effortlessly conflict-free, we pay in deferred integration effort. Under the circumstances, probably a good tradeoff for most.

Assuming I have a correct understanding of MOB_WIP_BRANCH_QUALIFIER, how do we ensure that everyone always sets it correctly every time, as effortlessly as possible? My best idea so far is for everyone’s shell .profile or equivalent:

MOB_PROG=$(command -v mob)

mob() {
    echo >&2 "nope, run mob1 or mob2"
}

mob1() {
    MOB_WIP_BRANCH_QUALIFIER="mob1" ${MOB_PROG}
}

mob2() {
    MOB_WIP_BRANCH_QUALIFIER="mob2" ${MOB_PROG}
}

Then nobody can run mob directly and risk not having sufficient configuration; instead they have to run the one whose name matches which session they’re in. This is information we're always aware of, so it seems okay to rely on.

My idea here only works with Bourne-derived Unix shells, and mob1 and mob2 are not available to be called from other programs.

Are there other ways of addressing the need that are as portable as mob is? Possibly by enhancing mob?

[hollesse]

Hi Amitai, do you know the flag -b for mob start? Example mob start -b mob1 this sets the branch qualifier for that specific command.

[schmonz]

Ah, that's a more concise way to express MOB_WIP_BRANCH_QUALIFIER. Still interested in how mob might (optionally of course) more natively and automatically ensure that (1) each simultaneous ensemble is on a distinct branch and (2) within each ensemble, every participant has the same branch.