Taylor Blau <me@xxxxxxxxxxxx> writes: > The conversion in c7c4bdeccf was mostly plug-and-play, with the snag > that some instances of strvec_pushv() became guarded with a NULL check > to ensure that the second argument was non-NULL. Thanks. I am not sure about the reference to plug-and-play but most of the changes in the commit were mechanical conversion akin to @@ struct child_process C; expression V; @@ -C.env = V; +strvec_pushv(&C.env_array, V); but a few have rewritten them to @@ -C.env = V; +if (V) strvec_pushv(&C.env_array, V); And this hunk is one of them. If we know V is never NULL, the check becomes redundant, and we can tell tmp_objdir is not NULL at this point in the control flow. If C.env (or C.env_array) weren't empty by the time the control reaches at this point, however, this rewrite changes behaviour X-<. Another one in connected.c looks like this: diff --git a/connected.c b/connected.c index 35bd4a2638..ed3025e7a2 100644 --- a/connected.c +++ b/connected.c @@ -109,7 +109,8 @@ int check_connected(oid_iterate_fn fn, void *cb_data, _("Checking connectivity")); rev_list.git_cmd = 1; - rev_list.env = opt->env; + if (opt->env) + strvec_pushv(&rev_list.env_array, opt->env); rev_list.in = -1; rev_list.no_stdout = 1; if (opt->err_fd) If opt->env can be NULL, the original code would have just cleared existing rev_list.env but new code instead keeps whatever is in rev_list.env, which technically speaking changes the behaviour but we are saved by the fact that rev_list.env hasn't been touched at this point in the control flow since it was initialized, which makes it a correct conversion. There are few others in editor.c::launch_specified_editor(), run-command.c::run_command_v_opt_cd_env_tr2() and run-command.c::run_hook_ve(), but the story is identical in these places. > diff --git a/builtin/receive-pack.c b/builtin/receive-pack.c > index 9aabffa1af..f673e0e76e 100644 > --- a/builtin/receive-pack.c > +++ b/builtin/receive-pack.c > @@ -2214,8 +2214,7 @@ static const char *unpack(int err_fd, struct shallow_info *si) > close(err_fd); > return "unable to create temporary object directory"; > } > - if (tmp_objdir) > - strvec_pushv(&child.env_array, tmp_objdir_env(tmp_objdir)); > + strvec_pushv(&child.env_array, tmp_objdir_env(tmp_objdir)); > > /* > * Normally we just pass the tmp_objdir environment to the child So, in short, looks good, well spotted, and will apply. Thanks.