metze-samba

@lhoward can we at least revert commit 1cdc9d5f3cff0288846c29c35ee91b6056a2e2bb for now, so that krb5_init_creds_step() doesn't get any external users with the broken prototype?

> Sounds like a sensible idea. @nicowilliams? (Sorry I don't have time to work on this right now.) See https://github.com/heimdal/heimdal/pull/974

@lhoward I think this should be labled as bug

> Looks good in principle although I do not have time to review it now, so @nicowilliams or @jaltman will have to. > > I didn't look too closely so...

> > Looks good in principle although I do not have time to review it now, so @nicowilliams or @jaltman will have to. > > I didn't look too closely...

> Yes, this is a better approach. Suggest the following symbolic names: > > ```c > #define KDC_REQUEST_KV_PA_SUCCEEDED_KVNO "pa-succeeded-kvno" /* heim_number_t */ > #define KDC_REQUEST_KV_PA_FAILED_KVNO "pa-failed-kvno" /* heim_number_t */ >...

What's the best way to bisect the test failures? Is there a way to a travis ci build without a pull request?

It's failing with this: AIL: check-context =================== Creating database Doing database check Starting kdc ./check-context: line 176: 10513 Segmentation fault: 11 (core dumped) KRB5_CONFIG=${KDC_KRB5_CONFIG} ${kdc} --detach --testing kdc failed to...

Ok, it no longer segfaults, but we get this: FAIL: check-context =================== Creating database Doing database check Starting kdc Test gss_acquire_cred_with_password test_context: gss_acquire_cred_with_password: gss-code: 851968 Miscellaneous failure (see text) --...

Heimdal has the check_pac option on by default, we could just use a verified PAC as indication to skip the transited check, that would not need any additional options. I'll...