On Wed, 2021-10-13 at 10:47 +1300, Andrew Bartlett via samba wrote:> On Tue, 2021-10-12 at 21:24 +0100, Rowland Penny wrote: > > On Tue, 2021-10-12 at 22:13 +0200, Ingo Asche wrote: > > > Hi Rowland, > > > > > > I asked Andrew the same, here's what he answered: > > > > > > - Yes, you start from 4.14.0rc1 as this is the branch point where > > > master > > > - split into 4.14 (working) and what would eventually be 4.15. > > > > > > Regards > > > Ingo > > > > > > > Well, no one has complained about 4.14.x version, the problem only > > seems to exist on 4.15.0 (and I cannot get the problem to not work > > for > > myself), so surely it is something in 4.15.0. I await Andrew > > commenting > > on this. If you do have to start from 4.14.0rc1, then you are going > > to > > be at it for sometime. > > G'Day Rowland, > > A bisect does a binary search, so even across large version gaps it > the > increased workload in testing is minimal. That is why we use that > approach, rather than (say) linearly selecting all commits. > > It is exceedingly unlikely the issue was introduced after 4.15.0rc1, > so > we must start before that. The correct spot to start is 4.14.0rc1 as > discussed, master development that became 4.15 started from that > point. > > Somewhere between... that point and ...> 4.15/master (they are so alike it doesn't change > much) this regressed, and we will find it. I've got an idea for one > thing it might be but only testing changes speculation into > verification. > > Also manual bisect testing is something our users can do that I don't > have the time to handle right now, so it is incredibly valuable. > > Confirming that 4.14.0rc1 works *in the test being done as the > validation* is important, as otherwise we could be chasing the wrong > thing as the decision basis. > > I trust this clarifies, > > Andrew Bartlett > -- > Andrew Bartlett (he/him) https://samba.org/~abartlet/ > Samba Team Member (since 2001) https://samba.org > Samba Team Lead, Catalyst IT https://catalyst.net.nz/services/samba > > Samba Development and Support, Catalyst IT - Expert Open Source > Solutions > > > > > >-- Andrew Bartlett (he/him) https://samba.org/~abartlet/ Samba Team Member (since 2001) https://samba.org Samba Team Lead, Catalyst IT https://catalyst.net.nz/services/samba Samba Development and Support, Catalyst IT - Expert Open Source Solutions
Then I will just proceed... ? Andrew Bartlett via samba schrieb am 13.10.2021 um 00:56:> On Wed, 2021-10-13 at 10:47 +1300, Andrew Bartlett via samba wrote: >> On Tue, 2021-10-12 at 21:24 +0100, Rowland Penny wrote: >>> On Tue, 2021-10-12 at 22:13 +0200, Ingo Asche wrote: >>>> Hi Rowland, >>>> >>>> I asked Andrew the same, here's what he answered: >>>> >>>> - Yes, you start from 4.14.0rc1 as this is the branch point where >>>> master >>>> - split into 4.14 (working) and what would eventually be 4.15. >>>> >>>> Regards >>>> Ingo >>>> >>> Well, no one has complained about 4.14.x version, the problem only >>> seems to exist on 4.15.0 (and I cannot get the problem to not work >>> for >>> myself), so surely it is something in 4.15.0. I await Andrew >>> commenting >>> on this. If you do have to start from 4.14.0rc1, then you are going >>> to >>> be at it for sometime. >> G'Day Rowland, >> >> A bisect does a binary search, so even across large version gaps it >> the >> increased workload in testing is minimal. That is why we use that >> approach, rather than (say) linearly selecting all commits. >> >> It is exceedingly unlikely the issue was introduced after 4.15.0rc1, >> so >> we must start before that. The correct spot to start is 4.14.0rc1 as >> discussed, master development that became 4.15 started from that >> point. >> >> Somewhere between > ... that point and ... > >> 4.15/master (they are so alike it doesn't change >> much) this regressed, and we will find it. I've got an idea for one >> thing it might be but only testing changes speculation into >> verification. >> >> Also manual bisect testing is something our users can do that I don't >> have the time to handle right now, so it is incredibly valuable. >> >> Confirming that 4.14.0rc1 works *in the test being done as the >> validation* is important, as otherwise we could be chasing the wrong >> thing as the decision basis. >> >> I trust this clarifies, >> >> Andrew Bartlett >> -- >> Andrew Bartlett (he/him) https://samba.org/~abartlet/ >> Samba Team Member (since 2001) https://samba.org >> Samba Team Lead, Catalyst IT https://catalyst.net.nz/services/samba >> >> Samba Development and Support, Catalyst IT - Expert Open Source >> Solutions >> >> >> >> >> >>
Hi Andrew, Hi Rowland, found a bad one: 4.15.0pre1-GIT-997fbcbc902 After installing this one login failed. How should I proceed? Regards Ingo Andrew Bartlett schrieb am 13.10.2021 um 00:56:> On Wed, 2021-10-13 at 10:47 +1300, Andrew Bartlett via samba wrote: >> On Tue, 2021-10-12 at 21:24 +0100, Rowland Penny wrote: >>> On Tue, 2021-10-12 at 22:13 +0200, Ingo Asche wrote: >>>> Hi Rowland, >>>> >>>> I asked Andrew the same, here's what he answered: >>>> >>>> - Yes, you start from 4.14.0rc1 as this is the branch point where >>>> master >>>> - split into 4.14 (working) and what would eventually be 4.15. >>>> >>>> Regards >>>> Ingo >>>> >>> Well, no one has complained about 4.14.x version, the problem only >>> seems to exist on 4.15.0 (and I cannot get the problem to not work >>> for >>> myself), so surely it is something in 4.15.0. I await Andrew >>> commenting >>> on this. If you do have to start from 4.14.0rc1, then you are going >>> to >>> be at it for sometime. >> G'Day Rowland, >> >> A bisect does a binary search, so even across large version gaps it >> the >> increased workload in testing is minimal. That is why we use that >> approach, rather than (say) linearly selecting all commits. >> >> It is exceedingly unlikely the issue was introduced after 4.15.0rc1, >> so >> we must start before that. The correct spot to start is 4.14.0rc1 as >> discussed, master development that became 4.15 started from that >> point. >> >> Somewhere between > ... that point and ... > >> 4.15/master (they are so alike it doesn't change >> much) this regressed, and we will find it. I've got an idea for one >> thing it might be but only testing changes speculation into >> verification. >> >> Also manual bisect testing is something our users can do that I don't >> have the time to handle right now, so it is incredibly valuable. >> >> Confirming that 4.14.0rc1 works *in the test being done as the >> validation* is important, as otherwise we could be chasing the wrong >> thing as the decision basis. >> >> I trust this clarifies, >> >> Andrew Bartlett >> -- >> Andrew Bartlett (he/him) https://samba.org/~abartlet/ >> Samba Team Member (since 2001) https://samba.org >> Samba Team Lead, Catalyst IT https://catalyst.net.nz/services/samba >> >> Samba Development and Support, Catalyst IT - Expert Open Source >> Solutions >> >> >> >> >> >>