Technical strategy and performance

[Jason Azze]

On Thu, Jun 30, 2016 at 8:31 AM, Eric S. Raymond <esr at thyrsus.com> wrote:
> Jason Azze <jason at azze.org>:
>> > The thing about these users is they're intensely conservative.  They
>> > won't buy a time service implementation that doesn't reassure them by
>> > looking and smelling like what they're used to.
>>
>> This is why I try to make noise when things are broken on RHEL/CentOS
>> 6.x.
[snip]
> OK, then this is a good time to ponder the big question: from your P.O.V.
> are we doing enough to support 6.X?  If not, what needs to be fixed?
> I agree this is important and am sure Mark will too.

I'm finding it harder to answer this question than I should.

Issue #63 was the first blocker to compiling on RHEL 6. It was fixed
in a reasonable (to me) amount of time given what I know about the
workload of the developers and compared to other projects I'm familiar
with. However the problem was discovered by the user community rather
than by the project's build automation. I think it would be an
improvement to add a CentOS 6.8 builder to buildbot so that build
failures will be visible on commit.

The documentation build on RHEL 6 is broken right now, but I have been
remiss in writing a formal report for the issue tracker. I hope to
find the cycles this weekend.

That said, at the current maturity level of the project, I think you
are "doing enough".

I would be happy to one day be able to type yum install ntpsec on a
CentOS 6 box to get the goods. To do that, I think you need to start
upstream in Fedora and be able to make it all the way through their
package building automation without error.

Caveat: I am one sysadmin working in a Red Hat / SuSE dominated field.
My perception is certainly colored by this. The Usenix LISA conference
is in Boston this December. I'm going to try to gather some intel
there. (Maybe we can get Daniel to give the time talk!)