Problems building bind 9.18.1 on FreeBSD

[Ondřej Surý]

> On 25. 3. 2022, at 17:43, Dennis Clarke via bind-users <bind-users at lists.isc.org> wrote:
> 
> The entire ISC *preocess* has become gradually more toxic for at
> least a decade.

The only thing that’s toxic here is your current and previous communication
to the bind-users mailing list. Please, stop.

> Many systems and architectures are slowly dropped
> and they fall to the wayside to be forgotten and abandoned.

Yes, and that’s a good thing.

Also you cannot expect so few developers to support every possible
nook and cranny you can find in the operating system world. Naturally,
the developers pick the stuff they are going to work on.

The supported and unsupported platforms are clearly
described in the documentation:

https://bind9.readthedocs.io/en/v9_18_1/requirements.html#supported-platforms

including a paragraph on the community maintained platforms:

> These systems may not all have the required dependencies for building BIND
> easily available, although it will be possible in many cases to compile those
> directly from source. The community and interested parties may wish to help
> with maintenance, and we welcome patch contributions, although we cannot
> guarantee that we will accept them. All contributions will be assessed against
> the risk of adverse effect on officially supported platforms.

The requirements for building BIND 9 are aligned with what you would
expect by system created in 2022.

> Regardless if you have a decent compiler or really substantial processors and
> memory and various standard compliant headers etc. One may file a bug
> report and then be politely told to go away.

Stop spreading FUD. I cannot find any issues opened by you that
would support your claims. The last time your communication involved
accusations of “gnuisms” (which I showed you that are not gnuisms)
and system tests.

The system tests have more requirements than the standard build process
and we (the developers) are more lenient when it comes to dependencies
to build and run system tests.

> Expect that. That is what will happen.

And more FUD. This is not simply not true. We do carefully consider every
issue. We are just not going to impose more work and more pain on us
and use C89 to implement everything from scratch to just support some
niche system.

We focus on building DNS server, which means it’s OK to use modern C
specification, external libraries (networking, TLS, HTTP/2, etc…) because
we should do what we understand best - DNS. But as a matter of fact, we
have contributed back to many of the libraries we are using to build BIND 9
(libuv, cmocka, …). This is how the open source should work - everybody
improving the reusable components which works for everybody’s benefit.

> The build process has become more toxic and complicated
> and even outright obfuscated to the point that it is hopeless

This is just pure nonsense. On the contrary, the build system in 9.18+ become
more standard - it uses autoconf + automake + libtool and the syntax is more
declarative (to the extent possible with automake). That makes the build system
easier to understand and easier to improve.

> to even bother looking at a system running FreeBSD on RISC-V or some
> other UNIX on just about any architecture.

Hmm, there are no issues opened about RISC-V support:
https://gitlab.isc.org/search?search=RISC-V

So, more FUD?

In fact, we are happy to support modern architectures when they
become commodity. I am not aware of anything that would intentionally
prevent building modern BIND 9 on RISC-V or any other new architectures
like OpenPOWER. There’s very little code in BIND 9 that’s architecture
dependent. In fact, the only place that I am aware of is isc_rwlock_pause(),
that in fact has a graceful fallback - if the processor “pause” is not available
it does nothing.

> Even Python3 was slammed into the mess
> for a code base that was always pure clean portable C.

No, that’s not true. Python3 was optional in BIND 9.16 and the
Python components were replaced dnssec-policy (in C) in 9.18+.

We require python + pytest as testing framework for system tests.
The system tests are primarily meant for developers.

> My opinion is
> that ISC is all about the "Support Subscription" business and quite
> frankly it stopped being any sort of a welcome "community" a long time
> ago.

Stop spreading FUD. You are the one behaving rude and arrogant here.

> To the point that I now consider it a waste of time to even bother
> dealing with BIND.

That’s obviously not true since you still lurk here. Maybe you should
consider putting your foot where your mouth is?

> If it works ( at all ) on the future FreeBSD 13.1
> release then be happy and say nothing. Don't expect it to stay that way.

More FUD not supported by any evidence. FreeBSD is one of our primary
platforms.

> Expect future problems and more toxic traffic

Even more FUD not supported by any evidence at all.

> until it is all just a Linux SystemD service.

It’s **systemd**, not SystemD. And we are not planning to depend on
specific supervisor, but in fact, we are happy to add **optional** support
for any available supervisor **if it makes sense**.

Ondrej
--
Ondřej Surý (He/Him)
ondrej at isc.org

My working hours and your working hours may be different. Please do not feel obligated to reply outside your normal working hours.