[BlueOnyx:16162] Re: Scientific Linux 7.0 released

[Michael Stauber]

Hi Ernie and Gerald,

> Scientific Linux 7.0 has been released, they took a fair while this time
> compared to CentOS.

Thanks for the update on this.

After wrapping up the transformation of PKGs for the new GUI I've
started looking into BlueOnyx 5209R for EL7 (CentOS 7 & SL 7) again.

However: I'm not sure if or when it will be ready.

It's certainly nothing to hold your breath for.

I still have to overcome some fundamental platform changes. For example
some of our C++ code in CCED needs an overhaul to make it compile with
the GCC compiler that's included on EL7. That is mostly done by now,
minus some uncritical compile warnings.

The PHP modules i18n and CCE that we need for the GUI need to be adapted
for PHP-5.4 (and later), too. However, the API for PHP modules has
changed between PHP-5.3 and PHP-5.4. I went through the module code to
clean them up and got them to compile, but it's not yet working as
intended. Need to do some more digging there and maybe need some help
from someone who knows a lot more about C than I do.

Lastly: Anyone who's also reading the kernel mailing-list is probably
aware of the heavy in fighting that's currently going on between the
supporters of Systemd and those that heavily oppose it. This has been
cooking for some time and turned into a bloody war between heavily
entrenched supporters on each side.

See http://www.vitavonni.de/blog/201401/2014012201-the-init-wars.html
for a short run down.

EL7 sadly shipped with Systemd instead of the usual SysV init. The
change to Systemd is such a pain in the ass that I've even contemplated
to *not* base the next BlueOnyx on EL7. I didn't reach a decision on
that yet, though.

I'm strongly opposed to having a multi-process and multi-libraried
process with the PID #1 that also thinks it must be a jack of all
trades. SystemD soon gets its own Terminal service *and* NTP server.
Imagine that. It is bloody ridiculous. Next they'll probably let it have
its own webserver, build in Skype and once it gained self-awareness they
give it it's own seat in the UN security council.

The net result is this:

[root at 5209r ]# ls -hals /usr/lib/systemd/systemd
1.2M -rwxr-xr-x 1 root root 1.2M Jul 30 09:22 /usr/lib/systemd/systemd

[root at 5208r ]# ls -hals /sbin/init
152K -rwxr-xr-x 1 root root 147K 25. Jun 2013  /sbin/init

If you try to make the master-process a "jack of all trades" you end up
with a bloated blob of 1.2 megabytes. Instead of a lean and mean
/sbin/init that's just 147 kilobytes and *just* does what you expect it
to do.

Security wise Systemd turns into a nightmare as these heaps and piles of
extra-code (beyond starting and stopping processes) contains a lot of
"baggage" that constantly needs to be audited.

This *will* bite us in the ass sooner or later.

But we'll take that road when we get to that crossing. At the moment the
main priority for 5209R is not the choice of OS, but getting i18n and
the CCE PHP modules ready for PHP-5.4 and newer. Then I'll look at
Systemd again and if I can overcome my heavy resentment about it there
might be a 5209R on CentOS7 and SL7. Or something entirely different.

-- 
With best regards

Michael Stauber