systemd Breached One Million Lines Of Code In 2017

[aht0]

Principal design nuances of systemd which specifically have been causing problems for people:
-Hardcoded compile-time "FallbackDNS", which many distros have simply set to 8.8.8.8
-systemd-resolved does treat all configured DNS servers as equivalents while in real life they may very well not be so. Despite RFC's Poettering keeps pointing at and closing the requests for a change in behaviour. For example: When you do have your own company's DNS with a whole bunch of custom settings including the LAN, having systemd-resolved suddenly jump over to 8.8.8.8 breaks things.
-systemd-resolved switches between these DNS servers more or less randomly. Anyway, it's supposed to try another when it's failing to connect it's current DNS server. It does not.

You of course won't have anything to complain about it when public DNS suffices for configuration. Which are most home users with a PC.

Folks who would now recommend to recompile systems-resolved with different FallbackDNS configured, switching a distro or anything other along the lines - please sit on something pointy and long..

[sthalik]

Javascript for quoting text parts doesn't work for me, responding with old-style quotes. Sorry for that.

pal666,

>> The proper engineering solution is having multiple projects, each maintained by a separate group of people.
> where do you get this shit from? one of reasons of open source success is that it allows cross-project development
>> Leaving a big blob of a project in hands of a monolithic group of committers creates an echo chamber.
> what is this "big blob of a project" and "monolithic group of committers" shit is? are you off your meds?

First off, watch your insults and your tone. They're only telling about yourself, not me. As for the matter at hand:

All the heavily-reused projects like pixman, freetype2, zlib, qt, gtk, glx et al. must and do agree on an API. This provides a separation of concerns, as each library's internals aren't exposed to its consumers. Familiarize yourself with the concept of "information hiding".

For a monolithic project, the developer needn't hide any internals, or keep a stable API/ABI. Ideally we'd have a public API, a private API for each subproject, and an internal API each subproject exposes to other parts of the monolithic projects. We have that in the Linux kernel, except for the "stable internal API" part. Since there's no natural separation in monolithic projects, it takes a significant engineering effort to keep sane practices (thus preventing a large project from falling apart).

>> This should never be a single repository!!!
> why??????? it is 50 times smaller than kernel !!!!!!!!!!!

That's not any measure of complexity. Also, the kernel is mostly drivers using internal APIs for the subsystem they're exposed in.

> isn't it funny that you chose to contrast systemd with behemoth reimplementing everything in its sight, including its implementation language?

It's cruft from C++03 days. I agree they should deprecate it more heavily now.

Also it's an example of a large project with saner engineering practices. Somehow I managed to use qtbase without being obliged to pull qtmultimedia, qtquick, qtwebkit, or plenty of other stuff. There's a healthy separation between them.

I suggest you familiarize yourself with project management and sane engineering practices before insulting me.

[pal666]

Why'd I post this and not 'original' bug report?

because you were not competent enough to read last report's message which explains that it is only a matter of configuring proper search domain on interface

And scrolling downwards you can see 400 loc shell script being offered

if you are against all shell scripts now, scroll more and find nice c code, provided for systemd for that https://github.com/systemd/systemd/pull/7576/files

[pal666]

The proper engineering solution is having multiple projects, each maintained by a separate group of people.

where do you get this shit from? one of reasons of open source success is that it allows cross-project development

This way, user of a dependency will spot if the API is crap, or otherwise non-functional.

while when user of dependency is the same person as api author, he can make it functional from the start

Leaving a big blob of a project in hands of a monolithic group of committers creates an echo chamber.

what is this "big blob of a project" and "monolithic group of committers" shit is? are you off your meds?

This should never be a single repository!!!

why??????? it is 50 times smaller than kernel !!!!!!!!!!!

Their tests are also short and useless, barely checking of anything works at all. Proper practice here is adding tests for bugs that just got fixed.

show me sysv tests in comparison. or write them better tests, they will surely accept your valuable contribution

Compare this with Qt 5, which is a big project in its own right. I just build qtbase, qtserialport, and qttools. Only the former is "big" by any standard.

isn't it funny that you chose to contrast systemd with behemoth reimplementing everything in its sight, including its implementation language?

[pal666]

And before you start insisting it's a isolated case: try simple thing. Type into Google: systemd dns vpn problem
See how many returns you get. I got in excess of 100k.

i typed linux dns vpn problem
and got 578k hits. which proves what?

[pal666]

Who kept pointing out how .bad and cumbersome sysV init and shell scripts combined are, and how systemd is supposed to make it all go away.

systemd is supposed to make all bad sysv shell scripts go away. but systemd does not forbid shell scripts, it allows to run any programs in any language. who told you otherwise?
and in any case it is a script for openvpn, not for systemd

[pal666]

Yes, I can disable journald completely (which I do) but it won't help me in creation of all system logs completely in ram in a god damned circular buffer.

obviously, to get all system logs completely in ram the first step is to enable journald. is it so difficult to understand?

Why timesyncd?

why can't you read its man?

How many implementation of time syncing existed before? Wasnt' that enough?

they are all included in slackware 1.0
wat's wrong with you? go back to your slackware 1.0

Why my machine is leaking dns now over a vpn connection? Ah, my distro has switched to the new shiny system-resolved, thanks a lot.

ah, evil poettering makes such nice things that everyone wants to switch to them

Why devuan and alike even exist

the correct question is why morons are bitching on forums instead of switching to those wonderful distros

Is it because these people are so damn stupid or is it because they see flaws in the systemd design?

let me think, are marginal distros for idiots made by idiots or by brightest people on planet?

And no, I can't switch to a devuan or alike because my hardware vendor only ship ISO's with systemd inside.

then you should thank systemd, because otherwise your hardware vendor would ship you empty iso, idiot

And no I do not have time nor money to tinker with yocto.

if you don't have time and money, then all you can do is report bugs to your upstream (which is not systemd in case you can't understand such trivial things)

[pal666]

I don't think

thinking is hard. consider also don't posting shit

[pal666]

I feel that maybe SystemD is not designed for embedded systems.

i feel that maybe you should learn how it is spelled before accusing it of bugs of whole new distro?

[pal666]

sorry, can't help you with a distro

it is not like fedora needs your help