Debian drops redis non-systemd support

[golinux]

Another step towards making systemd mandatory and eliminating choice. The decision is Debian's. Upstream maintainer will continue to support other inits. It does not please me to say, we told you this would happen. There has been discussion on the dng mail list for those who are interested.

[n_hologram]

https://lists.dyne.org/lurker/message/2 ... 35.en.html

antirez just replied me on the fly because he is traveling and
confirms that official Redis position is that of supporting all init
systems. He also says that: the most "obvious" way to start redis is
from init.d scripts. distros may do whatever they want, but the
support to start redis from a sysvinit script will never be removed.

I would like to know here if anyone knows in detail the "reasons"
Debian is removing the support for sysvinit scripts in the redis
package. Is there a limitation of sorts, or a problem generated by
having the init scripts left in the package, or is it just vandalism?
is there a discussion or bug open?

Since even the program's creator understands that forcing an init system is ludicrous, it looks like, at the least, Debian[-based] users will still be able to access the scripts. Debian's approach just seems like overkill:
https://lists.dyne.org/lurker/message/2 ... f3.en.html

> https://github.com/lamby/pkg-redis/comm ... 2d2bc07e5c

This diff clearly shows that redis-sentinel example scripts provided
by upstream redis developers for compatibility with sysvinit are being
*deleted* from the package, de facto removing even the documentation
on how to start/stop from init.d despite it being just an example.

it is even worse than I initially thought.

[Bulkley]

Is redis required for non-systemd? My Stretch with systemd does not use it. These are the only references to redis.

Code: Select all

$ locate redis
/usr/lib/libreoffice/share/gallery/www-graf/gredisk.gif
/usr/share/doc/debian/FAQ/ch-redistrib.en.html
/usr/share/doc/debian/FAQ/ch-redistrib.html
/usr/share/nmap/nselib/redis.lua
/usr/share/nmap/scripts/redis-brute.nse
/usr/share/nmap/scripts/redis-info.nse


I'm just curious to know if there is a way around redis. Is the non-systemd world dependent on it?

[golinux]

Is redis required for non-systemd?

I have gathered that redis is required to provide certain functions for complex programs. But I am certainly not an expert and had never heard of it before.

I'm just curious to know if there is a way around redis. Is the non-systemd world dependent on it?

I suggest you read the thread posted above. It appears to be a crucial component for many admins on the dng list. It also seems to be an app that desktop users would not engage directly or care about.

I wonder how those using Debian with sysvinit would be able to use redis with that support removed. Maybe there's a shim. If not, then this could be the first salvo at eliminating sysvinit entirely and forcing Debian users to systemd. The hardcore admins over at devuan have their knickers in a twist over it.

Thankfully the maintainer is adamant about continuing to support sysvinit so there is a way around the obstacle that Debian has created.

[debiman]

https://lists.dyne.org/lurker/message/20171019.104326.9b567835.en.html

antirez just replied me on the fly because he is traveling and
confirms that official Redis position is that of supporting all init
systems. He also says that: the most "obvious" way to start redis is
from init.d scripts. distros may do whatever they want, but the
support to start redis from a sysvinit script will never be removed.

Since even the program's creator understands that forcing an init system is ludicrous...

forcing?
ludicrous?
did you get this from the post quoted above?
because i don't see anything so judgemental and loaded in it.
you are putting words into someone else's mouth (assuming antirez is the "program's creator"), filling them with your very own emotional subtext.

[n_hologram]

>"emotional subtext"
Please forgive me. It was difficult to write though the tears. Now that I'm no longer hyperventilating, I'll attempt to clarify what I meant.

My point is that, in the emails, it was strongly implied that someone who appears to be a developer, who, despite Debian removing the sysv scripts, retains sysvinit compatibility whether or not he or she believes it ludicrous to restrict his or her program to a singular init, takes a position that includes more than ons init system -- although the mystery of whether or not requiring a specific init system is a kind of forcing or imposing remains to be verified.

I agree with this decision and speak for only myself in the assertion that systemd imposes many undesirable features, and restricting the entire Linux ecosystem to using it, which I see as a natural consequence of removing support for alternative inits in order to use programs, is -- again, in my personal opinion only -- ludicrous.

In any case, it was an inference: "He also says that: the most 'obvious' way to start redis is
from init.d scripts." I interpreted the 'obvious' part to suggest that he wouldn't be thrilled to remove them, but maybe he doesn't find it ludicrous to remove them and thereby require, well, systemd.

[Soapm]

Years ago, I remember insisting I would never own a front wheel drive car because it just made more sense to push something as opposed to pull...

Here we are years later and I wouldn't give you much for a rear wheel drive car in the snow, FWD sure eliminated a lot of pushing...

Not sure why, but I thought I would throw that in there...

[debiman]

In any case, it was an inference

that's what i meant.
good we agree.
please stop using neutral comments like the one you quoted to push your agenda.
comments made by people whose opinions you know nothing about - you might be misinterpreting them, and next thing you know another false internet rumour is born.

[n_hologram]

Asserting that I have an agenda, rather than making an inference after I admitted that's what I did, is ludicrous. Since you made your point, and since I even validated it with clarifications, I ask that you take this conversation in a constructive direction, rather than another critical commentary, and I intend to report any posts which focus on a trivia rather than the post's topic -- redis.

To attempt rectifying the main topic, I think that supporting other init systems, like sysv, is a constructive decision, and I applaud the developers in maintaining this decision; although it might serve as a profound annoyance to build this from source in order to maintain, for instance, projects like Devuan, I still believe it better than the alternative, an approach which it would appear Debian appeared to endorse at the time of the topic's creation, of omitting that choice completely.