GNU Shepherd 0.10 Released For Guile-Written Init/Service Manager

[phoronix]

Phoronix: GNU Shepherd 0.10 Released For Guile-Written Init/Service Manager

GNU Shepherd 0.10 is out today as the latest release for this GNU-backed service manager and init system that is written in Guile...

https://www.phoronix.com/news/GNU-Shepherd-0.10

[Ericg]

Honestly that service file syntax alone makes it a total non-starter for me and probably many others. Compared to ini/TOML style of systemd… I literally can’t make heads or tails of that syntax and where different things begin or end

[flower]

Honestly that service file syntax alone makes it a total non-starter for me and probably many others. Compared to ini/TOML style of systemd… I literally can’t make heads or tails of that syntax and where different things begin or end

the syntax really isn't great but tbh it's not a real problem.
what's interesting about it is that it is a real programming language. this could (sometimes) be helpful.

but well... i am a systemd fan anyway and learning multiple init systems is just painful. too many details to know

[scottishduck]

As a GNU project we can expect a 1.0 release sometime this century.

[F.Ultra]

the syntax really isn't great but tbh it's not a real problem.
what's interesting about it is that it is a real programming language. this could (sometimes) be helpful.

but well... i am a systemd fan anyway and learning multiple init systems is just painful. too many details to know

Turing complete configuration languages are a massive security risk in the making. E.g that is why sendmail experienced decades of horrible exploits.

[jacob]

Honestly that service file syntax alone makes it a total non-starter for me and probably many others. Compared to ini/TOML style of systemd… I literally can’t make heads or tails of that syntax and where different things begin or end

The syntax is not the worst part, at least it's consistently parseable and machine-readable. The worst is that apparently it still uses pidfiles.

[dragorth]

As a GNU project we can expect a 1.0 release sometime this century.

You are a very optimistic fellow.

[fwyzard]

but well... i am a systemd fan anyway and learning multiple init systems is just painful. too many details to know

Then we strongly disagree :-)

I'm glad to see (m)any alternatives being worked on. At the very least it will bring diversity and improvements to the whole ecosystem.

[WiR3D]

Honestly that service file syntax alone makes it a total non-starter for me and probably many others. Compared to ini/TOML style of systemd… I literally can’t make heads or tails of that syntax and where different things begin or end

Fully agree, just the thought of having to code review this compared to systemd units gives me chills.

[Delgarde]

what's interesting about it is that it is a real programming language. this could (sometimes) be helpful.

It's a mixed blessing. Using a real programming language gives you power when you need it — but serious downsides too. Security risks, certainly... embedding a Turing-complete parser and runtime just to read config files massively increases attack surface. But it also makes it all but impossible for tools to manipulate those files, so they pretty much have to be hand-written.