OpenPrinting Now Developing Upstream CUPS, Apple Bows Out

[phoronix]

Phoronix: OpenPrinting Now Developing Upstream CUPS, Apple Bows Out

Back in 2007 Apple effectively acquired the open-source CUPS project and in 2017 then decided to no longer develop CUPS under the GPL but instead the Apache 2.0 license for this widely-used Unix/macOS/Linux print server. But then at the end of 2019 the CUPS lead developer left Apple and following that public development of CUPS seemingly halted. Fortunately, now there is a happy next chapter to the CUPS printing story...

https://www.phoronix.com/scan.php?page=news_item&px=Apple-No-More-CUPS

[Chugworth]

Over time Apple will probably phase out CUPS from their OS and eventually dump it completely. It's a system that really seems quite bloated for workstations that just need to occasionally print a document. I mean right now on my laptop I can browse to http://127.0.0.1:631/ and see that it's constantly running a web server on my machine for managing print jobs. Do things really need to be this complex just to pass a document to a printer?

[Ipkh]

Sure, the answer was to make the printer run the server components.

[Sonadow]

Wasn't moving away from CUPS supposed to be the whole point?

I don't understand why PAPPL still requires CUPS.

[wizard69]

Over time Apple will probably phase out CUPS from their OS and eventually dump it completely. It's a system that really seems quite bloated for workstations that just need to occasionally print a document. I mean right now on my laptop I can browse to http://127.0.0.1:631/ and see that it's constantly running a web server on my machine for managing print jobs. Do things really need to be this complex just to pass a document to a printer?

I have to agree that CUPS is bloated. The problem for Linux is that little effort has been put into a modern solution. For Apple I think you are right CUPS no longer serves their needs. Maybe we will see this acknowledged at WWDC this year with a replacement.

the funny thing here is that I don’t even have my printer hooked up. The more I think about it CUPS is probably a waste of energy.

[oiaohm]

Wasn't moving away from CUPS supposed to be the whole point?

I don't understand why PAPPL still requires CUPS.

There are still the more complex network and multi user cases where you really do need authentication proxy that kind of where CUPS Sharing Server goes. Its kind of like why when I have http servers do at times I need https://varnish-cache.org/ reverse proxy. Like one of the reasons for having a reverse proxy is to at timed add filtering to reduce the damage a denial of service attack can do.

You also have to hand legacy applications this is where CUPS local server per user fits in.

PAPPL does solve other problems. PAPPL removed the problem of one printer driver basically screwing all printers up in one hit. But we cannot be sure that PAPPL will be flawless and are safe to directly expose to the network always. So there is still a need for a printer proxy for likes of reverse proxy on a web server not all uses may end up running it but the need is still there for cases where PAPPL drivers are either flawed or there need to be some other constraint on access.

Yes it really simple to miss running servers does not mean you don't need proxies on those servers.

[oiaohm]

the funny thing here is that I don’t even have my printer hooked up. The more I think about it CUPS is probably a waste of energy.

Without the structure changes with CUPS 3.0 where the local can start on demard with a lot of applications on linux they need to talk to CUPS to be told there is no printers or they get really upset. So having no printers does not mean you don't need something to tell applications that there is no printers. The I have had great doh moment a few times when attempting to make ultra compact Linux installs of strip out cups and something I need to work no longer does because it was needing the no printers answer.

[c117152]

Over time Apple will probably phase out CUPS from their OS and eventually dump it completely. It's a system that really seems quite bloated for workstations that just need to occasionally print a document. I mean right now on my laptop I can browse to http://127.0.0.1:631/ and see that it's constantly running a web server on my machine for managing print jobs. Do things really need to be this complex just to pass a document to a printer?

I had the same concern a few years ago so I disabled BrowseWebIF in /etc/cupsd.conf. It barely scratched a few kb off the RAM overhead. My conclusion was that since cups needs to support network printing anyhow, the web interface is practically free.

Btw, cupsd is mostly suspended until the system wakes it up when something knocks on the relevant ports so it's actually far more efficient than you might think.

[wizard69]

Without the structure changes with CUPS 3.0 where the local can start on demard with a lot of applications on linux they need to talk to CUPS to be told there is no printers or they get really upset. So having no printers does not mean you don't need something to tell applications that there is no printers. The I have had great doh moment a few times when attempting to make ultra compact Linux installs of strip out cups and something I need to work no longer does because it was needing the no printers answer.

Never considered that but in a way that is an applications bug. Maybe what we need is a very light weight app/damon that tells apps “ no printer installed “.

[lumks]

I mean right now on my laptop I can browse to http://127.0.0.1:631/ and see that it's constantly running a web server on my machine for managing print jobs.

In that case, why do you have it running all the time? CUPS is able to start when you're about to use it via systemd socket activation
See: https://wiki.archlinux.org/title/CUPS#Socket_activation