Is there something wrong with CUPS?

[rcrsn51]

There is a wide-spread belief in the Puppy community that CUPS is a fundamentally bad piece of software. Words like "fragile" and "unstable" are often used. This is completely untrue and here is an example.

Samsung makes a Unified Driver for installing its printers. If you run this package in Lupu, the CUPS web interface immediately crashes (although the CUPS daemon is still running happily in the background). Reinstalling CUPS won't fix the problem.

It turns out that the Samsung installer changes the permissions on some device nodes like /dev/null. This may make sense in a multi-user distro like Ubuntu, but not in Puppy. CUPS stops working because you have taken away its right to do so. CUPS isn't broken - it is behaving exactly like a Linux program should.

Once you restore the permissions to their correct values, CUPS starts working again. No other repair is required because none of the CUPS files were altered.

There is nothing wrong with CUPS. The problem is running CUPS in a single-user, all-root environment like Puppy.

[tater]

I've had zero problems with cups across my network printing to a windows xp shared printer. It's an HP Deskjet 1050 I recently bought at WalMart for $40. (great deal considering that it's an all in one printer).

It was quite simple to set up. I just accessed the cups interface in my web browser typing in "localhost:631" and added the printer using the address "smbc://192.168.0.7/HPDeskje". It works like a charm.

I did have to install the extra HP printer .pet for the working driver.

Just my 2 cents.

[rcrsn51]

It was quite simple to set up.

Exactly. People run into trouble with CUPS when they install third-party packages that aren't compatible with Puppy. It is generally believed that you can click on any DEB file from anywhere and safely install it into Puppy. If that installer modifies the ownership/permissions of Puppy, CUPS may recognize the changes and quit working.

[Bernie_by_the_Sea]

There is nothing wrong with CUPS. The problem is running CUPS in a single-user, all-root environment like Puppy.

Sounds to me like the problem is permissions, not single-user, all-root. A true single-user system needs no file permission system. Puppy should take the next step and remove the whole permissions business, a bad hangover from UNIX. This won't happen so for CUPS the solution is to dump the browser interface and provide a gui interface for CUPS written specifically for Puppy. I just searched for alternative interfaces and they seem to be either huge or abandoned.

In my very tiny experience with printers in Linux I've found "close" may be good enough. Brother offers a specific Linux driver for my printer but I don't bother with it since a foomatic driver for a different Brother printer already in CUPS works for everything I want to do. Beginners should know that they don't always need a driver specifically made for their printer. Others may work just as well.

[rcrsn51]

... so for CUPS the solution is to dump the browser interface and provide a gui interface for CUPS written specifically for Puppy.

It doesn't matter what kind of interface is used. If CUPS has the wrong permissions, it won't work. The easiest solution is to abandon all new CUPS versions and go back to CUPS 1.1. In that series, CUPS ran its print jobs as the root user, so any modifications to the ownership/permission structure were irrelevant.

But the whole point of the new CUPS is to NOT run as root, thereby protecting the user from exploits. Like it or not, CUPS is designed to work in the big, bad world where that kind of thing is important.