Technically, the DEC DECwriter series were dot-matrix printers, not line printers. They differed from their Teletype predecessors only in print-head technology, but both printed a single character at a time. Daisywheel printers were similar.
Right....
Line printers are distinguished not by the width of the paper but by the printer having enough print heads to print an entire line of output at a time. That speed advantage made them the preferred output device for many-page program listings, as opposed to a teleprinter terminals which were more suitable for interactive computing.
There were originally two styles, the drum printers which DEC sold(e.g. LP20) and the chain printers that IBM offered (e.g. 1401). The drum had all the characters in each of the 132 columns (the upper case only printers were faster because the alphabet was on the drum in two places). The IBM ones has slugs on a rapidly spinning chain that was horizontal (and parallel) to the line being printed. The chain was easily replaceable by the operator - which was one of the duties we would have. When a user queued a printer a set of symbols (i.e. the chain of the needed output characters) was specified and the system queued it until the printer had been properly provisioned. For instance, CMU printed checks with a special chain and film ink, so once a night the operator would configure the printer, and tell the queue to print them). Some chains were faster than others, the standard one had N copies of each character.
In common to both schemes is that each both styles had 132 hammers and when the proper character was in the position needed, the hammer fired to make an impression the ribbon on the paper, which was caused the noise people associated with computer printers. The high-end IBM 1401 had a hydraulic cover that came down over it and was controlled by the channel processor (it would auto-open when it needed to be serviced - like a new box of paper). But even with the cover down it still loud.
The original DEC ones were OEM'ed from Centronix and were noted for always being a little random on the hammer timing and thus the print on the paper often looked like the characters bounced on the line. I remember the ones we had on the PDP-10s were awful and the issue with BLISS is that the dot operator is extremely important to your code and the dots were sometimes notoriously missing.
Cabling could be difficult too. They were parallel devices and were supposed to have shorter cables (i.e. in the machine room). IBM used its own interface, but by the mid-1970s the Centronix printers were pretty standard on the mini-computers and their parallel interface became the standard (in fact the IBM PC supplied a Centronix parallel interface).
There were dot-matrix line printers of the late 1970s made by Printronix, which is apparently still around.
IIRC, in 1979 the Printronix cost about $5K, plus another $300-$500 for an Arduino sized parallel to serial converter that they sold so the printer could be remote on a 9600 baud serial line. Until the cheaper lasers came about, these were often the standard printers that UNIX sites had [I was told once that the original Lion's book was printed on one]. They were about ½ the cost of the DEC printers and since it will all pins, did not have the bounce issue the drums had. When the UNIX boxes started to appear at CMU we used them, and IIRC @ UC Berkeley, we had a number of them around Cory Hall also.