In message <slrnd5j8l3.ina.martijnb@atlas.ipv6.stack.nl>, Martijn van Buul writ
es:
>It occurred to me that Steven M. Bellovin wrote in gmane.os.netbsd.general:
>>>So, yes, lpa is very tied to the parrallel port device, and you really
>>>shouldn't use it to begin with, if you can avoid doing so.  You really don't
>>>want a polled driver if you can afford an interrupt driven one.
>>>
>> My experience is just the opposite -- I found that polling drivers 
>> worked much better.
>
>I'll admit immediately that I haven't done a *lot* of testing, but I noticed
>that my trusty old LaserJet wasn't printing any faster while using lpa
>when compared to lpt. That's probably because my laserjet isn't setting
>any landmark records.
>
>However, what I *did* notice was that the lpa driver was taking a
>generous amount of resources in comparison to the lpt one.
>
Right.  And on my system -- this was a few years ago, with a Laserjet 
5M on a 800 Mhz i386 -- it didn't work at all with interrupts, I 
assumed because of the latency in fielding interrupts.  Nor am I the 
only one who had troubles; for an old summary, see

http://mail-index.netbsd.org/port-i386/1996/06/16/0007.html

In theory, interrupt-driven may be better.  "The difference between 
theory and practice is that in theory, there is no difference, but in 
practice there is one."  

I don't claim to be an expert on the issue.  I'm just trying to note 
that this particular question is more complex than it appears.

		--Prof. Steven M. Bellovin, http://www.cs.columbia.edu/~smb