[cups.bugs] [LOW] STR #2824: Misleading error reporting.

Kim Bruning kim at bruning.xs4all.nl
Mon May 19 19:51:52 PDT 2008


On Mon, May 19, 2008 at 05:06:18PM -0700, Michael Sweet wrote:
> 
> [STR Closed w/o Resolution]
> 
> Please contact your Linux distributor for this issue.

In essence, I am my linux distributor. 

> 
> Unfortunately, we can't document or work around every distribution,
> driver, or printer quirk. In particular, we can't tell you to go looking
> for a printer driver when we don't know how to do that for your OS of
> choice!

Sure, no problem, but on the other hand, if everything in logs and UI
blithely asserts  that CUPS can , will, and is successfully communicating
with the printer (but in fact there is something wrong), it's going to be
rather tricky to deduce that the problem is that there's actually a
printer driver missing.

I don't expect anyone to document or work around every distribution,
driver or printer quirk, that would be unreasonable. What I would like to
see is that when there is something wrong, the software at least breaks
cleanly. 

In this case, it would be possible to get an id-string back from
the printer (on linux, lsusb already does this, for instance) and
compare it to an entry in the driver file.  If the two don't match, you
can warn that something is not quite right.  "Warning: Printer ID doesn't
match ID in driver file.  Possibly you have the wrong driver?" 

If multiple devices somehow can't be told apart, then that could also be
warned about. "Note: FooCo unwisely doesn't let us tell their printers
apart. This could be a Barmatic 1234 or a Bazmatic 5678, but this driver
only works for the Barmatic 1234. "

Finally, in combination with the warning, it is not outside the realms
of possibility explain where to start looking for a solution, at least
to some degree.  "No other suitable drivers could be found at this time.
You may need to install additional drivers. Please refer to the
documentation provided by your linux distributor, or contact them for
help".

Of course, that might be asking too much. A simple crash and "Fatal
Error: USB ID 04c8:0812 does not match FooCo-Barmatic_1234" in the error
log would still be vastly preferable to the situation I found myself in.
Or is that still too much to ask, for some reason I don't quite understand?

> "man" and the on-line help at "http://localhost:631" (which is included at
> the bottom of every man page and in the README file for CUPS) is the sum
> total of all CUPS documentation we maintain.  

That's fine too. In the html interface, would it be hard to make
error messages in the following form? :

'<strong>Usb backend failure.</strong> Check your cables, and don't forget
to click <em>start</em> to bring the printer back online. <small>For more
information, see our <a href="error_help.xhtml#usb_failure">error
documentation</a>, or check the <a href="http://www.cups.org">cups
forums</a> or <a href="http://www.linuxprinting.org>linuxprinting
forums</a> </small>'

> You can also post questions
> on the CUPS forums here on cups.org, on each Linux distro's support
> forums, and on linuxprinting.org.

I certainly would, but in the two cases mentioned, there were no clear
error messages to post. (In one case no error was reported while there was
clearly an issue, In the other case a misleading error was being reported
where no error existed).  That's what I've just reported. 

With one medium and a small number of minor tweaks to error checking and
reporting, the problems I had would change from potential showstoppers
taking hours to solve, to something that can be handled in minutes on-site
(with an apt-get or emerge or rpm or etc in one case, or a simple mouse
click in the other). Fair enough?

sincerely,
	Kim bruning


-- 
[Non-pgp mail clients may show pgp-signature as attachment]
gpg (www.gnupg.org) Fingerprint for key  FEF9DD72
5ED6 E215 73EE AD84 E03A  01C5 94AC 7B0E FEF9 DD72





More information about the cups mailing list