2014/1/10 Didier 'OdyX' Raboud <[email protected]>: > Control: tags -1 +moreinfo > > Hi all, > > While hunting down old src:cups bugs, I stumbled upon #578079, which I > think is not-a-bug, see below. > > Le jeudi, 28 avril 2011, 15.56:06 Martin-Éric Racine a écrit : >> I'm starting to suspect that upstream made some backward-incompatible >> changes to the way queues are handled, because we get a similar >> problem with CUPS-PDF, but only whenever the package gets upgraded; if >> it's installed from scratch or purged then re-installed, it works as >> it should. > > From the code: > > $ git grep -B1 'Upgrade Required' cups/ > cups/http-support.c- case HTTP_STATUS_UPGRADE_REQUIRED : > cups/http-support.c: s = _("Upgrade Required"); > $ git grep 'HTTP_STATUS_UPGRADE_REQUIRED ' cups/http.h > cups/http.h: HTTP_STATUS_UPGRADE_REQUIRED = 426, > /* Upgrade to SSL/TLS required */ > > … which means that the 'Upgrade Required' error is not a "Please upgrade > your cups client to a more recent version because we broke backwards- > compatibility"-error, but a "Hey client, please talk to me over SSL/TLS > because I require it"-error. This can be enforced using the "Encryption" > cups.conf statement: > > https://localhost:631/help/ref-cupsd-conf.html#Encryption > > I therefore propose to close this bug as it doesn't appear (to me) to be > a bug but mostly a miscomprehension of the error message.
Way back when people reported installation errors for CUPS-PDF, the general concensus was that using the -E option to enable encryption with 'lpadmin' was a source of problem, because the SSL certificates needed by CUPS might not have been yet unpacked and configured on a fresh install, so the option was removed from CUPS-PDF's maintainer scripts. Yet, according to the above, connections do require it. I'd really appreciate more precise info for when it's required and when it's not. Once I've received this, I can upgrade CUPS-PDF's maintainer scripts to match. Still, TBH, the error message itself is a textbook example of obfuscated code that provides misleading feedback to the user. *sigh* Kudos to Didier for taking the time to figure this one out for everyone else. :) Martin-Éric -- To UNSUBSCRIBE, email to [email protected] with a subject of "unsubscribe". Trouble? Contact [email protected] Archive: http://lists.debian.org/capzxpqe5ednbgcfwocc4pmp8tn5tievmhtk8j+rxyrdzacm...@mail.gmail.com
