>From: Matija Nalis <mnalis-deb...@voyager.hr> > So, original question was "what broke"? You still didn't answer that. > Package being upgraded is not breaking. If you lost some functionality, that > might have been breakage (from your POV), and that is what I (and Gerrit, I > believe) are interested in details of.
Well, one can no longer run `build-ucspi-tcp` (from ucspi-tcp-src) and install the package and expect it to remain installed. I rate that as breakage. As the packages are fairly similar one can state the breakage isn't a problem, but I must still rate it as breakage. It would have been perfectly okay for Gerrit Pape's ucspi-tcp package to have a lower version number and let people choose to install that, rather than inflating it and FORCING its installation. > For example, if main/ucspi-tcp lacks some feature provided by > non-free/ucspi-tcp-src, and you upgraded by mistake (despite being warned) > and your system stopped working correctly because of that; and *that* > missing feature is what is actually bugging you (ie. upgrade removed some > feature which is usefull), the correct way to go would be file a feature > request for that feature -- and certainly filing not a major bug requesting > the removal of the package! Well, two major items come to mind: Item #1: 10 years of packaging bugfixes. While some of those may no longer be relevant due to changes in Debian packaging procedures, many are still relevant. Figure that every single bug report ever filed against ucspi-tcp-src needs to be cloned and applied to Gerrit Pape's ucspi-tcp package. Item #2: Gerrit Pape's ucspi-tcp lacks the IPv6 patch. While there is a ucspi-tcp-src package, this means Gerrit Pape's ucspi-tcp package is in no way equivalent to what had already been done with packages from ucspi-tcp-src. (BTW, if there was concern that that patch breaks some systems and thus should be in the default package, I think there is ample evidence to suggest otherwise given ucspi-tcp-src) > That may be. I see you have made at least one such proposal. My proposal > would instead be to abandon obsolete packages as there is no need for them > anymore (*-src packages are extremely ugly kludge used only when legal > issues prevent debian binary distribution) to prevent useless duplication of > efforts; maybe accompanied with feature-requests on ucspi-tcp to add some > missing patches/features if they were present in other package but are not > present in currently official one. I must return that suggestion right back to Debian (Gerrit Pape specifically). Why was "ucspi-tcp" created from scratch and given an inflated version number, instead of working to get ucspi-tcp-src into main? > Yes. Which does not change the fact that ucspi-tcp-src is *still* in > non-free, and as such is still *NOT* part of Debian (I've quoted you the > Debian Social Contract which you seem to be unaware of; even if it is the > most fundamental of all Debian documents). > > Yeah, the ucspi-tcp-src maintainer could've made an effort years ago to move > it to main when Dan announced going PD (assuming none of the additional > patches have license issues with that). He didn't. He still can try, > although it is much more complited, and very much more pointless now when > somebody else has already made the effort. Okay, looks like the license change was 12/28/2007 while Debian lenny came out 02/14/2009. I'm actually pretty impressed with Gerrit Pape being able to produce a package so quickly. That is actually pretty fast by the traditional Debian timeframe. > Anyway I hope you will realize that I'm "not out there to get you" and that > you'll be able to see what the question being asked is (and maybe even > understand why it is being asked). If you do, I'd be happy to try to help > you (by porting patches or validating their licences or whatever is needed). > If you however would rather makes this some crusade or whatever, I'm not > interested in that. I never made such an accusation. I guess I'm most annoyed by the /way/ this was done, rather than that it was done at all. I've seen quite a few mentions of the problems with using epochs, plus this overrides people who have been using ucspi-tcp-src for a long time. I'm quite concerned people may not have realized anything occured, since they might skip testing the mailserver expecting it to have been left alone... I'm also greatly concerned about the loss of existing bugfixes in the ucspi-tcp-src package. -- (\___(\___(\______ --=> 8-) EHM <=-- ______/)___/)___/) \BS ( | e...@gremlin.m5p.com PGP F6B23DE0 | ) / \_CS\ | _____ -O #include <stddisclaimer.h> O- _____ | / _/ 2477\___\_|_/DC21 03A0 5D61 985B <-PGP-> F2BE 6526 ABD2 F6B2\_|_/___/3DE0 -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
