By Jake Edge
December 3, 2008
An effort to get the qmail mail transfer agent (MTA) into
Debian repositories has run aground due to various concerns, but the
overriding one seems to be a distaste for qmail itself. Distributions make
package availability decisions based on "taste" all the time, but they are
generally made strictly on technical grounds, which does not seem to be the
case here. While it
has its share of detractors, qmail is a relatively popular MTA—with
an excellent security track record—and one of the main impediments,
its license, has changed in the last year. Because of that, it makes it a
bit hard
to understand why qmail would be kept out of Debian.
More than six months ago, Gerrit Pape had uploaded qmail and related
packages to the ftp-master system, but they have yet to be added to the
official Debian archive. He recently outlined his efforts in a
post to debian-devel trying to see if he
could break a kind of standoff between him and the ftpmasters, who are the
folks that decide which
packages get moved into the official archives. More than two months after
his first upload of the packages, Pape got a reply from Joerg Jaspert outlining multiple
technical reasons why the packages were being opposed, but also containing
the following disheartening verdict:
Aside from these technical - and possibly fixable - problems, we (as in the
ftpteam) have discussed the issue, and we are all of the opinion that qmail
should die, and not receive support from Debian. As such we *STRONGLY*
ask you to reconsider uploading those packages.
After that, Pape addressed some, but not all, of the technical complaints
and uploaded updated packages along with a reply
to Jaspert's rejection on September 1. Since that time, there has been no
action on the packages
nor any further communication from the ftpteam, which is what led to the
debian-devel post. Responses there mostly backed the ftpmaster's
"decision"; qmail, it seems, is not very popular with many Debian developers.
Unfortunately, some of the complaints are based on old or faulty
information. There is a reasonably active upstream and, since Daniel
J. Bernstein (aka djb) released the code into the public domain, there is
no longer the need to patch qmail to get a sensible MTA. There are some
legitimate concerns, in particular the backscatter that gets created by the
default qmail configuration, but it is rather disingenuous to list security
as one of those problems.
While not as bulletproof as djb would have it,
qmail does have a long record of few security problems. In response to
claims that the Debian security team would have more work because of
qmail's inclusion, Moritz Muehlenhoff makes it
clear that the team won't block qmail. Florian Weimer puts it this way:
Like Moritz, I don't see issues with security support, provided that
the number of additional patches is rather small. (To my knowledge,
badly patched qmail with a SMTP AUTH bypass vulnerability was one of
the few MTAs which were actually exploited to send spam in recent
times.) I'm also not sure if upstream can be considered dead, and
arguments along that line are not very convincing because similar
criticism could be brought against our default MTA.
I can understand that people have strong feelings. I'm willing to
provide security support, but it's extremely unlikely that I'll run
qmail on production MTAs ever again. 8-/
In the end, it comes down to emotions, largely. People generally feel
strongly about qmail, either hating it or loving it, with few who know much
about it anywhere in between. Clearly the ftpteam has the responsibility
to reject packages on technical grounds, but are they the arbiters of taste
for Debian as well?
An earlier thread
about including qmail, from shortly after djb freed the code, showed a
fair amount of interest in qmail, along with some opposition. It is
unlikely that all Debian developers are happy with all of the packages
currently supported by the distribution, so singling qmail out seems rather
arbitrary. As Wouter Verhelst notes:
As long as qmail is free, packaged
properly, and integrates well with the rest of Debian, I don't see why
anyone should oppose its packaging.
Whether or not it's a good MTA, the fact is that it's a *popular* MTA.
That alone should be a good reason to package it.
Installing qmail has always been painful; it is a package that cries out
for distribution integration, which Pape is trying to provide. Whether it
gets into the official repositories or not, unofficial qmail packages do
exist. If the problems with qmail are largely packaging-related, it is
hard to see how they will get fixed by staying unofficial. But if the problems
are based on an emotional response to qmail itself—whether based in
technical concerns or not—it is hard to see how a developer can
overcome them.
(
Log in to post comments)