[Dapper] Who decided to hose wpasupplicant?

[Stephen R Laniel]

I was just trying to figure out why my network connection
was uncommonly slow, so I did an iwconfig and noticed that I
was connecting to a neighbor's unencrypted access point
rather than to my own WPA-encrypted one. This would explain
the slowness. So I did a 'ps aux' and noticed that
wpasupplicant was not in fact running. This was odd. Next
step was to run '/etc/init.d/wpasupplicant start', which ...
doesn't exist anymore! In its place is

-rwxr-xr-x 1 root root 1782 2005-10-30 07:43 /etc/init.d/wpasupplicant.dpkg-bak

Well that's odd. Where did it go? Look in the
README.Debian file ... ah ha! Aparently all the
functionality has been moved from /etc/wpa_supplicant.conf
into /etc/network/interfaces. Except not really: no one
bothered to migrate my existing WPA configuration to the
new format. And it's not a simple matter of copying the
entries from the old wpa configuration into
/etc/network/interfaces.

So much about this migration is wrong that I don't know
quite where to begin. Coming as it does on the heels of an
X.org migration that has turned X into a television test
pattern, I am more than a little vexed. Let's try a few
points, for starters:

1) No upgrade should, under any circumstances, break a
   user's network connection without clearly advertising the
   fact that it may do so.

2) If it's going to break fundamental parts of his system's
   operation, it should provide an upgrade path to make this
   workable. In this case the upgrade is fairly minor: add a
   wpa-conf line into the relevant stanza within
   /etc/network/interfaces. Which leads to point 3.

3) Wherever practicable, the upgrade should include a script
   that handles this migration for the user. Even better is
   if this migration is transparent, perhaps leaving behind
   a little note in (in this case) wpa_supplicant.conf,
   stating, "Settings have moved to the appropriate stanza
   in /etc/network/interfaces. Sorry for the inconvenience!"

4) Yes, it's development software, so RTFM and
   don't-expect-cleanliness and all of that. But it seems to
   me that the development of Ubuntu and Debian software has
   a nasty record of letting bad mistakes out into the
   wild -- missing semicolons in some cases, and unspoken
   migrations like this one in other cases. I want to help
   the cause of improving Ubuntu, but cases like this very
   much encourage me to stick to stable releases and not
   work on new ones. I *expect* instabilities in new Ubuntu
   releases, but I also expect due diligence before the
   public gets hold of alpha releases. Alpha releases are
   not a license for carelessness.

Am I being unreasonable here? If Ubuntu expects users to
help with, say, Dapper, it has to treat them better than
this. A reasonable response is that power users should
expect some carelessness. I would retort that if
Ubuntu wants to get more people helping with development
releases -- not just power users, but ordinary users too --
it has to be much more careful in what it lets out. The more
ordinary users we have helping with development, the better,
right? Dapper is in no position to let ordinary users into
the process, and that's a shame. It needs to do better.

-- 
Stephen R. Laniel
steve at laniels.org
Cell: +(617) 308-5571
http://laniels.org/
PGP key: http://laniels.org/slaniel.key
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 191 bytes
Desc: Digital signature
URL: <https://lists.ubuntu.com/archives/ubuntu-users/attachments/20060325/8b7b59b7/attachment.sig>