Well, I have a wishlist as well (of course all items should be configurable):
Save cronjob output to a file
Append that output to log files on the system, after applying regexp filters
Send E-Mail to a number configurable addresses
Have a list of regexps that the output is filtered through before being included in the e-mail
Allow the preceding operation to be done multiple times for a single mail, allowing to include different "verbosity levels" in a single message to have the admin decide which to read
Allow certain parts of messages to be tacced (printed with line order reversed)
Modify message subject if certain conditions are met (for including tags like [OK], [ALERT] etc)
Allow arbitrary texts (preferably using macro expansion) in between message parts
This is only what immediately comes to my mind. I will keep this wishlist updated, and have a Wiki Page.
Looks like there is no passwd-compatible crypt(1) for the command line. htpasswd, unfortunately, uses a different algorithm.
This short perl script might be a replacement:
#!/usr/bin/perl -w
use strict;
while(<>) {
my $seed = `apg -a 1 -m 8`;
chomp;
print crypt(“$_”, “\\$1\\$$seed”). “\\n”;
}
Or do we have something better already in the distribution?
Update: looks like mkpasswd (from the whois package, whatever makes it belong in there) does the job quite nicely, but the script shown above takes care of automatic salt creation as well. Any ideas how to do that more elegantly, without requireing apg?
A lot of recent systems I have to work with have Tigon3 ethernet interfaces, which behave strangly when used under Linux in settings that are non-trivial, networking-wise.
While browsing Debian Planet, I happened to have the desire to comment on one article innutmeg's Blog. What I found was a blog I couldn't comment on, but one whose format I know pretty well: It looks suspiciously like a Debian changelog. nutmeg then told me that he uses a script from Romain Francoise which indeed builds an RSS feed from a Debian changelog to feed his blog to Debian Planet.
This is something I love Debian for. Use existing tools for new challenges. Well done.
In a fit of carelessness, I decided to pull one of my test notebooks to current unstable, including the Xfree 86 => X.org migration. To my surprise, the migration was not only absolutely painless, but the new X server worked on the first try. This is a new experience with X11 for me.
Splendid work, Guys! Keep it up!
After the weekend (where I need my main notebook to be operational), I am going to try moving my main work vehicle to X.org.
The update of my main work notebook was as painless as expected after the test with the unimportant box. aptitude, clean up dependencies, and you're back in business. On the first starts (from the command line), fonts were broken (too large), but after rebooting and logging in from kdm, the fonts are ok again.
Thanks to Jörg Hoh, who submitted a number of cleanup patches against adduser, the package has changed quite bit. This is the reason for the 3.68 upload going into experimental. Please go ahead, test, report bugs, submit patches. Thanks!
exim4, sarge's default MTA, uses gnutls for the obvious license reasons. However, gnutls does seem to have issues of interoperability, which have manifested themselves in a list of bugs, most prominently being #297174, which we are at a loss to debug.
Neither Andreas nor me have the knowhow to debug gnutls, and Upstream uses openssl - the gnutls patch was contributed and the author of the original patch doesn't seem to be around any more.
Looks like I finally got rid of my last woody system on 2005-07-26, so two months after sarge's release it is now time to discontinue supply of woody backports on zg.debian.zugschlus.de. Cleaning up the actual package pool will take a whileis finished and reduced the pool from 803MB to a mere 35MB, and it looks like I'm gonna miss again the opportunity of converting my local package-pool-helper scripts to debpool.
I have again worked on the automatic package build, test and publishing process for clamav-data, which runs twice an hour on a volatile.debian.net host. Details of the process are described in VUA 2-2 on debian-volatile-announce
Surprise: Packages depending on debconf should have an alternative for debconf-2.0, so Joey says in a Mail to -devel. I must have missed something.
I have never heard of that requirement before, and I surely am not the only one. I won't be able to update my packages quite soon, but I'll try to get things sorted out in August.
I'll have to touch:
adduser - doesn't use debhelper, manual change committed to VN
clamav-data
clamav-getfiles
snoopy - fixed in 1.3-10, now properly using debhelper's macros
Packages that I co-maintain which are affected as well (so I need to keep an eye on):
aide
exim4 is not affected, Andreas has thought of that. Good.
The softmodems that can be found in today's notebooks have been a mess from the very beginning. This especially holds when such a modem is to be used on Linux. Today, we have gazillions of different (and incompatible!) methods of using these gadgets, but none of them works with all softmodem flavours out there. This article describes what I have tried to bring my hp compaq nc8000 online, and how I finally succeeded.
The Debian BTS has recently learned to track versions. Bugs can be closed and "found" for certain package versions, so that it is easier now to determine whether a bug is present in a distribution or not. However, Colin still needs to write a "metric shitload of documentation", and I still fall way short of understanding how to use the feature.
Packages that need to change their init script execution order during an update have a problem: sysv-rc's and file-rc's abstraction layer (which is access through update-rc.d and invoke-rc.d) doesn't offer read access. Hence, it is impossible to see whether the execution order has been locally modified without interfering with the internal mechanisms of the appropriate package.
