2007-04-08: Back from vacation
I was badly overdue for a vacation, having not gotten away in four months,
so for the last week I was wandering about Vancouver Island and visiting
friends. I love British Columbia; it reminds me of home. The Pacific
Northwest climate always feels more natural to me than the Northern
California climate.
This year, we got out to the west coast of the island. Waves crashing on
rocks and sand is soul-calming, but even more impressive was touring the
temperate rainforest and bog. If you're ever on the west coast of
Vancouver Island near Tofino, go walk the rainforest trails. It's an
amazing experience.
I also got a chance to wander Protection Island, a small island in the
Nanaimo harbor (and thus in the Strait of Georgia) large enough to have a
gravel road network that the residents navigate largely with golf carts.
It's a ten-minute ferry ride from downtown Nanaimo and the ferry runs every
hour. I'd love to live there. The atmosphere is wonderful; artistic and
small-neighborhood than rich mansions, away from the bustle of Nanaimo, but
still close enough to have easy access to all the benefits of a city.
We had a wonderful time, with plenty of opportunities for long conversation
or quietly enjoying each other's company.
Stanford's relatively generous with vacation for a US employer, but I still
wish I could work for an employer with European benefits and get away a
little more often. Although to be honest the limiting factor isn't so much
the amount of vacation I get as it is the amount of work that builds up.
Starting tomorrow, I dig myself back out again.
Debian etch released while I was out of touch, which is a great surprise to
come back to. That will mean a ton of work, of course, but I like the look
of etch and we're already running so much etch software via backports.org
that it shouldn't be too bad of an upgrade. This also means that I can
start working on all the package uploads that were put off pending the etch
release (although to give people a chance to set up the archive software,
I'll probably not start until towards the end of the week).
Congratulations to Sam Hocevar on his election as Debian Project Leader.
After long consideration, he was my top-ranked choice; hopefully the good
feeling I had about him will be correct. It's very difficult for the DPL
to change anything substantially, so I'm not expecting too much, but
perhaps I'll be pleasantly surprised.
2007-04-10: kstart 3.10
The first of the post-etch flood of updates to my packages. This release
fixes a build issue with later versions of MIT Kerberos on systems that
don't have a /usr/include/com_err.h link. It also contains the first
attempt at overhauling my Kerberos configure logic to use the new AS_IF
macro in Autoconf 2.61 so that Autoconf can do better dependency analysis
inside conditionals. Here's hoping I didn't accidentally break anything.
You can get the latest version from the
kstart distribution
page.
2007-04-10: pam-krb5 3.5
The second package of the post-etch flood. This release includes a couple
of fixes that had accumulated (KCM caches, use of freed memory when doing
debug logging), and does better Kerberos status code translation. This is
the first 3.0 release uploaded to Debian, although the PKINIT code isn't
active yet since the Debian MIT Kerberos doesn't have PKINIT yet.
You can get the latest version from the
pam-krb5 distribution
page.
2007-04-10: lintian 1.23.29
Post-etch flood, part three. I finally got around to fixing the debconf
dependency analysis code and adding the checks for error templates, and
then I fixed the --color implementation so that it really worked the way
that I hashed out in the bug report.
145 lines of changes in this version. I need to upload new versions more
often (although in this case it's because we were waiting on the etch
freeze).
lintian is still over 100 bugs (although as always nearly all of them are
wishlist). There are a few more easy ones that I may tackle in a bit, but
I need to get caught up with the rest of my post-etch uploads first.
2007-04-11: pam-afs-session 1.3
Fourth in the post-etch upload series. This release incorporates a couple
of patches, including one from Jason McCormick that implements an option
not to obtain tokens. I love it when people do my own TODO items for me!
There are a few other minor fixes as well, and with this release I've now
uploaded the package to Debian unstable with a transition package for
upgrades from the old libpam-openafs-session.
In preparing the Debian upload, I discovered that this package (and
therefore probably libpam-krb5 as well) didn't support DESTDIR on make
install, so I'll get that fixed in the next release.
You can get the current version from the
pam-afs-session
distribution page.
2007-04-11: Various Debian uploads
Fifth, sixth, and seventh in the post-etch upload series are new revisions
of the rssh, opensaml, and shibboleth-sp packages. I took some time with
rssh (a restricted shell that only allows certain operations, which I use
as the shell of a backup user on one system so that I can do backups
automatically via rsync from other systems) to clean up the debconf mess
that had accumulated in the package and deal with most of the outstanding
bugs. There's still one change to the debconf templates pending that will
require translators to retranslate the template; the call is out and I'm
going to upload another new version in May.
The opensaml and shibboleth-sp packages fix C++ compliation errors with gcc
4.3, which will probably be the default compiler for lenny with any luck.
I took care of a few other minor cleanups while I was at it.
Still to go: gnubg, remctl, and kftgt, although the highest priority is to
fix the various serious bugs with the OpenAFS 1.4.4 package in experimental
and get it uploaded to unstable. I'll probably work on that tomorrow.
2007-04-12: Don't worry, part one
I have an occasionally nerve-wracking relationship with my body. I feel a
little guilty about that given how many people I know in chronic pain or
with serious long-term medical conditions, since I'm in excellent health
and my body has always worked just fine. Still, I'm one of those people
where "me" doesn't feel like my physical form, and my physical body
occasionally does baffling things without much warning. And when that
happens, I get very nervous until I understand them, and tend to assume the
worst.
Sort of a borderline hypochondriac, except that I don't invent things, just
worry about things that show up that I don't understand.
The Internet is wonderful for understanding medical conditions, and has
helped a lot with easing me through moments of panic. However, it is
notably lacking in one thing. There are very few sites devoted to
explaining all the things that aren't wrong with you, or that at
least aren't nearly as serious as you might at first think.
For example, today I noticed a hard bump under my arm near the middle of my
armpit that was tender to the touch and felt like a hard, round swelling
under the skin. If you do a search on the Internet, you will find many,
many sites that will confirm this to be a swollen lymph node, one of those
things that's usually harmless but might warrant a trip to the doctor (and
has cancer associations, so it can be particularly scary).
However, I happen to know, just from one past experience which freaked me
out, that I can, in fact, get pimples in my armpit. I know this because
the last time I had exactly these symptoms and was worrying over whether it
was a swollen lymph node, the bump turned into a red swelling in the skin,
formed a head, and eventually healed exactly like a pimple. I've yet to
find a single site out there that will tell you this, and some will say
that if your lymph nodes are red and tender, you should go to a doctor.
I'm going to give it a week or two and see if it turns out to be another
pimple, since that's way more likely.
I started, some time ago, keeping a medical log so that I could remember
things like this for which the Internet is decidedly unhelpful. I think
I'll start posting about a few of those, to throw them out there in case a
search helps someone else who has the same tendency.
Here's another: If you suddenly develop a grey-black object in your vision,
something that looks vaguely like an insect or cobweb but which doesn't go
away and which moves around when you move your eye, don't panic. You're
almost certainly not going blind, and you almost certainly don't have a
detached retina. You have what's called a floater. They happen to some
people and not to others. They're perfectly normal and usually don't
indicate a deeper problem.
If it's the first time you've gotten a floater, you might want to make a
appointment with an opthamologist to peer in your eye and double-check, but
unless you have dozens of them, are seeing bright flashes of light in your
peripheral vision, or have vision loss, it's not an emergency.
Yes, they're incredibly annoying. Yes, they can go away, but more likely
they'll come and go or be noticable or not noticable depending on the
brightness and the weather. You can sometimes get them out of the center
of your vision by rolling your eyes a few times. Yes, billions of people
get these and just don't appear to ever talk about them so that everyone
can experience the same initial panic for themselves.
First of an occasional series.
2007-04-12: More Debian uploads
Post-etch upload spree part eight was a new OpenAFS package. I uploaded
1.4.4-1 earlier to experimental and it failed to build on many
architectures. I was ready for some serious debugging, but it turned out
to be a missing build dependency causing aclocal to fail. That should be
fixed, and now 1.4.4 will go into unstable.
There are still some pending bugs that I need to deal with and I think I
want to migrate the location of upserver and upclient to match upstream's
policies. That's going to be a bit tricker.
Part nine and ten were new kerberos-configs (just a new translation) and
xml-security-c (portability to gnu-kfreebsd) uploads.
For a while, today was looking unproductive due to a variety of annoying
reasons, but then I managed to pull it out. That's a good feeling.
2007-04-13: filter-syslog 1.20
For some reason, I thought filter-syslog removed trailing whitespace from
syslog lines before applying its filtering regexes, but it doesn't. I came
to like that behavior from writing logwatch rules, so I added it. This is
an incompatible change from previous versions in that regexes written to
expect the trailing whitespace to be present will no longer match.
Hopefully that won't cause major problems.
You can get the latest version from the
filter-syslog
distribution page.
2007-04-13: More Debian uploads
Post-etch upload spree parts eleven and twelve go to remctl and kftgt, both
packages that I'd been building and uploading internally to Stanford during
the etch freeze. I cleaned up various issues and uploaded new packages
that include all the intermediate changes.
This about does it for the pending uploads except for MIT Kerberos 1.6
(which Sam may be taking care of — I haven't asked yet) and gnubg.
The latter will probably require more work, since I want to package the
0.15 stable branch snapshot.
Next week, I need to get back to project work, so the pace of Debian work
will probably drop off.
2007-04-13: eyrie.org downtime
eyrie.org may be off-line for a while. A machine room network
reorganization means that all of the systems need to be re-IP'd, and
unfortunately in the very short notice that I was given, I've been unable
to reach the person who handles DNS for me (Stanford has network
restrictions on running DNS servers due to a Linux vulnerability from long
ago).
What that means is that while the .stanford.edu names of my systems will
resolve to the right IP addresses, the eyrie.org names will not, and
therefore much content won't be accessible, including this journal. Posts
will accumulate and feeds should catch up once DNS comes back. Hopefully
that won't be too long.
I'm trying not to worry about it too much, since there really isn't
anything I can do about it. I feel too much responsibility towards
eyrie.org; it isn't the end of the world if it's down for a little while.
2007-04-14: eyrie.org back
A little after noon, DNS for eyrie.org was fixed, and now we're back. I
still need to go move some network cables around so that I'm not confused
later, but the IP move is complete and now the systems are on the right
network.
Less stressful than I was afraid of.
2007-04-14: Debian work
Post-etch upload number thirteen goes to gnubg. I took the time this
afternoon to package the 0.15 release snapshot, which fixes a bunch of
minor bugs. I'm not sure if the problem running it on non-GL screens is
still there; I'm guessing so, which means that I should forward it
upstream.
I was also going to upload a fixed version of libauthen-krb5-perl that
fixed the options reading from krb5.conf, but after looking at that, the
root problem is that the module is using deprecated interfaces. So I sent
mail to the upstream author pointing that out and offering to craft a
patch.
I also checked on whether I should go ahead and package the latest version
of GNU GSS.
Hopefully tomorrow, I'll get a GIMP plugin package that I volunteered to
sponsor checked and uploaded, but relaxation is the order of the weekend.
2007-04-14: Various hauls
Another major Powell's order plus a separate order have come in, I got some
used books from a friend, and today I went to the Friends of the Palo Alto
Library booksale and picked up a few things. So it's about time for
another of these.
Lloyd Alexander -- The Black Cauldron (sff)
Lloyd Alexander -- The Castle of Llyr (sff)
Margaret Atwood -- Oryx and Crake (sff)
M.M. Buckner -- Hyperthought (sff)
Arthur C. Clarke -- Against the Fall of Night (sff)
John DeChancie -- Starrigger (sff)
Michael Flynn -- Eifelheim (sff)
Laurell K. Hamilton -- Danse Macabre (sff)
O.R. Melling -- The Hunter's Moon (sff)
Richard Morgan -- Woken Furies (sff)
Mervyn Peake -- Gormenghast (sff)
Frederick Pohl -- Heechee Rendezvous (sff)
Geoff Ryman -- The King's Last Song (mainstream)
I also managed to get several duplicates (not reflected above) via various
means. Ah well. And there are several things at the library book sale
that, now that I'm home and can check my list, I actually want. So I
should go again next month (or tomorrow, but that's unlikely) and get more.
Not that I really need more books, of course.
The Laurell K. Hamilton is probably a huge mistake, but I have a very hard
time stopping reading a series. (I still think about going back to Robert
Jordan from time to time, and even selling my Terry Goodkind books hasn't
gotten me to stop thinking about them.)
The DeChancie is a pure impulse thing. The M.M. Buckner is because I'd
like to read the earlier books in her series before reading War
Surf, which won the Philip K. Dick award. The second one looks like it
may be hard to find.
Eifelheim, as a Hugo nominee, will be the next book I read after I
finish the one I'm currently working on.
2007-04-20: afsdb-load 1.20
Our nightly load of AFS volume information into our reporting database was
missing some volumes due to additional (more minor) format changes in the
new vos listvol format with OpenAFS 1.4. This version should now cope with
timestamps of "Never" and with a missing access count. The latter is
rather odd; I'm not sure why suddenly some volumes would have no access
count while others still show an access count of 0.
You can get the latest version from thet
AFS reporting database
distribution page.
2007-04-21: lintian.d.o back up and running
lintian.debian.org is now back up and
running, and I've sent my first-ever mail to debian-devel-announce. I
started the full archive lintian run with the latest version of lintian
back on Thursday and it finished late last night. There was a brief holdup
when the ftp.debian.org mirror on gluck wasn't updating, but that kicked
off this afternoon and I was able to do the necessary incremental.
Now a cron job is set up and it should be back to a nightly update cycle.
I already see a ton of interesting data from the tags, including several
false positives and some things that could stand a bit of attention. I
don't have time to do lots of lintian work right now, but this will make
various things much easier.
Special thanks to Bill Allombert, who gave me all the necessary information
to backport lintian to sarge and found all the bugs I would have run into
before I had to.
2007-04-21: Caught up on USEFOR
For the first time in months, I'm completely caught up on USEFOR. I
proposed text for two outstanding issues around constructing the Path
header field and then finally digested the huge thread about Injection-Date
and reinjection, wrote a new draft that addresses the issues and adds a lot
more explanation of history databases, and sent out the diffs to the list.
Now there will be a bunch of discussion, and I expect consensus is still
some distance off, but it's off my plate and on to others for a change.
Today was a great day. I got a whole pile of stuff that had been hanging
over my head taken care of, which feels great.
2007-04-22: Term::ANSIColor 1.12
Another documentation-only update. (I'm going to be surprised if there are
ever any code changes in this module, although maybe someone will come up
with extended color sequences.) Somehow, I'd missed documenting the cyan
and white colors all these years in one place in the docs.
You can get the latest version from the
Term::ANSIColor
distribution page.
2007-04-24: WebAuth 3.5.4
It's been quite a while since the last WebAuth release (September of last
year), and a variety of small fixes had accumulated. I finished resolving
some long-standing minor cookie handling issues in WebLogin and added code
to display the true final destination when authenticating for a Shibboleth
IdP, and that seemed like a good excuse for a new release. Also in this
release are configuration fixes for mod_webauthldap, better configure
handling with multiple Kerberos libraries installed, and better Heimdal
support.
The next release, 3.6.0, will have multi-realm support and the remaining
code that we need for guest account handling. I'll need to write quite a
bit of additional documentation for that, since the details of cross-realm
and multi-realm support are tricky.
You can get the latest release from the WebAuth distribution page. (I
still need to put up a regular software page for this on my own site
rather than just maintaining the service page.)
2007-04-27: PGP::Sign 0.20
Another case where I didn't really change anything about the module, but
someone correctly reported that the program that creates the module with
build-time configuration didn't flush its output before prompting. While
fixing that, I also improved the test suite output a little and then
uploaded a new Debian package with some minor updates and fixes.
You can get the latest version from the
PGP::Sign
distribution page.
2007-04-28: lintian 1.23.30
The last few days, I've had a ton of other things that I should be working
on but all I've really managed to focus on was lintian. I think it's
because each piece is nice and small and I get a quick sense of
satisfaction from finishing it. Today I dug into the reporting scripts,
figured out how they worked and what needed to be fixed, and tested new
HTML page generation that deals with several long-standing bugs. That
seemed like a good point to kick out a new release with lots of accumulated
bug fixes, so it will soon be in unstable.
This version should take care of a ton of false positives. As soon as it
ages into testing, I'll update lintian.debian.org and regenerate the full
tag list (which will also mean we'll start keeping override statistics).
2007-04-29: Some hierarchy catchup
Today was mostly a relaxation day, but I did start on processing the tons
of updates and improvements that Iulius has been sending to
news.admin.hierarchies. Several hierarchies should now be in better shape,
and he did identify a bug in the software (it was accepting control
characters in newsgroup descriptions).
I think I'm slowly clawing my way up to being in contact with all the
things that I think I should be doing. I wouldn't go so far as to say that
I'm feeling caught up or in control of them, but at least there aren't many
things left where I haven't touched them at all in a very long time. I can
accept making slow progress on a variety of fronts; leaving things entirely
alone is a lot harder.
Of course, now I have to go take my car in to be smog checked at one of the
test-only stations again, which is two hours out of the middle of some day.
*mutter*. I do agree with smog checks, but I wish they'd get less
aggressive about them once your car keeps consistently passing with flying
colors.
All posts