2006-02-01: WebAuth ported to Heimdal
I had a difficult time getting started this morning, answering lots of work
mail and reading a lot of random other mail before being productive. Then
I fiddled around with making a distribution of the new version of fetchmail
(local Stanford thing, unrelated to and predating ESR's program of the same
name) that does K5 GSSAPI authentication. I think that's now ready for
builds on all supported platforms.
But this evening, I made up for the slow start; the URL stripping fix and
WebAuthStripURL documentation is done for 3.4.1, the et/com_err.h fix is
now in, and then, just because I felt like it, I ported WebAuth to Heimdal.
This took a bit of doing since WebAuth completely disassembles a ticket to
pass it along inside a WebAuth token and reassembles it on the other end;
I'm probably going to handle portability by pulling four or five functions
out into a separate file, one for MIT and one for Heimdal, and then
conditionally including one or the other based on Autoconf results. But it
wasn't as nasty as I expected, and I managed to get all of the tests
passing.
Tomorrow, I'll do the appropriate configure glue and then run the acid test
of actually building the module and WebKDC and seeing if they work. If all
goes well, I'll probably release 3.4.1 tomorrow, or maybe Friday if I run
out of time.
2006-02-02: Scattered work
Well, given that I was feeling rather behind on things, I was going to work
late tonight and try to catch up. I ended up working late, but I got
distracted into historical archeology and DNS name cleanup and didn't
actually get much work done.
Oh well.
Today was mostly a meeting, procedure, and write-up day anyway. Tomorrow,
I only have one half-hour meeting, so I should be able to get a new WebAuth
release out and then try to catch up on Debian work. Maybe, if I'm really
lucky, I'll also get a bit of OpenAFS documentation work done as well.
Geoff Ryman's Air is utterly awesome. I need to write a review of
it. I'm currently reading Salt: A World History and Paul J.
McAuley's Fairyland. The latter, so far, is doing nothing for me;
hopefully it will get better.
Not sure what I'm going to work on this weekend. Feeling really productive
and launching into INN work would be a great thing for me to do, but I
don't know if I'm going to feel up to it. Sunday, I'll watch the Super
Bowl, at least. I don't think I have anything in particular planned around
it, but I may see if anyone at work wants to get together for the game.
I'm still staying relatively caught up with e-mail, but I could be
quite a bit better, and I'm sitting on some things that I really need to
deal with. I could use a whole week of pure catch-up, but I'm too
interested in programming that I keep diving into some project and then
getting distracted. There are worse problems to have, I suppose.
Still owe people a few replies in comments; maybe I'll do at least one of
them now.
2006-02-03: No WebAuth yet
Alas, I didn't manage to even get the WebAuth release out, let alone catch
up on Debian or do the other things I was hoping to work on. I did make a
lot of progress on WebAuth, but discovered while testing Heimdal that I had
a few more things I had to fix (like not using deprecated OpenLDAP
interfaces). Not sure if I'm going to work on it this weekend or switch to
doing something else.
But I did have a good time playing table tennis tonight, and rather enjoyed
watching a basketball game. Tomorrow, food shopping; Sunday, Super Bowl
party.
2006-02-04: Bleh
Well, absolutely no energy today. No energy to walk, to go shopping, or to
get anything done on-line. I watched sports, took a nap, played video
games, wrote a little, and that's about it. As a result, I'm now feeling
very behind and rather frustrated. Getting angry at the politics I haven't
managed to ignore doesn't help.
I still need to do a bit of shopping for tomorrow. Ideally I should do
that tonight. It remains to be seen whether I do that before I go to
sleep.
Hopefully I'll have more energy tomorrow.
2006-02-05: Super Bowl Sunday
Well, that was fun, and worked out quite well. I went shopping in the
morning, Digant and Jon both made it, there were lots of snacks and a bit
of time for video games before the game, and we ended up enjoying the game
quite a bit. It wasn't the best football game -- in fact, it was rather
sloppy all around -- but there were still a few good plays. And I enjoyed
the largely slapstick-themed commercials.
I'm still feeling horribly behind on everything, but it's hitting me less
today than it was yesterday and I'm feeling a bit more energetic. It's
possible some of the problem is stress at work from trying to set things up
for new projects and new staff members combined with a lot of social
activity lately for me. I may need to be a hermit for a while and recover.
I also think I'm fighting off or coming down with a cold, which overall
may not be the worst thing in the world; it would be a good excuse to take
some time and recover and catch up mentally. Accordingly, I'm not horribly
worried about it; if I catch it, I catch it.
I have at least been playing some video games lately, mainly Outlaw Golf.
I can provide a litany of its failings, but despite that, I've still gotten
more than my $10 work out of it already. It's a fairly decent golf game
(although approach shots are incredibly frustrating), it moves right along,
and it scratches an itch. I don't know that I'll play lots of it in the
long run, but for right now, I'm enjoying it.
I haven't walked in a few days and I'm feeling rather bad about that right
now, so I'm going to go force myself to get some exercise in now and then
read until I fall asleep and not worry too much about when or whether I
make it into work tomorrow.
2006-02-06: WebAuth 3.4.1
Finally found enough time at the end of the day today to finish up the last
bits of work and get the release out the door.
The big news in terms of work is that WebAuth has now been ported to
Heimdal. I tested it with 0.7, but hopefully it should work with 0.6 as
well. The only thing that might break with 0.6 is that it relies heavily
on deep internal knowledge of the data structure of a credential, so if
that was rearranged, things may break.
The big user-noticable change is that I had to revert the change to not
modify URLs to content that isn't WebAuth-protected. Unfortunately, one
can't really tell whether WebAuth is going to be applied, and with this
change WebAuth didn't strip the WebAuth information from URLs protected by
.htaccess. This then broke certain CGI and PHP scripts. In partial
recompense, I documented the WebAuthStripURL option, which had always been
there, and it's now officially supported as an option.
There are a few other minor changes. I discovered that the LDAP module was
using deprecated OpenLDAP interfaces, so it no longer does, and the build
system should now find the com_err headers on recent Red Hat releases.
You can get the latest version from the WebAuth site.
2006-02-06: Non-fiction haul
I gave in and ordered more books in the middle of last month, since of
course I can't manage to read books faster than I buy them. This time, I
went mostly for non-fiction and mainstream fiction, since I have a ton of
SF already that I've not read.
L. Timmel Duchamp -- Alanya to Alanya (sff)
Marcus du Sautoy -- The Music of the Primes
(non-fiction)
Thomas Marton -- The Seven Storey Mountain
(non-fiction)
Vladimir Nabokov -- Lolita (classic)
C.S. Lewis -- English Literature in the Sixteenth Century
(non-fiction)
Rebecca Ore -- Alien Bootlegger (sff)
George Orwell -- Volume 1: An Age Like This (1920-1940)
(non-fiction)
Mary Doria Russell -- A Thread of Grace (mainstream)
Richard Seaver, Terry Southern, & Alexander Trocchi (ed.) --
Writers in Revolt (mainstream)
Edward R. Tufte -- Political Control of the Economy
(non-fiction)
Oscar Wilde -- The Picture of Dorian Gray (classic)
I'm not sure when I'm going to get to any of this, as of course I have all
those other books I bought still to read and some of these volumes are
rather substantial. But some of this non-fiction I'm really looking
forward to reading.
2006-02-07: Work before noon!
Not today, of course. Today there was an all-hands meeting from 10:30 to
noon, and then volleyball, and then lunch, so once again I didn't start
doing work until 2:30 in the afternoon and then got constantly interrupted.
Which sounds pretty much like Monday.
However, tomorrow is my day to work from home, and I seem to have recovered
my energy, so hopefully tomorrow I'll be able to get quite a few things off
my plate. Unfortunately, there are still some compilation issues with the
WebAuth 3.4.1 release, so I might finish a 3.4.2, then try to catch up on
Debian, and maybe either start the remctl work or do the hopefully final
releases of kftgt and S/Ident, now with more portability.
Maybe a book, or at least a magazine, review too, while I'm at it.
2006-02-08: S/Ident 3.6
Well, I thought I might be able to use my Autoconf macros from kstart and
remctl verbatim, but S/Ident needs both GSSAPI and the Kerberos v4
compatibility libraries plus has various other oddities of its own with
library linkage, so that didn't really work. I ended up writing new
Autoconf macros based on the ones I did elsewhere, the same as I did with
WebAuth. But that's done now, and S/Ident now uses krb5-config and has a
much more modern Autoconf setup.
That's pretty much the only change in the new release, except that I also
took the opportunity to make Kerberos v4 support optional. MIT is
eventually going to drop the Kerberos v4 backward compatibility libraries,
and I want to keep S/Ident in Debian for those few applications where it's
still useful.
You can get the new release from the
S/Ident distribution
page.
I expect this is the last release I'll do for quite a while. Maybe someday
I'll clean up the code a bit more, but given that we're going to stop using
S/Ident for nearly everything, if not everything, I really shouldn't spend
a lot of time on it. There's a lot of other work to do.
2006-02-08: kftgt 1.12
For kftgt, I could just use the macro that I developed elsewhere. So now I
do, and it uses krb5-config like everything else. I also fixed a problem
with make install when building in a different directory than the source
directory.
You can get the latest version from the
kftgt distribution
page.
Hopefully this is the last release of kftgt that I'll do for a very long
time, if ever, since soon we'll be eliminating the need for it.
2006-02-15: Mood crash
I've been really quiet lately since I'm going through a bit of a mood
crash. This is normal for me, particularly after a spurt of strong
productivity such as, well, pretty much all of December and January.
Unfortunately, while they usually last two or three days, this time it came
at a time of a lot of stress at work and a bunch of other things going on,
so it's been lingering for a while. I've been fighting mood for the past
couple of weeks and the end of last week and I've been very short on energy
and motivation the end of last week and the beginning of this week.
I'm slowly starting to dig myself out from under it, but I'm having to do
aggressive prioritization since right now I feel like I'm horribly behind
on everything. I'm also trying to take it easy, not push myself too hard,
and tackle things one step at a time. So don't expect to hear a lot from
me for a little bit.
On top of that, it's also the Winter Olympics, and I watch the Olympics
fairly aggressively. The TiVo has been programmed to pick up nearly
everything I can with an old single-channel TiVo, and I've been submerging
myself in sports I love and rarely get to see, like curling. All of my
reading and exercising time are going into that right now, so this month
will likely be very short on book reviews. I have another magazine review
and another book review pending, and then I'm hoping to finish a few other
short books but it may be a four or five review month.
Tomorrow, the goal is another WebAuth release and then some Debian work. I
don't think I'm going to get to remctl work this week, but I can start that
next week. Thankfully, the upcoming weekend is a long one, so hopefully I
can use that to shed the rest of my low mood.
2006-02-17: WebAuth 3.4.2
Well, that was a full week later than when I first intended to release it,
but it's finally released. This release fixes some additional compilation
portability issues on Red Hat and with Heimdal and documents the Kerberos
credential encoding for WebAuth tokens.
You can get the latest release from the WebAuth v3 site.
2006-02-18: New review indices
I did this work a while back, but I never mentioned it here, so I will
belatedly. For my reviews, I've now added a couple of additional indices.
I now have a list of
book reviews sorted by
review date. I've also indexed all of my short SFF reviews from both
short story collections and from magazine reviews in a new
SFF short fiction
index. I'm not sure how useful either will be for people other than me,
but I find both interesting and am using the latter to track which short
fiction writers I like.
Also, just tonight, I've added the
Sunburst Award to
my list of SFF awards. It's a fairly new award, starting only in 2001, but
it seems to be picking some interesting books. So far, it's the best award
in terms of my rating because the only winner I've read is Air. I'm
sure the perfect track record won't last.
2006-02-19: podlators 2.0.4
Yitzchak Scott-Thoennes pointed out that Pod::Simple doesn't provide any
parse_from_filehandle compatibility routine, so this is yet more patching
around Pod::Simple's dislike of backward compatibility for Pod::Parser
parsers. In retrospect, it was probably not a good idea to use the
Pod::Parser API as the API for Pod::Text and Pod::Man, but I wish that
Pod::Simple would have gone to a little work. Oh well.
You can get the latest version from the
podlators
distribution page.
2006-02-20: Link round-up, comment notes
A few interesting things I've run across recently:
Jo Walton wrote a long post in her journal about
Marion Zimmer Bradley's
Darkover series and what she liked best about them, couched in the form of
a review of two of the continuations of the series by other authors. Even
if I don't agree with her about Mercedes Lackey (whose books fill the same
role for me as the Darkover books apparently do for her), this is a very
interesting discussion of what goes into good, memorable comfort reading.
Hal Duncan (author of
Vellum,
one of the best bits of experimental, non-linear fantasy that I've ever
read and one of the best books of last year), has made multiple posts about
his intense dislike of monotheistic religions. It started as yet more on
the cartoon thing, which frankly I'm sick of hearing about (mostly because
nearly everything I've read ignores complexities in order to push their own
agenda), but then it got much more interesting. It's horribly long-winded,
as Hal Ducan tends to be in his blog, but embedded in here are a lot of the
reasons why the whole Christian mythological structure just doesn't work
for me. See
one,
two,
and particularly
three,
and
four.
Worth skimming, I think, and the third post really clicked for me.
Note that I'm not interested in discussing what Hal is saying here. He
doesn't like monotheism, he particularly doesn't like Christianity, and if
you're Christian and are going to take attacks on your religion personally,
you're probably going to hate his posts. Please, spare yourself the stress
and don't read them if you don't want to deal. Otherwise, read them, don't
read them, it's up to you.
That leads to the other note, which is that after thinking about this for a
while (and stressing over it), I'm going to drop the political discussion
in the earlier comments on my terrorism essay. I realized that I wasn't
getting anything out of talking about this more, I don't really believe in
my ability to convince someone else that they're wrong, and I don't feel
like having the discussion can possibly accomplish enough to justify the
(very high) emotional energy cost for me. And by this I mean nothing
against the people who commented; I'm more apologizing for not being in the
right mental space to be able to usefully have a discussion.
I'm thinking about cutting way back on my exposure to political discussion,
actually, as it's not doing much good for me and it's making me angry and
frustrated right now. Many of my friends have done the same, several have
dropped the subject completely in their journals, and I've already stopped
discussing politics in one place where I used to. I feel some sadness at
the breakdown of meaningful political discourse, and I suppose that if
people who are capable of it stop talking about politics with those who
disagree it will only get worse, but at this point politics is so polarized
it requires the patience of a saint to try to have a meaningful
conversation.
On a different note, and unrelated to the terrorism thread, please note
that this is my personal journal and I can, will, and have simply deleted
comments from people who act like asses. The most recent instance of this
was from someone who decided to decided to call me fifth-grader names
because I didn't like Green Rider. If you want to disagree with a
review, please, feel free, but I expect either a polite statement of
personal opinion or some real analysis. If you start throwing around
insults, if you're incoherent, if you start saying things that piss me off,
and particularly if you spam the comments with multiple made-up names, I'm
going to just delete your comment without a second thought. This is my
soap box and you're a guest; if you want to say whatever you want, go get
your own soap box. Livejournal gives them away for free.
2006-02-21: Some Debian work
Well, I'm still struggling to get things done, but at least I did get some
Debian work done today. This afternoon, I updated the OpenLDAP packaging
repository to OpenLDAP 2.3.20 and showed Quanah how I did that, getting us
closer to a recent version of OpenLDAP in Debian. And just now, I uploaded
a new version of krb5 to Debian with a few bug fixes.
Hopefully even more tomorrow. I need to do something about gnubg; one
problem I know how to solve, but it's also failing on AMD64 systems, and
that problem I'm finding a bit more mystifying. Although I do now have an
AMD64 build system where I can start poking at this.
2006-02-23: Push, push, push
Now I'm starting to push through and get things done, even though it's like
going uphill. Today, a tiny bit of OpenAFS work, a small bit of machine
building work, and more documentation of group standards. Tomorrow, maybe
even more. The Olympics are almost over, after which it will be much
quieter, I'll have fewer distractions, and I expect that a more productive
mood will come more easily.
kstart 3.1 was released today to add a flag to make group-readable ticket
caches, but it turned out that we couldn't use them anyway since K5 cares a
lot about the cache ownership. So I'm probably going to just take the code
out again and release kstart 3.2. I'm still pondering whether I want to
try to add something else for that release, just so that I don't have two
releases with no forward progress at all. Maybe some documentation of
usage examples, or some code refactoring. Probably the latter.
I'm still way behind with INN and Debian work, of course, but eventually
that will get fixed.
2006-02-24: Weekend!
And am I ever ready for one.
Once again, I got a lot of little stuff done today but no big stuff. Some
of the little stuff was rather important, though, so I can't really
complain. I think my productivity is clawing its way back and I should be
in good shape after the Olympics are over, or maybe a week after that.
This weekend, I'll hopefully get to a bit more catching up, but I'm still
not going to push myself too hard. The only things I'm really going to
push myself to do are watching the Olympics and writing some reviews.
2006-02-27: gnubg and gtimer
Yes! Actual productive work happened today!
I finished catching up on the Olympics (which didn't require much, since I
stayed up until 3am last night watching the evening program) and then, at
about 2:30pm, decided that I really needed to get out of the house and away
from the TV and do something completely different than watching TV.
One cross-campus walk later, I'd picked up my mail, cleared my head, and
decided to try to knock some things off the top of my to-do list in the
nice, quiet, empty office rather than going back home. New versions of
gnubg and gtimer have now been uploaded and I feel less guilty about my
Debian packages now (although libpam-krb5 still needs serious attention).
I didn't fix the AMD64 crash in gnubg. It turns out that I don't have an
AMD64 system where I can really test and the valgrind output wasn't
helpful. I did poke around and discover how to log on to a porter system,
only to discover that it works fine in text mode. So, appeal out to the
debian-amd64 list, and hopefully someone can track down the problem.
I did, however, retire the gnubg-bearoffs package in favor of including the
smaller database in gnubg-data and building the larger database at
installation time if desired. In the process, I learned how to write
debconf prompts from scratch. gnubg now runs all the autotools at build
time (another huge learning experience) and I tracked down and cleaned up a
bunch of other minor problems. The FTBFS on hppa and m68k should hopefully
now be fixed as well.
quilt rocks my world. It's almost like using a real version control
system, plus has fancy patch features. Both dpatch and (particularly) dbs
start annoying me quickly, but quilt just feels smooth.
2006-02-27: More gnubg
So, while some of my other changes to this package were still worthwhile,
the fix to link against libsupc++ didn't fix the build failures on hppa and
m68k. Still all the same unwind and gxx_personality symbols are missing.
I'll have to do more poking.
Also, the AMD64 crash appears to be inside Guile somewhere. Suggestion:
disable Guile. Well... I can do that, but that sucks. There have actually
been requests for that functionality. I don't want to just turn it off and
not try to figure out what's wrong. Ugh.
Other than that, a scattered but not too bad day in which I learned more
about managing our DNS servers and learned far more about Alsa and udev
than I wanted to know.
2006-02-28: OpenAFS docs
I got some more work done tonight, continuing to work on the things I've
been feeling guilty about. This time it was OpenAFS. The patch that
should fix builds with AMD64 and Linux 2.6 is now pulled up to the release
branch, I answered the old mail about the new reference docs, fixed a few
things, added a bunch more to the to-do list, and fixed some issues with
the HTML conversion.
The biggest fixes are that all of the interpage links should now work and
the synopsis should be wrapped nicely and hence more readable. The fix for
the synopsis is a little ugly, but it's livable.
You can see the current working tree at
http://www.eyrie.org/~eagle/tmp/openafs/ (although note that if you read
this much after I wrote it, this link will probably be dead). There's an
ever-growing list of things to do, but we're getting there.
Looks like this month will end with only five book reviews, but I have a
book pending review and will likely finish another tonight or tomorrow, so
I'll catch up again. On the Olympic month, and a short one at that, I get
to fall a bit behind schedule.
All posts