2007-01-03: pam-krb5 3.1
This is a bug fix release only, fixing some obvious problems in 3.0. The
3.0 release would go into an infinite loop if someone entered the wrong
password, due to some reshuffling of the code without updating the exit
condition.
I was considering completing the TODO list by adding some macros to handle
option parsing, but I wanted to get this release out there fast and didn't
feel like working on that tonight. I'll probably do that for the next
release.
The PKINIT support hasn't been tested yet and should still be considered
alpha.
You can get the latest version from the
pam-krb5 distribution
page.
2007-01-05: svnlog 1.14
For my editing work on the Netnews protocol draft RFC, I wanted to enable
commit notification for one specific file (to notify the working group of
any changes made to the draft). The commit notification scripts that come
with Subversion have lots of options in this area, but I prefer the output
from mine. I've now added a very simple option to limit notifications to
only those commits that affect files matching a set of regexes. If any of
the modified files matches one of those regexes, the regular full commit
notification will be sent (so this may not be entirely what one wants).
I also added a somewhat ugly hack to allow user confirmation of whether to
send the notification message before sending it. It's ugly because
post-commit Subversion hooks don't have a stdout or stderr available to
them (but do have stdin — odd), so I opened /dev/tty to print the
prompt. This will probably break horribly in any scenario other than a
local repository accessed via a file: URL, but that's all I needed.
You can get the latest version from the
svnlog distribution
page.
2007-01-07: 2006 in Books
I've been slow in writing this up (although only one day slower than last
year). Here's the summary of my reading for 2006. These details don't
count the four reviews I've written so far in 2007.
| Books read: |
93 |
| Total pages: |
31,659 |
| Average rating: |
7.03 |
| Pages per day: |
86.7 |
| Days per book: |
3.92 |
I read less this year than last year on any metric: 8 fewer books and 4,710
fewer pages (meaning that the books I read were on the average shorter as
well, despite reading four George R.R. Martin books). Not entirely sure
why that was, although the Winter Olympics did take two weeks away from
reading on top of the regular company, vacations, and other interruptions.
The first half of last year was just slow for reading.
This coming year, I'm going to drop any explicit goals on how much I read.
I enjoy finishing books and seeing the numbers of reviews go up anyway, so
I'll have plenty of motivation, and I don't want to add stress to a
recreational hobby.
My reading was far more diverse this year, which was one of my goals. Only
61 of the books I read were SF: 17 genre classics or award winners and
another 20 books that were published in the last couple of years (the
remainder being various books from my long backlog). To that, I added 14
books in other genres (including three literary classics, a category I want
to increase), 15 non-fiction books (only three of which were SF analysis,
although James Tiptree, Jr. could arguably count as a fourth), and 3
graphic novels. This feels like a good mix, although I'd like to add more
graphic novels.
Three books stood out above all others this year: Geoff Ryman's brilliant
Air and Nicola Griffith's mainstream series The Blue Place
and Stay. Other notable and recommended SF books were Parable
of the Sower (Octavia Butler), Slow River (another Nicola
Griffith), and Komarr (Lois McMaster Bujold, one entry in the long
Miles Vorkosigan series). Ellen Raskin's The Westing Game is an
exceptional young-adult novel.
The remaining notables are all non-fiction. I put some effort into
learning time management techniques this year, and can wholeheartedly
recommend Time Management for System Administrators by Tom
Limoncelli and Getting Things Done by David Allen. I also finally
read Strunk and White's The Elements of Style and wish I had done so
years ago. Other non-fiction notables were James Tiptree, Jr.
(Julie Phillips), Zen and the Art of Motorcycle Maintenance (Robert
Pirsig), and About Writing (Samuel Delany).
2007-01-07: reminder 1.6
When editing a reminder the other day to change the repeat period, I got a
strange error message complaining about the repeat interval, couldn't fix
it, and then when I aborted, it deleted the reminder. Today I finally got
a chance to look at that bug. I'm still not sure what error I got and what
it meant, but now reminder will report correct error messages and not
delete them on aborted edits.
While I was at it, I also made a change that I'd been intending to make for
a while: reminder will now reuse the reminder numbers of completed
reminders rather than always using a number at the end of the sequence.
You can get the latest version from the
reminder distribution
page.
2007-01-08: Some new year thoughts
I don't do new year resolutions. There's too much baggage and weight on
the concept. I do use any long holiday to consider how my life is working
and think about ways to change it, though, and the Christmas holiday is
particularly good for that since I've often done a lot of thinking over my
long October holiday, tried to put it into practice for a couple of months,
and have some results to try to internalize and analyze.
One thing that has been mostly successful over the past couple of months is
my plan to multitask less. I used to do this a lot when I was a college
student, and shortly thereafter, and I probably could get more done that
way. But I've increasingly found that multitasking makes me irritable and
frustrated, and having the TV on or other background noise going while I'm
doing things really does decrease my productivity and make it harder for me
to get things done. So I'm going to continue to watch much less TV, not
have sports on regularly on the weekend, and try to focus on doing one
thing at a time.
The major challenge for this will be in the evenings, when I often have
left-over work that I want to get done, I have reviews that I want to
write, and I'm also spending time writing fiction and talking with my
friends. It's going to be hard for me to avoid multitasking there and
still get things done that I want to do, like writing reviews. This will
require more thought.
The other goal I have is to take more definitive weekends. Right now, I'm
very overloaded with work projects and non-work projects that I want to do,
and I've been packing my weekends too heavily. This has to stop, even if
it means that I simply do less and some things (like INN) just won't get
done the way that I want them to. I'm tackling this on a few fronts:
first, I'm trying to spend less time working on work projects and just
letting them take longer. Second, I'm taking Saturday as a day to do
things off-line, trying to minimize how much time I spend on the computer
and instead catching up on reading, doing the laundry, doing mindless
things around the house, playing video games, or watching DVDs. This gives
me a recovery day and leaves Sunday for non-work projects if I feel like I
have energy. And if I don't have energy, I'm trying to give myself
permission to take Sunday off as well, even from non-work projects.
I'm getting back on top of to-do lists. More work projects are heating up,
but I'm trying to stick to having more time to myself, more time to pursue
non-work things, and therefore work may just have to slip. I've been
devoting too much time to work and not enough time to myself the past
couple of years, and I've started making that shift after October and it's
helped a great deal.
2007-01-10: Asimov's, October/November 2006
Review: Asimov's Science Fiction, October/November 2006
Editor: Sheila Williams
Issue: Volume 30, No. 10 & 11
ISSN: 1065-2698
Pages: 240
This is the fall double-sized issue, which among other things means that it
has a Norman Spinrad book column, always something to look forward to. As
he has before, he complains about a trend in SF publishing (pushing series
over single novels) while writing detailed and interesting reviews of
several books. He liked
River
of Gods better than I did, but his book review columns are always worth
reading.
Of the other non-fiction, I liked James Patrick Kelly's discussion of web
sites and Locus Online. This issue also had a much-appreciated letter
column.
The fiction is, as it would have to be for this large of an issue, a mixed
bag, but there are several good and thought-provoking stories here and a
wide variety of ideas.
"A Billion Eves" by Robert Reed: A future society finds a way of
travelling to alternate worlds, which is then abused by a man who kidnaps a
sorority house of women into a new world with the intent of using them as a
harem. He founds a new religion based on this practice, very reminiscent
of certain fundamentalist religions. Several worlds down the line, a woman
growing up in one of the derived societies fights with its expectations,
beliefs, and abuse of the world to find another belief and set of goals.
This is the perfect antidote to nearly every story Orson Scott Card has
written, a story about how parallel worlds would really interact with a
patriarchal, fundamentalist faith. It is, in turns, disturbing, remarkably
perceptive about how religions form and diversify within a doctrinal base,
and realistic but still desperately hopeful. Even in a stifling and
twisted ideology, people can still find their own lessons, interpretations,
and ways of breaking free. (7)
"Biodad" by Kit Reed: More on the disturbing side, this is the
story of a family reunion gone wrong after a woman tracks down her
anonymous sperm doner. Good characterization, but not much of a plot apart
from the disturbing twist in the end. I didn't get enough feeling of
narrative flow to get into this one. (5)
"Dawn, and Sunset, and the Colours of the Earth" by Michael F.
Flynn: A Seattle ferry disappears into a strange disturbance, taking all
of its passengers with it. The families of those who went missing try to
cope, the city and others investigate, and scientists attempt to develop
theories as to what happened. In one of those unusual structures that one
can only get away with in short fiction, there's no story in the sense of a
narrative arc, just a set of reactions from different individuals (although
moving forward in time so some of the investigation unfolds). There's no
true conclusion, just a profile of how various people react to the
disaster, written in very different styles. I found it surprisingly
compelling reading, though (although I would have preferred more of an
ending). (7)
"After I Stopped Screaming" by Pamela Sargent: Written with
another unusual narrative style, this "story" is a short interview, late in
her life, with the woman kidnapped by King Kong. It becomes a
deconstruction of not only the movie but of the process and reactions of
celebrity, as well as a parody of the sort of answers celebrities give in
retrospective interviews. Not horribly memorable, but entertaining. (6)
"The Small Astral Object Genius" by James Van Pelt: I loved the
idea of this story. We've invented a way to send small cameras to anywhere
in the universe not close to the earth's star system, have them take a
picture, and return instantly. This odd technology, full of limitations,
is being used as a way of exploring the galaxy, and SETI-like is farmed out
to hundreds of thousands of individuals all over the world for the price of
a subscription and the hardware (and a lot of batteries). The protagonist
is a kid from a troubled marriage who works at this search obsessively as a
way of escaping the rest of his life. The ending involves a tenuous
connection between this search and the boy's family life that felt
artificial to me, which left me unsatisfied, but well worth reading for the
idea. (6)
"1 Is True" by Ron Collins: Another good idea story suffering from
a weak ending, this one starts with a corporate intrigue story featuring
the remnants of a programming team who invented a way to program illusions
using brain implants. I liked the way it started, drawing the protagonist
back into the next, more dangerous version of the original idea and
intermixed with memories of the woman who originally formed the marketing
and contract part of the partnership. Characterization is excellent,
particularly of the woman who never appears in the story and is only seen
through memories and interactions. The ending was a let-down, though: the
conclusion of the intrigue is too pat and the story of Yulani has a deeply
unsatisfactory resolution. (6)
"The Seducer" by Carol Emshwiller: This is a borderline-horrific
story of a seduction turning into a love story, mixed with a haunting of
sorts with clear psychological overtones. The element of the supernatural
may be entirely in the head of the protagonist and serves as a symbol for
dealing with childhood trauma. It fits together, but I didn't find the
completed puzzle that interesting, in part I think because I never felt
much reason to care about the characters. (4)
"Saving for a Sunny Day, or, the Benefits of Reincarnation" by Ian
Watson: The background here is that everyone in the world is reincarnated,
and it can be proven via computer who they were previously. As a result,
debts and assets now carry over into the next life, given out when someone
reaches a certain age, and people either work hard their whole life to
overcome debts from a previous life or live lives of luxury based on their
own inheritence, replacing all normal inheritence methods. The world
background is wryly humorous and well worked-out, and I liked the twist
ending. More amusing than serious, but true to human nature. (7)
"Foster" by Melissa Lee Shaw: A grieving widow takes on caring for
stray cats and discovers, as she tries to keep abandoned newborn kittens
alive, that she's being haunted by ghosts of her old pets. The emotional
dilemmas are drawn sharply and painfully, but the protagonist keeps making
different decisions than I would have made and by the end of the story I
was fairly disgusted with her. Evocative enough to leave a bad taste in my
mouth because I thought she was so wrong. (5)
"Down to the Earth Below" by William Barton: This is the novella
of this issue, the story of a group of teenagers who explore a cave and
find a passage in it to a world out of the early SF pulps that they love so
much. Events there are a combination of pulp adventure and adolescent
symbolism for sexual awakening, growth, imagination, and fantasy worlds.
It suffered considerably from my dislike of essentially all of the
characters except the woman (who gets a few good bits and then becomes a
non-entity) and my lack of interest in adolescent sexual stumbling. The
ending is a weird sort of fantasy feel-good story that takes it firmly out
of the realm of psychological realism and into wish-fulfillment, which was
odd given that I thought the rest of the story was leading up to something
more profound. (5)
Rating: 6 out of 10
Permanent
review page
2007-01-11: USEFOR status
The USEFOR IETF working group adopted my new Netnews protocol draft as the
new working group draft going forward at the end of last month, and I put
in another substantial chunk of time responding to mailing list messages
and issues raised with it. Now, it's time to polish it up for last call,
which means we've switched to a more formal discussion system. From this
point on, protocol changes will only be made in conjunction with issues
tracked in RT. This way, the working group chairs can call consensus on
the issues and I'll update the draft according to their call, which lets me
have my own opinion independent of being the draft editor.
I'm seeing if I can find the time to continue as draft editor. I'm trying
to limit my involvement to two hours a week (although so far it's been more
like five) given all the other things I have going on. As anticipated,
we're now getting into the period where the people in the working group who
favored the previous draft's style and position are raising as issues all
the things that I intentionally changed, so the discussion is a bit
tedious. I'm trying hard not to repeat myself and to stay out of the
discussion once I've said my bit.
One serious problem for the working group is lack of participants. Right
now, Charles and I are the most frequent posters, and since we want very
different things from a Netnews protocol specification, that's not that
useful. Frank mostly agrees with Charles in trying to push policy into the
specification (in my opinion). The rest of the working group comes down on
the other side of that question in polls of the whole working group, but
they're mostly not participating in the threads. I'm not sure how we're
going to deal with this going forward.
Anyway, right now, the chairs haven't had time to start turning threads
into issues yet, so I only have two minor wording changes accumulated for
this weekend's batch of updates (and in one of them we're still bickering
over the wording).
I'm considering posting to news.software.nntp to mention that we're
starting to do more work and faster work again, and now's the time for
people who dropped out to turn up and listen in, particularly if they know
how news software works (and ideally have written some).
2007-01-12: Benefits of INN
I've been feeling guilty for a while now that I've not had a chance to work
on INN. I need to make another 2.4 release, and more importantly I need to
finish a 2.5 release so that all the hard work sitting on the development
branch can see the light of day.
It has been nice, though, to not have that work go to waste. I put a lot
of effort into developing a really good portability layer for INN, a lot of
utility libraries that are generally useful, and a solid test suite for all
of them. I've been pulling code from it ever since for other projects.
Today, I spent most of the day working on IPv6 support for remctl, which is
almost trivial since I can grab all of the supporting code from INN and
build on the techniques I learned to keep #ifdef away from the mainline
code. I'm extremely happy with the results.
remctl 2.4 should be out, with IPv6 support, early next week. Maybe this
weekend I'll find time to put more work into INN and get closer to a new
release.
2007-01-16: pam-krb5 3.2
Douglas Engert had a chance to test with the new Heimdal release candidate,
and as expected there were a bunch of compilation errors in the PKINIT code
in pam-krb5. Those should hopefully be fixed now. In the process, he also
uncovered a dubious way of handling Kerberos credentials, and when looking
at that I found a simplification that let me delete a whole layer of
indirection.
There are a few other minor fixes, such as proper man page installation
when building outside of the source directory and prefixing the default
cache name with FILE: to make the cache type explicit.
You can get the latest version from the
pam-krb5 distribution
page.
2007-01-17: remctl 2.4
The big change in this version is IPv6 support, using the framework and
infrastructure that I'd built for INN. I'm really happy with how well this
works. It keeps all the problems self-contained in the portability layer
and the code can use the new IPv6-aware interfaces like getaddrinfo and
getnameinfo without having to think about IPv6 issues.
I also fixed a problem with detecting network errors when sending tokens,
added a newline after server-provided error messages in the command-line
client, and added a distinguished error for too many command-line
parameters.
You can get the latest version from the
remctl distribution
page.
2007-01-23: krb5-sync 0.4
Today I finished writing and testing the code in our status and password
synchronization plugin to support queuing. AD password changes aren't ever
queued, but we fail a password change if there's already a queued change.
AFS password changes and AD status changes are queued if they fail so that
they can be retried later. I also finished the Perl script that lets one
manipulate the queue.
You can get the latest version from the
krb5-sync
distribution page.
I've also been talking with the MIT Kerberos developers about the best
plugin interface to use going forward so that the plugin hooks will be in
MIT Kerberos 1.7 and I can supply only the module and supporting utilities.
2007-01-24: pam-krb5 3.3
This always takes me longer than I think it will, which is rather annoying.
Also, doing a good job of supporting arbitrary preauth options in the PAM
module context is a pain in the ass. As a result, the code is now more of
a mess than I'd like, and for 3.4 I'm going to need to do some significant
refactoring and clean up the option parsing.
But, hopefully this should work. This is the first version with support
for the MIT Kerberos PKINIT branch, and it should also fix various
compilation problems on Heimdal. A fair chunk of the time I spent on it
this afternoon was to clean up the error handling to use the new MIT
Kerberos error functions, so it should also have better errors. (It's
really annoying to have to free the error string after using it.)
You can get the latest version from the
pam-krb5 distribution
page.
2007-01-25: Life update
This week has been a good week for getting project work done. I'm
basically done at this point with Kerberos password and account status
synchronization between MIT Kerberos, Active Directory, and an AFS
kaserver. The code is deployed in test and working. We need to do another
end-to-end test when I have the rest of the account creation infrastructure
in place, but I'm pretty happy with what we've got now. There's even a
remctl interface to manage the queue.
Other than that, I'm still feeling very behind. I'm feeling particularly
guilty about INN, on which I've done no serious work in about six months.
Realistically, I'm just not sure how much time I have to work on it, or on
other Usenet things (like News::Gateway) right now. I do hope this
changes, since I've really enjoyed working on it, and I've used the work
I've done on INN in a ton of other projects.
In terms of general life, I want more time to myself, and I need to do more
general apartment cleaning, reading of things other than books, and so
forth. I'm doing surprisingly well with book reading so far this year
(figures, right after I decided not to worry about reading 100 books that
I'm currently ahead of schedule), but I'm behind in other things, like
Locus. This isn't any sort of emergency, of course, but I like
being current in the SF field.
So, pluses and minuses. Thankfully, work is going to settle down soon, and
if I can get through some additional deliverables for the Kerberos project,
I should have time to breathe and think about other work I want to get
done. I'm a bit pessimistic about being able to pick up some of the open
source projects I've mostly dropped, but we'll see how it goes.
2007-01-28: kstart 3.7
Thomas Weiss noticed that somewhere in previous code restructurings, I lost
the code that paid any attention to the value passed to -H (to verify that
the ticket has at least that much remaining life). This has been fixed,
and the documentation of -K and -H has been improved.
You can get the latest version from the
kstart distribution
page.
2007-01-28: cvslog 1.54
Slava Girshman pointed out that the URLs generated by cvslog for cvsweb
(and similar web CVS repository viewers) contain the invalid CVS
pseudoversion NONE when a file is added or removed. I've now changed the
URLs to suppress ones for removed files and to link to just the file,
without diff revisions, for added files. Hopefully this improves the URLs
in general for everyone. It seemed reasonable to me.
You can get the latest version from the
cvslog distribution
page.
2007-01-28: pam-krb5 3.4
3.3 still didn't build with Heimdal 0.7, since it turns out that 0.7
already had the PKINIT functions but took a different number of arguments.
This was enough to get me to finally install a Heimdal development
environment on my box at home and track down all the remaining build
problems.
So, pam-krb5 now builds with the PKINIT support in Heimdal 0.7, although I
don't know if there's enough there to do useful things. Various other
problems on Heimdal are also cleaned up, the error handling is now a bit
better with recent Heimdal snapshots, and I never attempt to use
error_message with Heimdal since Heimdal's krb5-config doesn't include
-lcom_err.
You can get the latest version from the
pam-krb5 distribution
page.
2007-01-31: Life update
There are very few things in this world more boring than writing audit
code. But thankfully I think I'm almost done with it, and then get to see
how long it takes the audit to run. Afterwords, I should have lots of
interesting statistics and information about our Kerberos realm, along with
a bunch of things that I have to fix at some point.
This, like most things lately, is taking me longer to do than I expected.
What I'd really like, right about now, is a week of vacation, entirely by
myself, where I can generally catch up on the world. I could take that,
but I don't really think it would help as much as it feels like it would
help. Really, I'm fairly on top of stuff; there are just several things
that I want to have more time to finish, and I want each weekend to be
longer.
Hopefully the last two days of this work week will feature fewer
interruptions and more time to sit down and concentrate on finishing things
and putting more stuff in order. Then I'll have a weekend (although one
day is a Super Bowl Party, so I won't get as much of it to myself as I'd
ideally like).
I should get a few software releases out today or tomorrow.
All posts