Posts for September 2006

2006-09-01: Scholarly haul

It had been a while since I'd placed a book order, and once I got thinking about that, I could resist. The great reviews of the Tiptree biography didn't help my ability to resist.

Samuel R. Delany -- Silent Interviews (nonfiction)
Harper Lee -- To Kill a Mockingbird (mainstream)
Gabriel García Márquez -- One Hundred Years of Solitude (mainstream)
China Miéville -- Between Equal Rights (nonfiction)
Julie Phillips -- James Tiptree, Jr. (nonfiction)
Duncan Steel -- Marking Time (nonfiction)

I'm still working on A Feast for Crows. So long. So very long. (Also, honestly, not that great.)

2006-09-02: Reading day

Well, I had hopes for Friday letting me get a better jump on the weekend relaxation, but I think I needed a couple of days to recover. Work politics are always stressful. So, nothing really got done today (not that anything needed to, but I do have a list of Debian things that I wanted to get done) and I ended up reading for most of the afternoon. I've finally finished A Feast for Crows and have several reviews stacked up, which will hopefully be coming soon.

For tomorrow, maybe I'll focus a bit more on finishing a few things up, cleaning out e-mail, uploading new versions of lintian and libpam-krb5, and maybe getting the new OpenAFS build uploaded as well. Or not. It's the weekend -- we'll see how it goes.

2006-09-03: pam-krb5 2.3

With Quanah's help, pam-krb5 has been ported to Solaris (Solaris 8 at least). I had to add a flag saying not to delete tickets when the session closes, though, since Solaris closes the session before running the user's shell (bizarre).

I'd been hanging on to those changes for a bit to see if I had time to do anything else, and then got a bug report that the prompting function in the module didn't properly handle multiple prompts due to an incompatibility between the Linux and Solaris PAM interfaces. That's now fixed (and I also figured out how to enable the feature of retrying authentication against the master KDC in krb5.conf, which will be useful).

You can get the latest release from the pam-krb5 distribution page.

2006-09-03: svnlog 1.12

Today, I finally got around to making a change I'd been planning on making for a long time, ever since the last time I imported someone else's Debian package into Subversion and had svnlog retag all the bugs as pending. Taking an idea from the lintian commit scripts, svnlog will now only scan the commit message for bug closers, not the diff of any file named debian/changelog. The latter was a neat idea and the code was rather fun, but there were too many false positives.

Also following the lintian commit script, it now sends a separate mail message to the Debian control address to tag the bugs rather than cc'ing it on the (possibly huge) commit message.

You can get the latest version from the svnlog distribution page.

2006-09-03: cvslog 1.52

Dirk Jagdmann sent in a patch that allows one to specify options in the base cvsweb URL and still have cvslog put the file information into the appropriate place in the URL. I cleaned that up a little bit and added it and then discovered that I'd also rewritten the file wrapping code and had never committed that. So that's in this release too; hopefully it won't break anything.

You can get the latest version from the cvslog distribution page.

This isn't the major update that adds better option parsing based on what I did for svnlog. I ran out of time this evening to do that.

2006-09-04: End of long weekend

Today was a very good day, a social day unlike the rest of the weekend. That means I had a good time (and some very good sushi), but don't have a lot of concrete work to show for it. Alas, tomorrow it's back to work. I'm going to start by finishing our local Kerberos 1.4.4 package and testing that for an upgrade on Thursday, and then will try to continue my catch-up on the many, many things that I need to be doing.

Yesterday I did manage to get my to-do lists up to date and did a lot of e-mail catchup, and I may try to do some of that tomorrow as well if I have time (and if not tomorrow, definitely Wednesday). That always makes me feel better.

I finally finished A Feast for Crows on Saturday, and am now reading Nicola Griffith's Stay, which is spectacular. I may finish that tonight. Maybe tomorrow I'll get a review written.

2006-09-05: Cutting back on Usenet

This feels very strange.

I decided last year that I'd done Usenet politics for long enough and was going to start cutting back, including getting out of doing Usenet newsgroup creation. The deadline is next month, so now I'm starting to really trim back. This past weekend, I started figuring out what I was going to do about Stanford's news peering and how I was going to cut it back to something that looks more like a top-tier university news server instead of a strange expression of my Usenet hobby and personal whims.

So today I deleted a bunch of peers, started mailing other peers about dropping or reconfiguring feeds, and started getting a handle on who the servers are talking to again. (It's really a credit to this software that it's run with very little intervention for months at a time; I've not had time to pay attention to it.)

It's not really a good feeling. Whenever I touch Usenet again, I remember how much I like it, and I want to go off and fine-tune feeds and get peers to fine-tune them and find more peers and be a more central site again for text Usenet, and I have to keep reminding myself that there aren't enough hours in the day and I'm already involved in way too much. But it's painfully difficult to give up a hobby, particularly one that I've been playing with for ten years. (I'm not giving it up entirely, mind, but I'm doing an upgrade of the news servers to Debian this fall, fixing a few automation problems, adding spam filtering to our mail to news gateway, and then will be trying to leave them completely alone for at least a year.)

Part of the trouble is honestly an ego thing. I made the biggest splash of any community in my life on Usenet. For a long time, I've been a significant figure in Usenet politics, around Usenet software, and in Usenet peering. There's no getting around it -- that's a nice feeling, and it's hard to give it up. I liked being listened to and I liked having a lot of leverage, but if I'm not going to stay engaged in the politics and discussions, it doesn't make sense for me to try to keep the leverage or to continue to occupy that sort of position. I won't, even if I want to, and that's the way it should be. Still, I come away from that decision feeling smaller.

However, it's time to stop. I want to get the time back so that I can use it to work on Debian, or Kerberos software, or OpenAFS, or other things that involve less ego gratification but more productive and important work somewhat closer to the leading edge of technology (and somewhat more relevant to my day job). For that matter, I want to get the Usenet politics and peering time back so that I can use some of it to work on INN; I'm not going to give up the software side, and there's certainly a lot of work there that needs to get done.

Goal: By the time I go on vacation next month, unsubscribe from news.groups permanently, document and release the software that I use to maintain the ftp.isc.org archives, and finish some automation around the news.announce.newgroups posting so that I can get more time back there as well. By the end of the year, move all of Stanford's news servers to Debian (must decide if I'm going to create my own Debian packages or use Marco's), move the mail to news gateway to a separate system and spam-filter it, and possibly automate notification to peers that the server is running a backlog trying to talk to them. By next summer, release INN 2.5.

2006-09-10: remctl 2.2

Thanks to Quanah's testing, I had accumulated various bug fixes for Solaris portability, the test suite, and builddir != srcdir builds. The interaction between gcc and glibc's sys/wait.h in unstable then caused a strange problem where the W* macros could not be applied to the int member of a const struct, so I added a workaround for that (and fixed various casting problems with printf on 64-bit systems).

In other words, no significant new functionality here, just a lot of minor bug fixes.

You can get the latest version from the remctl distribution page.

2006-09-13: WebAuth 3.5.3

I'm pretty happy with how I'm managing to keep development moving forward rather effectively on the various random Kerberos and authentication packages I maintain. It helps that I've had an excuse to work on things due to Stanford's big Kerberos transition.

This release finally includes the (four-line) Apache 2.2 compilation fix that I've been sitting on for rather too long. I finally found the time to go add better logging to the WebKDC, so now it will spit out not only the IP address from which the request came but the IP address of the person logging in to weblogin if the weblogin script passes that in (and the standard one always does). This should help with auditing considerably. I also wrote a new manual section explaining what all that cryptic stuff in the webkdc log is.

You can get the latest version from the WebAuth pages.

2006-09-14: Concentration

The biggest challenge right now is to manage to concentrate on something for long enough to make real progress towards getting it done, since I'm getting interrupted by just about everything right now. This is very frustrating. It's particularly frustrating because in less than a month I'm going on vacation, and various things need to happen before then.

The first step, prioritization, I mostly finished last night. Now I have to actually get the work done.

Before October 10th, I want to get the Shibboleth SP packages finished and uploaded to Debian, I want to get a new internal mail server built using Puppet and port all of the necessary configuration from qmail to Postfix and from bundle to Puppet, and I need to have the test environment for the new password propagation and verification mostly finished.

That doesn't sound too bad. There are multiple steps to each one of those projects, but it should be doable. But I need to get more than five or six hours of real work done each week.

2006-09-15: Productivity, finally

We're adopting Puppet as a configuration management system, but to date most of the work has been done by my co-worker, Digant Kasundra, and my only involvement has been getting briefed on how it works by the author and discussing theory and implementation with him. We're warming up to a very aggressive timeframe for starting to use it to manage all of our systems at work and deploying an infrastructure system using it is on the short list of things that I have to get done before my October vacation. Last night, I finally started writing Puppet configurations and getting into the rhythm of it, and all of today was devoted to working on Puppet.

This was extremely successful. I think some of that is focus and the willingness to simply ignore parts of work not related to the thing that I'm trying to get done. Another success factor was a complete lack of meetings today. At the end of the day today, I've ported pretty much everything our current Debian build system does after the initial build into Puppet, discussed implementation strategies for some additional features that we'll be contracting for, implemented the initial Postfix configuration, and started keeping notes on the things I still need to deal with. Most importantly, I think I have a handle on what the overall shape is going to look like (although getting support for overriding classes is going to be critical for us).

It's nice to reach the point in a major project like this where we've committed to the implementation technology. Nothing is every perfect and everything requires local customization, so there's always a long, uncertain initial period where I'm not sure if we're going the right route, I see a lot of different possibilities, and I'm not positive we're jumping in the right direction. When we commit, it takes a lot of the stress away and produces a sea change in how I look at the tools. Now, Puppet is something we're simply going to make work, and if it doesn't do what we want, we change it until it does.

This means that I now need to learn Ruby. I think learning Objective CAML just got pushed back farther again.

2006-09-16: podlators 2.0.5

A bug-fix release, mostly in Pod::Man, that turns off hyphenation and justification properly on Linux and eliminates some unnecessary Unicode characters. It also adds a bit better backward compatibility with Pod::Parser.

You can get the latest version from the podlators distribution page.

2006-09-17: Stay

Review: Stay, by Nicola Griffith

Publisher: Vintage Crime / Black Lizard
Copyright: 2002
ISBN: 1-4000-3230-X
Pages: 303

This is the sequel to The Blue Place and is almost impossible to talk about without spoiling the end of that novel. The back cover of Stay even gives away the ending of The Blue Place. If spoilers bother you, you probably want to stop reading now until you've read the previous book (and try to avoid reading anything at all about this novel). I did have the ending spoiled for me and didn't mind too much, but it took away some emotional power.

The Blue Place was compelling but quite a mixture, moving between detective, travel, and romance plots and not always focused. Stay achieves a purity of emotion and tight observation that draws on some of the best parts of The Blue Place and sheds side plots with relentless focus. This book is intense. Some of the plot structure of a detective story and a thriller remains, but as scaffolding to hold the character study and the raw emotions. I've rarely felt as close to a character as I did to Aud over the course of this story.

Stay is, on its surface, about grief. Griffith writes effectively from the perspective of a character who withdraws, who reacts to pain by seeking an environment and interactions that she can control, and who gets frustrated and impatient with the ways people try to sympathize. This is a type of reaction to grief that I don't see enough, done as deeply and well as I've ever seen it. There is a plot, an investigation, rescue, and fallout from both, but it's Aud's grief that structures this story. The reader watches Aud go through stages of grieving and acceptance close enough to traditional patterns to provide satisfying verisimilitude and unique and personal enough to distinguish her as a person.

The character study also goes deeper than grief. Stay is about the interaction of a person with the world. It is the thematic conclusion of the arc of The Blue Place, which took Aud from a detached private detective into expressed love of both another person and a place. The arc of The Blue Place was partly character-driven and partly expressed through setting. Its conclusion in Stay is wholly character-driven and follows perfectly Aud's realignment of who she wants to be and the pains and adjustments that forces on her world. The plot intrudes on her life and forces her to interact with the world again. The violence comes from and echos her anger at the intrusion of the world and its destruction of her love. Rescuing another defies her ability to withdraw, forces her back into responsibility. And the rescue and the violence lead to more complications, leading her farther into the world, forcing her to make hard decisions about the right thing to do, and confronting her with a life after grief.

One of my favorite characteristics of this story is that every action has consequences, and often not obvious ones. Life is a mess of unforseen complications. Throughout Stay, the right thing to do isn't always obvious in advance, and when it is circumstances change. I love reading about someone who makes mistakes, is honest with themselves, deals with the consequences, and then acts again. The second half of the book seems headed towards more traditional thriller territory and the world again defies Aud's expectations.

All of this is done with subtlely handled exposition, emotion that comes naturally out of setting and reaction rather than direct description, and tight, effective language that is never maudlin or over-wrought. As with The Blue Place, Stay is told in the first person and crawls inside Aud's head, and Aud has no patience for extended emotional exposition. She and the reader both understand the real significance of some of her internal monologue without belaboring the point, which allows the book to keep moving forward.

Griffith succeeds here, even more than in The Blue Place, in taking a hypercompetent protagonist and making her painfully human. I found it the perfect blend between a strong plot and sufficient danger to create suspense, and deep enough observation of character to show the pain and reversals of growth. Stay centers in emotions and interactions rather than in the resolution of an external plot. The book is full of wonderful touches: the descriptions of Aud's cabin that describe her character through showing what she builds, the role of Julia that starts a bit creepy and becoming bittersweet, Aud's rhyming with her name, a late touch that would warm the heart of any reader. I can't recommend it highly enough, although do read The Blue Place first.

Followed by Always, due to be published in 2007. I will be buying that in hardcover.

Rating: 10 out of 10

Permanent review page

2006-09-19: More Usenet work

Well, work on my day job towards wrapping up projects before my vacation has been going swimmingly, so rather than use the treadmill and read tonight, I decided to tackle pending Usenet work that I had been putting off. Mostly this involves writing (in some cases quite long) e-mail to people to whom I owed some, but I'm also working on the evaluation of the new Big Eight newsgroup creation system due October 1st.

From time to time, people at work tell me I write too long of mail. I'm not sure how much people have seen that on Usenet, since I haven't had many cases where I needed to write formal proposals or evaluations, but they're going to get a dose of it this time. Given that this is the last thing that I'm going to do for the Big Eight newsgroup creation process, I don't want to do a halfway job.

This is still a rather weird feeling, and I think I'm going to go through a grieving process probably sometime in November or December. Right now, though, it mostly feels like relief. I'm not sure how long I'm going to stick around for subsequent discussion after October 1st, but as of my vacation, I'm done, and I'm not going to start up again when I come back.

I'm not sure if this is going to feel as significant in retrospect as it does now, but it really does feel like the end of a phase of my life.

2006-09-20: On community

An off-hand comment in Zephyr about whether Usenet is dead, plus being in just the right mood this evening, prompted me to finally put this up on the web.

This is a post that I made to the largely dead net.subculture.usenet about a year and a half ago, another one of those posts that get stuck in my head and that I stay up well after midnight writing. I'm not really sure what to say about it, other than it's the best I will probably ever do at capturing what I feel about Usenet now. If anything, the feeling has gotten stronger over the past year and a half.

There's still a lot of emotion here.

On community

2006-09-26: Bookstore haul

Last night, I went to a good-sized chain bookstore for the first time in quite a while and spent a lot of time pointing out books that I thought she'd enjoy. It was only fair for me to spend a bit of money as well.

Khaled Hosseini -- The Kite Runner (mainstream)
Hope Mirrlees -- Lud-in-the-Mist (sff)
Ellen Raskin -- The Westing Game (children's)
Steven Savile & Alethea Kontis (ed.) -- Elemental: The Tsunami Relief Anthology (sff)
William Strunk Jr. & E.B. White -- The Elements of Style (non-fiction)
James White -- Beginning Operations (sff)
Gene Wolfe -- Shadow & Claw (sff)
Gene Wolfe -- Sword & Citadel (sff)
Gene Wolfe -- The Urth of the New Sun (sff)

The Gene Wolfe books are the really beautiful trade-paperback printings of the Book of the New Sun series. I already own that series (and have read it), but have the old ugly Pocket Timescape mass market versions and wanted to have nicer versions that go with my nicer copies of his other series.

I'm a little wry about paying as much as I did for The Elements of Style, but I don't want an on-line version of a book. I want to be able to hold it in my hand. That's worth $8.

I'm mostly buying Elemental to support the cause, and it's probably going to be a while before I read the short stories in there. I have several other anthologies that I'm more immediately excited about.

The James White makes me very happy. James White was a mid-list SF author from Ireland who wrote a long series of books about alien medicine that are highly regarded as good, readable fun. The last few are still in print, but the first ones have been out of print for quite a while, and I was delighted to see a trade paperback reprint of the first three.

2006-09-28: Marking Time

Review: Marking Time, by Duncan Steel

Publisher: John Wiley & Sons
Copyright: 2000
ISBN: 0-471-40421-7
Pages: 398

It's hard to think of something more central to our daily lives and yet more strange and full of odd names and bizarre rules than the calendar and our time-keeping system in general. After reading this book, one realizes that there are also few areas as full of simplifications, misconceptions, and misunderstood history, and Steel takes a great deal of glee in pointing them out. Subtitled The Epic Quest to Invent the Perfect Calendar, this is roughly a history of the modern Western calendar, but it's full of digressions on weird aspects of timekeeping, the theory and purpose of calendar design, different dating systems, and even a fair bit of political conspiracy theory (but more on that in a moment).

Steel is an astronomer rather than a historian, although he's clearly done quite a bit of research for this book. His focus shows in attention to the implications of different time systems, the astronomical processes that they attempt to model, and the detailed mechanics and mathematics of the calculation of Easter (which was the primary motivating force behind the Gregorian calendar). One of the most interesting parts of this book for me was the clear explanation of just how complex the revolution of the Earth around the Sun is, the various different "years" that one can measure, the problems with connecting a lunar cycle with a solar cycle, and the long-term drift that will require that any calendar based on nature, no matter how well-designed, eventually be corrected. Sometimes this is a bit hard to follow (particularly in the appendices which lay out more of the detail), but Steel repeats the same material from different angles and usually hit on one that made sense to me on the second or third try. I think I understand precession of the equinoxes now, and some of the bits of trivia are wonderful. I expect some of them will be old hat to an informed reader, but possibly different ones for different readers; I already knew a lot about the origins and rationale of time zones, for instance, but had never realized that obviously the Earth rotates slightly faster than once per 24 hours since it's also revolving around the Sun and the day is based on the return of the Sun to the same place in the sky.

There's a surprising amount of background of this type required in order to understand the goals of a calendar and the possible methods for accomplishing those goals. As with so many of these topics, a modern sense of superiority over our ancestors evaporates when one realizes just how much they got right with far inferior calculation methods and just how complex the problem is. For instance, the Gregorian calendar is frequently criticized for being strange and inferior to other possible calendars, but while Steel does lay out one calendar that is better by most measures, Marking Time will give you an appreciation of just how good the Gregorian calendar is at its actual goal. It may drift somewhat from the tropical year (roughly, the average year based on all four solstices and equinoxes), but it keeps the year length close enough to the vernal equinox year as to not require correction for around 8,000 years. And stabilizing the vernal equinox, not the rest of the seasons, was the purpose, since the vernal equinox determines the date of Easter.

The history, on the other hand, is more scattered and less satisfactory than the math and astronomy. We're introduced to most of the major players and some of the social history, but Steel's ability to create a sense of story is hit and miss. He points out interesting factual errors in common understandings of the origin of the calendar (and opens one's eyes to just how chaotic the calendar has been at times in the past), but he also speculates frequently about motives and digresses into math more often than he needs to. The most interesting parts of the history also suffer the most: the dating of Easter and the switch from the Julian to the Gregorian calendar are complex topics that rightfully dominate much of the book, but he never convinced me of his theories about the motivations of British colonization. I suppose it's conceivable, given the religious feeling at the time, that the English really cared deeply about controlling "God's longitude" where the date of Easter would stay stable on a single day following the introduction of a proposed superior calendar no one actually used, but with a lack of supporting primary sources and much speculation in the book, I grew tired of having most major political events at the time connected back to this.

There are a few other flaws of focus, places where I think Steel missed an opportunity to write a more popular history. For instance, while there's a detailed discussion of the origin of seven-day weeks, the derivation of the names of the weekdays is dealt with only in passing and only in English. The origin of month names is told only in snippets and mentions here and there, and he never goes through the entire calendar. He also focuses almost exclusively on the western calendar except for one survey chapter, which means that one gets a belly full of Easter calculation methods and almost nothing about the Mayan or Chinese calendars.

Still, this is certainly more analysis, detailed history, and fascinating trivia about the calendar and about time-keeping in general gathered in one place than I've ever seen before, and I thoroughly enjoyed reading it. It takes a bit to get used to Steel's writing style -- he discusses the structure of his own book in advance quite a bit more than is necessary -- but he has an infectious enthusiasm for his topic and a good grasp of making the reader feel more intelligent than before. And if you want an exhaustive explanation of exactly how the date of Easter is calculated, with the rationale behind all the steps and a historical discussion of choice of date and system, you're not likely to find a better resource.

Rating: 7 out of 10

Permanent review page

2006-09-28: Telsa's last post

One of the first on-line diaries I read was Alan Cox's, quickly branching from there to read his wife's commentary on the other side of the story. I've never met either personally, but Alan's entries were often a useful way to follow Linux development (and now that Kernel Traffic has apparently stopped, it would be nice to have something like that). Telsa's, on the other hand, were just fun, often amusing commentary on day-to-day life and hacker doings.

Alan switched his diary to Welsh a while back and then stopped keeping it entirely, but Telsa kept on for a bit and I kept reading periodically. About a year and a half ago, the entries became rather intermittant, and then stopped entirely. She's now written a final post to close things down more formally.

Sad, a little, particularly for me as many things seem to be ending right now, but I certainly understand her reasons. Her paragraph on some of the ways in which the diary became more trouble than it's worth resonate with me and match some of my feelings about Usenet right now, particularly the point about an increasingly confrontational climate.

The archives are still there, linked off that last post. I quite enjoyed reading them the first time. Thank you from a random stranger for the moments of amusement and laughter.

Last modified and spun 2017-03-25