2011-09-02: C TAP Harness 1.8
The primary change in this release is to add checked memory allocation
wrapper functions to the basic C TAP library. The functions bmalloc,
bcalloc, brealloc, and bstrdup are now available. Each calls the
underlying C library version (without the "b" prefix), checks to see if it
fails, and calls sysbail to abort the test case if it did fail.
This release also fixes an edge case in the runtests wrapper where, if
SOURCE and BUILD were not compiled in, the -b option would be ignored
unless the -s option was also given. Now, one can set the build directory
without setting the source directory, although one would normally still
want to do both.
You can get the latest version from the
C TAP Harness distribution page.
2011-09-02: rra-c-util 3.8
There are a huge number of changes in this release, partly due to Julian
ÉLIE's work in reconciling the rra-c-util code with the portability and
utility layer in INN and partly due to the
opportunity to spend a couple of work days on switching
pam-krb5 over to rra-c-util. The results
from the latter will be coming hopefully soon.
The network connect utility functions now take an optional timeout.
fdflag_nonblocking is now implemented for Windows (required for the
non-blocking connect to implement timeouts). There's now also a function
to free the results of network_bind_all.
The generic PAM option parser now supports time configuration options,
which are a krb5_deltat if Kerberos is available. pam_get_item and
pam_set_item are now supported in the fake PAM library used for testing.
In the Kerberos and GSS-API Autoconf macros, use of krb5-config is now
avoided if the specific --with-*-lib or --with-*-include options are
given, since that normally means krb5-config will be wrong. The
environment variable to find krb5-config is now PATH_KRB5_CONFIG instead
of KRB5_CONFIG, since the latter is used for another purpose by the
Kerberos libraries. The C preprocessor symbol HAVE_KERBEROS is now always
defined by the Kerberos macros, not just the optional version, since it
makes it easier to reuse code.
Substantial new Kerberos portability and replacement functions are now
available, incorporating the portability code that was in pam-krb5.
The remctl Autoconf macros now support optionally building with remctl and
setting HAVE_REMCTL if remctl libraries are available.
The is_function_output TAP add-on supports taking an opaque data pointer
that's passed to the called function. There's a new run_setup TAP add-on
function that runs an external command and calls bail if it fails,
suitable for doing test setup in an external command.
There are a few other, more minor changes, including an update to the
latest C TAP Harness.
You can get the latest version from the rra-c-util distribution page.
2011-09-02: WebAuth 4.0.0
This is the culmination of most of my work over the past three months, as
well as quite a bit of work by other people here at Stanford. This is the
first version of WebAuth to support multifactor authentication, and also
contains a significant reworking of the WebLogin code to use more modern
programming techniques.
This should be considered a beta-quality release. There will be at least
one more release before we deploy this version in production for the
WebKDC and WebLogin server at Stanford. The changes to add multifactor
involved significant changes throughout the code base, and my
recommendation would be to wait a while as we shake out any bugs and
deploy the next release or the release after that.
The primary change, as mentioned, is addition of support for multifactor
authentication, including conveying initial login and session
authentication factors to the WebAuth Application Servers and from there
into the environment for use by applications. WebAuth-protected sites can
require particular authentication factors, or can require a (site-defined)
level of assurance. This code depends on a user information service
component that is specified in the documentation that comes with WebAuth,
but for which we do not yet have a public implementation. We hope to add
a simple version of this application as a separate software release in the
near future.
The WebLogin code has been rewritten to use Template Toolkit (instead of
HTML::Template) and to use CGI::Application. This will require many new
Perl modules be installed on the WebLogin server, and will require
rewriting all of one's local templates to use the Template Toolkit syntax,
but allows for much more powerful templating.
Both the libwebauth C library and the Perl libraries are halfway through a
rewrite in this version and will be changing substantially in the future,
particularly in making much better use of APR throughout the C library.
There are some other, more minor changes. See the release notes for all
the details.
You can get the latest version from the official WebAuth distribution
pages or from my WebAuth distribution
page.
2011-09-11: Vacation
The last vacation that I had was in April, and while that was lovely, it
was also a bit more stressful than the average vacation for various
reasons. Since then, I've been heads-down on one particular project
except for a small bit of traveling time, which is nice but not an
adequate substitute. But I have this week completely off, and can finally
unwind a little bit.
It's only Sunday and I was traveling more yesterday, so it's only starting
to hit, but I'm relaxing already from the lovely feeling of not having
anything that I have to do. (As usual, I had tentative plans to do
various things during my vacation, but this year they're going to get
thrown out even more than normal, since I just don't have the energy.)
One nice discovery from today is that while it's been some time since I
took a lot of pictures (since last April, in fact), it's still as
engrossing as always, and makes meandering walks lots of fun. So more
pictures than have recently appeared are likely for the next bit.
Hope everyone else's day was as lovely as mine was!
2011-09-12: Minecraft
Another photograph from near the Nanaimo River. We'll probably venture
farther afield later in the week, but it was still a little warm today and
we went for a long walk yesterday. Today's picture-taking was more
experimental, since it was somewhere I'd previously already taken lots of
pictures, so less of it worked, but I got a few neat effects.
The theme of this vacation seems to be Minecraft, a game that I've been avoiding since I really don't have
enough time to explore and play it to its full potential. This remains
the case even though I bought a copy, but I like supporting neat ideas.
It's unfortunately pretty badly buggy on my laptop (and I really dislike
the requirement for the Sun JDK, since I'm trying to switch completely to
OpenJDK), so I may not play much while on vacation and wait until I'm home
to poke at it some more.
But I'm finding that it's more fun currently to watch other people play
than to play myself. Since there are two other Minecraft games going on
at the moment, that provides lots of interesting things to look at and
speculate about. I really like the way the crafting system works in the
game and the complexity of it. I'm the sort who is looking things up
on-line, but I'm watching other people play who like discovering
everything for themselves.
It's really an excellent game if you like the free exploration sort of
thing. I'm not as sure on playing it on non-Peaceful mode yet, where
there are hostile enemies and one has to keep things well-lit, but I
haven't played it much myself so far and may change my mind.
I doubt anyone reading this hasn't already heard of it, but if you haven't
and you like exploration and construction games, take a look. (And most
people don't seem to be having as much trouble on Linux as I do.)
2011-09-13: More Minecraft
One of the pictures from today's walk along the Nanaimo waterfront. (The
pictures recently don't necessarily have much to do with the contents of
the post, but are just a picture I felt like posting from what I took that
day.)
Alternately, as opposed to what I said last night, I could get past the
bugs with Minecraft on my system and start playing it and end up playing
until 6am. And then go back and play more today. Good thing it's
vacation!
It's kind of amazing how much fun it is to build a giant staircase down
into a mine, or play with different ways to fence off a garden.
2011-09-17: End of vacation
This is a picture in late afternoon light of trees over one of the
colliery lakes at Colliery Dam Park.
Today's the last day of a really great vacation; tomorrow afternoon, I fly
back. We got in lots of walking, several nights of really good food (even
if the restaurant that I was hoping to go to again tonight was closed for
remodeling, I got to have the souvlaki earlier in the trip), and tons and
tons of Minecraft. Which will sadly have to get short shrift for the next
month or two, given all the other things I have going on and some
restrictions around what systems I can play it on, but I'm looking forward
to doing more.
As much as I love the game, I must say that this is exactly why I
generally only game on consoles. I could originally run it on my laptop
with only a few bugs and problems, but then 1.8 landed with lighting
updates and the GL support in my laptop video card was no longer
sufficient. I truly hate having to worry about or even think about this
stuff. With an Xbox 360, you know that if you have a 360 and you buy a
game for the 360, it will work. End of story. I don't like having to
mess with computers when I want to play a game; it's a completely
different mindset for me. (So here's hoping the Xbox 360 version of
Minecraft does materialize.)
But, apart from that, Minecraft really is as good as all that. It's one
of the more amazing games I've seen in terms of providing multiple ways to
play it. And, something that I don't think is advertised nearly enough,
there's a peaceful mode that disables all the combat and means you don't
have to deal with the first-person shooter aspect of things, which is how
I suspect I will always play the game.
I was planning on doing some things on-line during the vacation, but ended
up taking an almost complete break (and a complete break from work apart
from sorting mail). That was exactly what I needed.
2011-09-23: rra-c-util 3.9
This release is primarily a collection of various bug fixes, but it also
starts the process of rearchitecting and improving some of the TAP add-on
modules for testing to make them easier to use.
The replacement for the issetugid function was mistakenly called
issetuidgid and was therefore used on all platforms, not just on platforms
missing the function it was intended to replace. This has been fixed.
The Kerberos TAP add-on now registers its cleanup as an atexit handler so
that the caller doesn't have to explicitly clean up. It also frees all
memory for cleaner valgrind output, and the kinit and shell variants avoid
overwriting the user's AFS token on Heimdal.
There are also minor fixes to the GSS-API include handling, removal of the
non-portable INADDR_LOOPBACK define in the networking code, and use of
typedef instead of #define for socklen_t and sig_atomic_t replacement.
You can get the latest version from the rra-c-util distribution page.
2011-09-23: WebAuth 4.0.1
This is the first bug-fix release for the major 4.0.0 release, and is the
version that we expect to go into production with at Stanford, although we
have not yet done that.
The primary changes are bug fixes, closing some memory leaks and errors in
incorrect inititialization that affected login cancel tokens. This
release also fixes the logging level from mod_webauth when obtaining
additional credentials and logs the return URL of authentications to the
WebKDC.
The new feature in this release is that suspicious login information
returned by the user information service is now displayed by WebLogin on
the confirmation page (and the confirmation page is forced). This will
require updates to any site-local confirm page templates.
You can get the latest version from either the official WebAuth pages or
from my WebAuth distribution page.
2011-09-24: Let's forget that week
Looking out at the bits of fun within the stone of the work week. That's
how I'm feeling this weekend. I had an awful week coming back from
vacation, full of minor and larger annoyances, interruptions,
frustrations, and all sorts of other problems. Thankfully, it's now the
weekend, and I only have to survive another couple of weeks before I can
take another long vacation.
There's almost nothing about project scheduling and communication of
schedules that I think works properly right now. Everything seems much
too aggressive, much too high-pressure, and way too full of arbitrary
artificial deadlines. That's adding a lot to my stress at the moment, but
thankfully once we get through this push, we should be in a deadline-free
zone for a while. (Finally. Hopefully that's actually true this time.)
On the plus side, I got a Kinect for the Xbox 360 and spent a bunch of
time last night and today playing with it. It's a much different video
game experience (and something for a much different mood, I think), but
it's excellent exercise and it's entertaining enough that it's a lot more
fun than regular exercise. So that seems promising for increasing my
overall activity level. Hopefully jumping up and down on my floor won't
bother people below me too much.
All posts