I tracked this down to it sending Expect: 100-continue in the headers of the request it sent to its server. However, it was going via our web proxy at work, which is Squid. Recently I upgraded Squid to 2.7, and this seems to be the cause of the problems. Squid 2.7 is less forgiving about that particular header (which is apparently an RFC violation on the part of the client).

It’s easily fixed by adding ignore_expect_100 on to your Squid configuration.

tail somefile | ccze -A -o noscroll | grep ...

-A makes ccze output ASCII sequences, and -o noscroll turns off the scroll option – whatever that is exactly. The man page says it’s a good idea to turn it off if you’re redirecting the output. Omitting it seems to make the output appear in lumps :)

For example, keep an eye on squid logs omitting all MSN crap:

tail -F access.log | ccze -A -C -o noscroll | grep -v gateway.dll

This goes in your zshrc:

  vman() {
    if [ $# -eq 0 ]; then
      /usr/bin/man
    else
      if man -w $* >/dev/null 2>/dev/null
      then
        /usr/bin/man $* | col -b | vim -c 'set ft=man nomod' -
      else
        echo No man page for $*
      fi
    fi
  }
  alias man='vman'

And this goes in your vimrc (in a suitable augroup):

  autocmd FileType man setlocal ro nonumber nolist fdm=indent fdn=2 sw=4 foldlevel=2 | nmap q :quit

Now you’ll get nicely highlighted man pages with folded sections, and you can hit q to quit just like less.

The only problem I encountered was getting it set up to work with multiple networks in irssi. Configuring ZNC itself to put me on many networks was easy enough, and when I connected irssi to it for each network using /connect it worked fine. Using /server add to set up irssi’s server list didn’t work out so well.

Every time I tried to add another server (for connecting to another network), irssi would remove the first one. This is because irssi only considers a server’s hostname and port when trying to work out if it’s unique. When I added another server which differed from the first only by password and associated network, irssi assumed I was updating the existing one and changed it. Putting the required entries in irssi’s config by hand also didn’t work.

The solution was easy enough; one ZNC instance can listen on multiple ports, so I got it to listen on a selection of ports using multiple Listen directives in its config file, and then added each network’s server to irssi using a different port number.

Now I can take advantage of cool things like buffer replay and detaching from channels. Sorry, irssi-proxy – you don’t quite cut it now :)

Sit no more! Just hit Ctrl-t and fsck will reveal its secrets. (This sends the currently running process a SIGINFO.)

...
generalInterfaceTiledExamples.o(.text+0x36f): In function `writeTiled1(char const*, Imf::Array2D&, int, int, int, int)':
: undefined reference to `Imf::Channel::Channel(Imf::PixelType, int, int, bool)'
generalInterfaceTiledExamples.o(.text+0x3b3): more undefined references to `Imf::Channel::Channel(Imf::PixelType, int, int, bool)' follow
gmake[1]: *** [imfexamples] Error 1
gmake[1]: Leaving directory `/usr/work/usr/ports/graphics/OpenEXR/work/openexr-1.6.0/IlmImfExamples'
gmake: *** [all-recursive] Error 1
*** Error code 2

Remembering that in UPDATING there was something about this, I had another look:

20071008:
  AFFECTS: users of graphics/OpenEXR and graphics/ilmbase
  AUTHOR: nork@FreeBSD.org
...
  # pkg_delete -f OpenEXR
  # pkg_delete -f ilmbase       (if you already installed before CONFLICT)
  # cd /usr/ports/graphics/OpenEXR
  # make install
  # pkgdb -F

Nope, that didn’t help. pkg_delete told me ilmbase wasn’t installed. A quick Google turned up a PR about this, but it was closed with “look at UPDATING.” so that didn’t help any more.

I fixed it by rebuilding graphics/ilmbase anyway. Once that was rebuild and reinstalled, OpenEXR built fine. No idea why pkg_delete didn’t think it was installed.

A quick Google search this time lead me to this post which gives the answer:

cp /usr/local/lib/thunderbird/defaults/isp/rss.rdf /usr/local/lib/thunderbird/isp/rss.rdf

and suddenly it works! :hellyeah:

On FreeBSD, this doesn’t work so well. However, you can tell manually tell the Crucial site what motherboard you have and it will tell you exactly what sort of RAM is best.

This is fine if you know what motherboard you have.

if you don’t, the answer is dmidecode(8) (/usr/ports/sysutils/dmidecode). This utility grabs the same information which the Crucial thingy does, so all you need to do is find the info you need.

This is pretty easy:

# dmidecode -t baseboard
SMBIOS 2.3 present.

Handle 0x0002, DMI type 2, 8 bytes
Base Board Information
        Manufacturer: ABIT <http ://www.abit.com.tw>
        Product Name: 694X-686B (VP6)
        Version: v1.0 ~
        Serial Number:

Not masses of useful information returned from my old system, but still enough to identify the motherboard as an Abit VP6. This is all I need to know to find out what memory I should be buying.

I already know what size sticks of RAM I have, but just incase you need to check:

# dmidecode -t memory
SMBIOS 2.3 present.
...
Handle 0x0021, DMI type 17, 21 bytes
Memory Device
    Array Handle: 0x0020
    Error Information Handle: Not Provided
    Total Width: Unknown
    Data Width: Unknown
    Size: 512 MB
    Form Factor: DIMM
    Set: None
    Locator: BANK_0
    Bank Locator: Bank0/1
    Type: Unknown
    Type Detail: None

Again, not the most useful information ever seen, but enough to tell me I have a 512MB stick in bank 0. (On my newer desktop, I get much more useful information like the speed and voltage of the RAM.)

1. Released bMotion 0.1.0 – only about 3 years after I planned on it originally. 0.1.0 is greatly improved over the previous version, and a few weeks ago I finally decided I should stop pointless tinkering with it and get a stable releasable version :)

2. Finished Zelda: Phantom Hourglass on the DS – mainly in celebration of releasing bMotion.

3. Started using my Simple Monitor project at home and on my colo, and it’s working well so far. The only problem was that both machines needed Python updating to support the native sqlite class. Once that was fixed, both worked fine and are now monitoring their services. Once I get some other features I’m planning on written in, I’ll release it for other people to try.

In FreeBSD, clicking on the addressbar doesn’t do that. However, when you manually select the URL so you can replace it, it overwrites the contents of your clipboard and makes it so much harder to paste the URL you were about to look at.

Luckily, you can fix this :) Browse to about:config and stick urlbar in the filter box. Find the browser.urlbar.clickSelectsAll setting and change it (by double-clicking) to true.

To paste, you’ll probably have best success with Ctrl-V as middlemouse seems to paste the clipboard without overwriting the previous contents.