Hacker's Diary

A rough account of I did with Emacs recently.

April 30

I think I may have to reinstall Java on my home box.

At home, trying to back up the laptop so I can send it for repairs. I spent an hour poking and prodding at rsh and rsh-server until I got fed up and used a script wrapped around rexec instead. This stuff used be easy, you know.

April 29

Gonzo crashed rather hard this morning, and then played silly beggars when I tried to restart. So I'm currently upgrading to a 2.4.4 kernel and poking various bits of the system.

Bleah. NFS is somehow hosed on the new kernel. Wonder what's up with that?

Tweaked my property searching script to iterate over more than one page of search results. Still pretty gross, though.

April 28

Morrisa and nj visited. Much carousing and the like.

April 27

Spent the morning setting up a Linux box on my desk as a "monitoring station". *snork*

April 26

It appears that my role is morphing from admin to developer due to the lack of present requirements for the former by comparison to the latter. Java, JSP, Oracle. Looks like I'll be increasing my buzzword compliance.

April 25

Tracked down the silliness that was causing a database to be offline every morning. A bug in my bugfix, doh.

April 24

Yay, we have an Oracle DBA to play with now. Should make it easier to track down performance issues; that, or we can stand there pointing at each other going, "it's your problem!"

April 23

Tried out SyncBBDB for the first time. Observations: (1) BBDB gets REALLY SLOW when you have a lot of records listed and the GUI stuff is turned on; (2) I have way too many junk records in my BBDB; (3) I'm sure the matching could be smarter, and some of the data could be better accomodated. I'm also sure I could help integrate this better by doing a bit of work on the elisp end of things. After some mucking about I decided to revert to my old databases and try SyncBBDB again when I'm thinking more in a straight line.

April 22

Put Cygwin onto Klortho to try and debug some problems with building BBDB under Windows; a similar bug seems to affect BSD make builds, funnily enough.

Whee! My ID3lib patch for Grip has been accepted.

Hacked on the EMU10K1 driver some more. Still trying to get /dev/sequencer support to cooperate with the mainstream driver. The EMU guys, meanwhile, are trying to get some patches into the kernel.

AAAAAAAAAAAAAAAAAAAAAAAAAAAH. VMware is STILL taking down X Windows.

April 21

Picked up a USB<->serial convertor more-or-less on a whim. Now I need to find out how to make it work.

April 20

Phonecalls (3) from office on my first day off work.

April 19

JC's bon voyage drinks in Messrs. Maguire pretty much soaked up my evening. "soaked" probably being a pretty good choice of word.

April 18

Cheese-o-rama camera driver from the net doesn't work either. Yet VMware can see it. Blag. Something to hack some other time, if at all.

Poked at the /dev/sequencer stuff for the EMU10K driver and sent a patch to the CVS guy.

April 17

Recoded the NTLM stuff for my proxy-proxy in perl, so now it's an all-perl thing. Of course, the DES encryption bit is still through an add-on library, so I should see about doing a DES implementation in straight perl.

Had a quick poke through the driver source for the cheese-o-cam and determined that I'd hacked it to pieces and didn't have a spare copy. Time to redownload, I guess.

April 16

More tooling with BBDB - a BSD compatibility fix, of all things.

Grabbed and built the new xscreensaver, then tried to figure out why the rpm was missing most of the stuff from the driver directory. Eventually discovered that it was due to make -s. Also, %config(missingok) doesn't work like you think it should, at least when building rpm files...

April 15

Somewhat wacky San Marino Grand Prix with the Ferraris performing pretty badly (ok, so Ruebens got a podium finish, but really now..) and Ralf Schumacher getting his first win ever.

Hacked on BBDB some, since I've been neglecting it a little of late.

April 14

Hacked up a script to manage my kernel configurations across three machines so that I don't have to keep running xconfig or menuconfig.

Hmm. The formerly-mentioned problem with ssh2 port-forwarding dying after one use seems to have gone away. One less bug to work around. Unforunately I still can't seem to make the NNTP hack "just work" by putting the forwarding command into /etc/inetd.conf. I'm still missing something, I guess.

I've said it before and I'll say it again: after you've checked the patch with patch --dry-run, don't forget to drop the --dry-run parameter. Especially if you're doing a kernel build.

Finally got around to fixing my random .sig toy. It now pulls stuff straight from the quotes file, retrying if it finds more than 4 lines, and inserts my mail address and tagline if there's room. Much better than the old one!

Quickie: I patched the current CVS XMMS to make the specfile a bit better. Grab the XMMS specfile patch and play with it.

Tried out an all-new XMMS and Icecast setup, and I'm still getting the choppiness problem. Seems like it hit jwz on the DNA Lounge stream recently, but I'm guessing from comments elsewhere that he's gotten around it. I wonder what the hell it is about my particular setup that's causing it?

Hacked a little more on a Perl script I had started for generating a spec file from an rpm, either installed or sitting on your disk. Used that to convince up2date that I'm running a newer kernel than the one it recommends I download. Imperfect, but workable.

April 13

Finally got the NTLM proxy working with a combination of Perl and C. I need to see about converting the C stuff to Perl (it's the bit that does the DES encryption) so that I'm not passing around data in a snoopable fashion. Also I need to make the thing a bit more secure.

More RPM upgrade madness. You'd think I'd know better by now.

Finally nuked the last traces of Windows from Gonzo. Whee! Of course, I'm going to have to turn Klortho into a Windows-only machine before I send it in for repairs, which means putting Klortho's existing setup onto Gonzo. Bletch.

April 12

Spent some time playing around with code to do NTLM proxy authentication. Perl has me spoiled; I can't code dynamic memory stuff and pointer arithmetic any more without drawing diagrams and thinking a lot!

I broke Redhat's up2date toy last week by selecting a package for download that is no longer available. Since I've not had a reply from them to my bug report, I hacked around the problem at source instead. Man, but Python is one horrific language.

Went on another RPM updating rampage.

April 11

Happy Birthday To Me.

April 10

Louise has arranged for Compaq to have a look at my laptop with a view to repairing my dead IrDA port, so that's good.

April 9

Klortho continues not to go clicky-clicky, so that's cool.

Did a big upgrade session on Gonzo with GnoRPM, causing me to gnash and wail some and eventually add a bit of a rant to my wishlist page. It's all of this damned worse-is-better stuff.

April 8

Klortho seems pretty happy with the 2.2.18 kernel, and I can use the parallel port in VMware now - which means I can play with the cheesy parallel-port camera again. I wonder if the half-assed Linux driver code I have works on 2.2.18?

I broke RHN, the Redhat Network. More specifically, I can't run the damned thing on Klortho until they clear my download queue on the Redhat end, because there's a package queued that no longer exists, and the daemon is too stupid to cope. I could hack it, but that'd mean touching Python. Ick. No thank you. I've not yet recovered from the last time I tried dealing with the damn thing.

April 7

Rolled Klortho back to 2.2.18 to see if it stops the "hard drive go clicky" problem. Then I noticed that 2.2.19 had been released. Yeek.

Bought "PI" on DVD, largely due to peer pressure from Bob - who will, of course, deny everything under questioning. Not a bad movie. Not sure I'd have bought it without the peer pressure, though...

Installed netpbm-nonfree on the DSP server so I could go back to my nice, small GIF images for snorq.

April 6

Started installing toys on the Solaris boxen.

April 5

Did a little more with the Outlook hackery, and at this point I'm convinced that the only way I'll get it to behave exactly right is if I write my own Reply code from scratch. Yeesh.

April 4

Managed to hack Outlook into some semblance of reasonable behaviour today, but the API is still branedead. Like, for example, if you set the Body (text) of a MailItem, it reverts from whatever type it was to "the Default Editor Type" which despite my best efforts appears to be Rich Text. GAH.

Got another Solaris box to look after today, whee!

April 3

I guess my Windows machine wasn't "installed properly" since the admins' quest to get me Internet access today culminated in them reinstalling the system from scratch.

Before this happened, I was poking at the Visual Basic stuff for Outlook to try and make the Reply behaviour more sane. Turns out that the API is about as branedead as the application itself, in that hooking events on messages pretty much requires you to hook every single message in the mailbox - as best I can tell, since the documentation is also a bit slack.

Caught Klortho in the act of doing the skippy disk thing. Seems that the drive just gets stuck in a loop and then everything dies because of disk i/o being unavailable. Worryingly, reading one of the RHAT developer diaries on Advogato, I found a comment along the lines of "we can't figure out how any of the 2.4 kernels work" in reference to some apparently intractable bugs in the VM stuff.

April 2

First day in the new job.

Quote for the day: "Microsoft products work perfectly well as long as they're installed properly."

...which I've managed to disprove at least once since it was said to me, but heck, probably my anti-Microsoft bigotons got into my k0mp00ter...

Found a very silly bug in snorq that stopped it from snarfing Bobbins.

April 1

Site update for Micromail pretty much all for the day.

Discovered that my VMware bug is still alive and well. The hard way. Goddammit why does this always have to happen when I've an unsaved emacs session running somewhere?

previous month | current month | next month