< | >

Hacker's Diary

A rough account of what I did with Emacs recently.

February 28: Revived VMware somewhat through judicious abuse of the registry. I think possibly the ATI taskbar toy or some of the Compaq power management stuff is causing the problem. I'll restore the bits one at a time and see where it dies.

Progress Quest is exceedingly funny and bizarrely compelling.

Micromail site update, in time for our VAT rate increase tomorrow.
February 27: Updated the Ardmore Diving website for my brother. Mostly just new prices, but there were a few other things that needed tweaking as well. And if you're looking for a dive trip in October, check out the site!
February 26: Hrmph. Oracle Developer 2000 has got to be one of the worst development environments I've ever encountered. As I said to one of the guys, it was probably designed and built in 1983 and hasn't changed significantly since then.

ZDNet (and, by linkage, Slashdot) is running a story about how all programming languages are essentially the same and thus there's no choice for developers. Now, aside from the straight-ahead nonsense value of the statement, isn't this exactly what that nice Mr. Turing with his little machines proved all those years ago? Further skimming of even the opening paragraphs of the ZDNet article (I'm not going to link it. It's too lame.) make declarations about J2EE and .Net dominating the programming market, and better still, how VB is procedural while VB.Net is object-oriented. Uhhh. Right. Two things spring to mind here; first, this guy would probably make the same comparison about C and C++, falling into the simple error that object-orientedness is a language feature rather than a programming technique; and second, in terms of looking like an object-supporting language, well, the last time I did any VB hacking (middle of last year) it looked pretty much like an object-supporting language to me. But what do I know? I code for a living, as opposed to having been "a software architect for several start-ups" and "an independant consultant and speaker".

Funnily enough, the Slashdot thread on the article is more informed and so forth than the article itself. This amuses me greatly.
February 25: More hacking on PatchReport today. Also a new server turned up for me to play with, whee!

Damn, it happened again. My pilot-link patch turns out to have been incomplete, in that it was missing the Malsync support. Fortunately, the Internet Wayback Machine had a copy of my old patch with the missing bits on it, so I was able to fix the problem and update my patches page.
February 24: Good grief. Now my hub has died. How do hubs die, exactly? Friction from the packets? Anyway, I went back to CompuStore, which is expensive and poorly stocked, but is also the nearest place to me and open on Sundays, and got myself a 3com 4-port hub. Bah.

Monsters, Inc. is quite a fun movie. The short at the start (For The Birds) and the outtakes in the credits are absolute screamers, though.

February 23

Travelled to CompuStore, PC World, and finally to Maplin to get myself a new fan. Nobody wants to sell Socket 7 kit any more, it appears. I also picked up a monitor cable - again, far more difficult to find than it should have been; everyone seems to stock a million extenders (male/female) and no actual KVM-Monitor connectors (male/male). The one I found was in a shop-soiled box in the middle of a bunch of extenders, as it happens.

February 22

Went digging in Oracle, trying to modify an export-to-CVS script to not require privileged access. This is hard enough without the added bonus of a hangover... still, I managed.

Willy at Micromail sent me a complementary copy of "Homepage Usability: 50 Websites Deconstructed, noting that the Micromail site scores quite highly (preen preen). I've just started looking at the book, and I've already got a few ideas for changes. Plus I've been talking to AjD about the site, and he's suggested one or two things as well.

Tabbed browsing is just the coolest thing ever. Why have I not tried this before? Of course, it would be nice if... you could manually flip to the tab, rather than it popping up immediately, and it would be nice if closing all the tabs except one didn't lose you your entire Back-button history except for the remaining tab, but still. Pretty.

I made it into NTK again, whee! And not in the embarassing way that lead to my first NTK appearance, either.

February 21

Woop. Office night out. BEER!

February 20

I've been hacking on Joe Shamblin's PatchReport on and off for the last week (since d. mentioned it to me) and I've made it (I think) better. Once I've finished the current bout of hacks I'll be posting the modified version back to Joe to see what he thinks.

Cleaned up my source code directory some more, mainly removing an old set of duplicates from the Perl directory.

Did a bit of work on BBDB, continuing to kick some bugs out of the dialing stuff and also making a command-line interface for the testing.

February 19

Swapped out Gonzo's fan for the one in the 486 only to discover that that's also a bit whiny. I guess I'll need to actually purchase a fan, bah humbug.

Smarry mentioned a thing in his LiveJournal that I'd had once before for timesheet tracking: a desktop clock that you can use to monitor how long you spend on a given task. I asked him about it, and while waiting for a reply, I threw one together myself using my Perl Monitor module. Wow. Code reuse in action. I feel, uh, whatever. The new toy is in the workshop area.

February 18

Aie. Gonzo is definitely unwell. Just fell right over. Now, this could be because of my hardware fiddling yesterday, or it could be something wrong with the new kernel. So, I'll try rolling back the kernel to see if it improves things. Of course, I used my RPM cleanup script to wipe out the older kernel files only last night. Bah.

KABOOM! Some poking on the web indicates that this may be a problem with the USB stuff, which you can see features in the crash dump...

Feb 18 14:01:40 gonzo kernel: Unable to handle kernel paging request at virtual address 89cb4000
Feb 18 14:01:40 gonzo kernel:  printing eip:
Feb 18 14:01:40 gonzo kernel: c0112647
Feb 18 14:01:40 gonzo kernel: *pde = 00000000
Feb 18 14:01:40 gonzo kernel: Oops: 0000
Feb 18 14:01:40 gonzo kernel: CPU:    0
Feb 18 14:01:40 gonzo kernel: EIP:    0010:[do_page_fault+359/1200]    Tainted: P 
Feb 18 14:01:40 gonzo kernel: EIP:    0010:[<c0112647>]    Tainted: P 
Feb 18 14:01:40 gonzo kernel: EFLAGS: 00010203
Feb 18 14:01:40 gonzo kernel: EIP is at do_page_fault [kernel] 0x167 
Feb 18 14:01:40 gonzo kernel: eax: c408a000   ebx: c0b228e0   ecx: c408bf19   edx: 0000693d
Feb 18 14:01:40 gonzo kernel: esi: c46979e0   edi: c01124e0   ebp: 40ebd000   esp: c408bf1c
Feb 18 14:01:40 gonzo kernel: ds: 0018   es: 0018   ss: 0018
Feb 18 14:01:40 gonzo kernel: Process setiathome (pid: 6625, stackpage=c408b000)
Feb 18 14:01:40 gonzo kernel: Stack: c099f36a c408a000 c408a000 c099f768 c0166d5d 00030002 00000002 00006300 
Feb 18 14:01:40 gonzo kernel:        c1577a40 00000202 c1577a40 c1577a40 ca87191c c15af620 c0314920 c011cce0 
Feb 18 14:01:40 gonzo kernel:        00000246 c0314540 00000000 c02ff5a0 00000000 c011d2a7 c0314920 c0119a7b 
Feb 18 14:01:40 gonzo kernel: Call Trace: [flush_to_ldisc+221/240] flush_to_ldisc [kernel] 0xdd 
Feb 18 14:01:40 gonzo kernel: Call Trace: [<c0166d5d>] flush_to_ldisc [kernel] 0xdd 
Feb 18 14:01:40 gonzo kernel: [8139too:__insmod_8139too_O/lib/modules/2.4.9-21/kernel/drivers/net/+-18148/96] rh_init_int_timer [usb-uhci] 0x5c 
Feb 18 14:01:40 gonzo kernel: [<ca87191c>] rh_init_int_timer [usb-uhci] 0x5c 
Feb 18 14:01:40 gonzo kernel: [__run_timers+176/224] __run_timers [kernel] 0xb0 
Feb 18 14:01:40 gonzo kernel: [<c011cce0>] __run_timers [kernel] 0xb0 
Feb 18 14:01:40 gonzo kernel: [run_all_timers+23/32] run_all_timers [kernel] 0x17 
Feb 18 14:01:40 gonzo kernel: [<c011d2a7>] run_all_timers [kernel] 0x17 
Feb 18 14:01:40 gonzo kernel: [bh_action+27/80] bh_action [kernel] 0x1b 
Feb 18 14:01:40 gonzo kernel: [<c0119a7b>] bh_action [kernel] 0x1b 
Feb 18 14:01:40 gonzo kernel: [tasklet_hi_action+80/128] tasklet_hi_action [kernel] 0x50 
Feb 18 14:01:40 gonzo kernel: [<c0119990>] tasklet_hi_action [kernel] 0x50 
Feb 18 14:01:40 gonzo kernel: [do_page_fault+0/1200] do_page_fault [kernel] 0x0 
Feb 18 14:01:40 gonzo kernel: [<c01124e0>] do_page_fault [kernel] 0x0 
Feb 18 14:01:40 gonzo kernel: [error_code+56/64] error_code [kernel] 0x38 
Feb 18 14:01:40 gonzo kernel: [<c0107058>] error_code [kernel] 0x38 
Feb 18 14:01:40 gonzo kernel: 
Feb 18 14:01:40 gonzo kernel: 
Feb 18 14:01:40 gonzo kernel: Code: 84 b4 00 00 00 ba 01 00 00 00 eb 0d f7 46 14 05 00 00 00 0f

After several more crashes, I think I've found the source of the problem: the fan I tried to fix. Seems it had simply stopped dead, and the machine subsequently overheats. So the machine is currently running in case-open mode so I can prod the fan if it sticks again. The machine has been up for an hour without crashing, which is better than it was doing earlier.

Discovery #2: I did, in fact, miss some files when upgrading the laptop harddrive, but as previously stated they were on the old drive. Now I'm looking for a front-end of some sort for dump/restore that will allow me to point and click the appropriate files back into place and, well, I'm not finding one. So I've done a little manual reading instead and restored the files I want off the main tree for slow but careful reintegration.

Discovery #3: toys and such for the laptop! There's a page about installing Linux on the Armada M700 at Compaq's website, and there's also a tool for frobbing the M700's LCD brightness. Whee! The latter was linked off a page about Linux on the Armada E500, but much of the stuff seems to apply, so that's cool.

February 17

Came back from Bob's place with a hangover, a Microsoft Sidewinder 3D Pro joystick, and a dual-processer AMD box. That makes eight computers in the room right now, three of which run more-or-less constantly and two more of which are known-bootable. I really need to do something about that...

A little more tweaking and I had enough of a PXE boot working to get me a NFS mount of Klortho's backup files, and whoosh, restore. Followed by a little mad scramble to accomodate the fact that I'd moved the partitions around some. What I need to do now is get a mounting kit for the old drive (so I don't have to keep swapping the mounting kit around) and then set it up with Windows so I can just swap drives around as necessary. It's more trouble than dual booting, but I like the separation.

This is nice:

/dev/hda4             15559588   3485612  11283580  24% /

Giggle value: the files for the PXE boot are named with variations on "honk" and "oof". I just need an appropriate bootup graphic now... (hi AjD!)

HMM. I suspect I may have missed one or two files in the changeover. But they're still on the old drive, so I can sort that out easily enough.

Gonzo's fan is whining again. Problem solved by dismantling, cleaning, and remantling. I took the opportunity to clean up some of the cabling while I was rooting around, and it promptly rewound itself into a different sort of mess.

Hrm. Gonzo booted with a newer kernel as a result of the reboot, and SETI@HOME has crashed twice with some sort of memory allocation error. In fact, it's oopsed.

February 16: Fitted the new drive, then started playing with PXE which is enormously fun and I wish I'd investigated it before. Didn't quite manage to make a useful PXE recovery setup, since I had to abandon it all in favour of a party at Bob's place.
February 15: New computer bits arrived. 20GB hard drive for Klortho, and a Hauppage USB TV device. Things I should have researched before buying the latter: 1. Framerate; 2. Linuxability. The framerate isn't bad, but it's occasionally annoying - especially when the station isn't properly tuned and the screen is thus a complex mass of uncompressible snow. The Linuxability, alas, seems to be near nil at present due to sneaky, er, proprietary compression stuff. Gah.
February 14: Happy Valentine's Day. Bah humbug.

Micromail update today.

You know, the Microsoft bashing on the ILUG mailing list is really, really tiresome. I mean, sure, I have my own little bitches at them, but it's directed, not just arbitrary baseless comments for the sake of appearing to be one of the cognoscenti or something. Really, guys. Grow up.
February 13: YAY! Laptop returns! All healthy and so forth.

Finally I have defeated Reports in mortal combat. In fact, I've done a whole lot of really neat stuff that I can't tell you about because people get upset when I mention company-related stuff on my website. So you'll just have to imagine I'm describing a pretty tidy hack here.
February 12: Still not got the Reports server sorted. This isn't software; this is torture for sysadmins. Also, playing with Locales is amusing, for very, very, very small values of amusing and large values of LART! LART! LART!
February 11: Wrestling with Oracle Reports server in the office. It is not a pleasant piece of work. With the printing and the X and the screaming and the bwahah hey hey.
February 10: I really suck at X-Wing, btw.

It's kinda amusing watching Outlook trying to preview a 5MB httpd access log. Especially the way the whole interface pretty much locks up. Also, IE keeps jumping to the front. What's that about? If i put a window on top, I expect it to stay there, modulo modal dialog boxes from other applications.

I collect the strangest links sometimes... someone on LiveJournal has picked up one of my FROST.BOB II pix for some reason. And I've found one of my party pictures over on FARK.

February 9: Figured out what was up with Eudora: it was trying to download my entire website as a collection of mail folders. Uh. Not particularly clever, that. Especially not displaying any indication that that was what it was doing.

More X-Wing. Found my Tie-Fighter boot disk, which solved the EMS-vs-SoundBlaster problems. Notably, it's a DOS 6.22 disk, rather than a Windows recovery disk.
February 8: Rediscovered X-Wing. Spent some time trying to make an appropriate boot disk (remember those?). Also relocated Novastorm, but it keeps freezing up and otherwise flaking out.

For some reason, EMM386 (remember that?) kills the SoundBlaster. I tried swapping cards around in case it was a hardware problem, but no dice. And where did I get three SoundBlaster cards, anyway?

Installed Eudora 5.1 to see if I'd recommend it to my folks. Set up IMAP. Pointed it at the mail server. It promptly hung itself up, and eventually I killed it, after which it sulked. Even after a reboot. So, sorry, no dice.
February 7: I've been playing with the leftover bits of my brother's old P133 (well, it used be mine before it was his) and I've now got a Windows 95 OSR2 box on which to browse. It's not really healthy enough to do any hacking with, though.
February 6: And today, Klortho went back to the shop for... the fifth time. This time, I've included a sheet of paper on which I've described the fault exactly, and also included a floppy disk with an image which shows the fault. Belt & Braces approach.
February 5: You know, Oracle Application Server 9i is a big nasty hack. But I know a lot more about it as a result of some trial-and-error work today.

Also found out a bit more about Emulex Lightpulse cards. Having a "toy" Sun box is quite useful.

Mozilla 0.9.8 is out. I wonder if they've fixed the "can't delete a folder via IMAP" bug yet? Only one way to find out, I guess...

Had a look at x0rfbserver today, and was disappointed to discover that (a) it wants me to install a Windows-lookalike library, (b) it's written in C++, and (c) it will require abuse to compile on Solaris, which is exactly where I'd like to use it. Gack.

Did some work on BBDB's phone-dialing stuff. I need to figure out a useful way of testing it, though.
February 4: Waiting for the UPS express delivery goop to show up so I can ship the laptop back. Since my employer's office is out in the boonies, "express delivery" means they get it to you by three o'clock the next working day.

Installed XEmacs again to look for a few bugs in BBDB. I can't recall uninstalling it; must have been when I was cleaning up the mess last month.
February 3: The crossword toy is still running. Evidently I need some more rules for it.

February 2: Hacked up a thing to try arranging spaces in a crossword grid, given the list of clues and the length of each answer. Caw! Squawk!

Spent the evening dining, wining, and watching High Fidelity again.
February 1: Laptop returns. Accompanied by a note saying that no fault was found. Hello, Tri-Tec, you are IDIOTS. Phoned Compaq support again and explained the problem. Compaq issued a third RMA, and have promised to monitor the process this time to make sure I get the laptop repaired.

Hmm. pptp works fine on the laptop but I can't figure out what exactly I did to make it work, since it's not working on Gonzo.

previous month | current month| next month

Waider

Hey! Who stole the end of the month?