Did you know stuff happened today while you weren't watching?

This terminal was beckoning...and I'm at a lull. Figure I'll do something. Let's write about nothing.

ALL of yesterday was spent setting up zaphod. I mean all of it. It was maddening. In any case, it is alive. At least the windows head of it is (it will also have a linux head, representing the two heads of the humorously-fated Zaphod). It will probably get linux shot at it this weekend. It also needs more RAM. Any RAM. It's only got 16mb. I think I will take some out of my NetWare server...which seems to have stolen a whopping 40mb. I'm still not sure where it got that. It doesn't need more than 16. zaphod is running on a Cyrix 6x86 P150+ (running at 60x2.0=120mhz). 8gb drive (not sure where that came from). And the shittiest CD drive I've ever had. The actual details of yesterday are unimaginably ludicrous and I'm still figuring the probabilistics on all those events occuring within the same 12hr period.

The first half of today was spent installing VNC and VisualDSP on zaphod, and trying out some minor details of the 2181 ICE setup. Pete is aparently coming over later this afternoon. Hopefully it will be a more productive visit than yesterday's...

You know the one day I actually come out and say that its going to be a boring day, it would turn out to be a rather interesting one.

And a good chunk of Friday was indeed quite boring. That is, sans a few hours at the end; the part that involved people I'd never met before unexpectelty arriving at my door and hiring me.

So after they left, I went to bed eventually. And then got up at 0700 for no aparent reason. After sitting around being blatently unstimulated, Brock came over for yet another unaparent reason. Watched MP's Life of Brian. He had yet to see that. I think I'm still the only one that thinks its better than Grail. Oh well. He left. Then those strangers came back again. They kicked themselves out roughly four hours later. I had not yet had a nap at that point and decided 0100 would be a mighty decent time to take one.

Then got up at 0700 again. I had at least a half-reasonable excuse for it this time. Annika got baptised this morning (please be careful with that link...its got a load of largish pictures that are detrimental to my link, and a load of obsenely cute captions to go with them (that I had no part in writing)). That mess ended at noonish. Had a large family lunch (which was, interestingly enough, the first time I've eaten in three days...I keep forgetting to do that). They then had a "baby shower" in the living room. I wouldn't know about that. I avoided it. (Aparently Brock witnessed part of it.) [Living rooms are an interesting place to shower babies, but I suppose she needed one badly enough after being dowsed with all that "clean but holy" water.] Anyway. Brock and David (who has remarkably returned just as mysteriously as he departed...I didn't know he left, nor did I know he was home...so why did anyone tell me?) came over with Undead. Undead is a rather interesting machine. Not so much because its wholly remarkable, but more interesting in how cheap it was. Anyway, it had some jumper settings that were rather inconsisent with what it was aparently doing, along with needed some serial port work. [It's worth noting that just after I wrote that, Brock and David entered, sat down, and promptly killed my 8227 (disconnecting me from writing this). They just left, after watching a bit of Star Trek Four (on my Big Project console)and picking up David's box. I also took a few moments to attempt to remember how to bind the RPL NLM to the net card in NetWare 4, since I had neglected to put that in my autoexec.ncf, and somehow the server was accidently killed sometime today. Okay, back out of the brackets...] That machine did get running. We ended up putting the 233mhz MMX chip at 75x3.0, since it doesnt seem to have a valid x3.5 setting (proper 233 is 66x3.5). Probably could have figured it out. But I dunno. Might even be able to push it up to 83x3.0=249MHz again (it was running that way for a while...but it started doing some wacky filesystem stuff...but that may have been related to mixing SIMMs and DIMMs, which is aparently a Bad Thing for this board).

So after that I attempted to sleep for a bit. Failed. Watched Simpsons (the one with the hilarious Make-up Gun sketch). Got pizza. Attempted the sleep thing again. Failed again (damn distractions). Brock and David came back. They left. That leaves me here. Watching the second half of ST4.

It's been a weird weekend. Just plain odd.

Television commercials are far more disturbing with the sound off...

Hrm. Well, it appears I haven't written in a bit.

I really haven't been doing much. Tuesday, of course, I got my 8227 booted. And then made more failed attempts at the RPL protocol. Wednesday was spent mucking with NetWare (4.1...mmmm...NDS...). This was because NetWare also does RPL. And its slightly different. But it hasn't helped me much. We'll see. No good protocol info yet. I should really find out what RPL is based on (I'm told its a NetBIOS protocol, but I can't find any verification of that). Thursday was really a useless day. Mostly read mail and browsed around. Though I played with LISP a bit. That was fun. Oh, and I also installed LiteStep on the Win95 part of my laptop. Looks neat. Also have WindowPictures installed, which is just like gtk themes for windows. So my Win95 feels a lot like E now. But still no sloppy focus. And it has the habit of randomly losing windows... I also tried to take the Compaq survey they asked me to do. But it never worked for me. But they did give me the $30 Amazon gift certificate anyway, just for putting up with their brokenness. Also, diz was watching Special Edition Star Wars (actually Empire). And as usual, he piped the closed captioning into linuxnet #starwars... Amusing to read. "Beep beep blip."

I was watching CNBC the other day and they had Alan Hassenfeld (head of Hasboro toys). He says that they're going to start producing a new Furby. No, not just any old Furby. But a Yoda Furby. With Yoda ears that move up and down.

Today isn't looking to be much more productive than yesterday. Oh well.

Hrm.

They all came home on Sunday. Had a big lunch. Watched The Fisher King (good movie). Then Pirates of Silicon Valley (so-so movie...I wish it would have been less fictional...the real story is far more interesting than the one they mostly made up). Then Simpsons, et al. After talking to Sara for a bit, I finally WENT TO BED (at 2200).

Got up around 0500 Mon. Dinked around for a bit and then got Jeremie's 8227 dump. Then spent an hour or so figuring that out. Pretty neat. The problem occurs when I actually try to make MY 8227 listen to ME. It doesn't. Either I'm still not getting the ethernet/SAP headers correct, or there's something in the Source MAC field that the server sends out that is particularly odd. In fact, there is. Later in the dump, the 4006 LTD type specifies the source MAC directly, where in the first LTD it sends out, its somewhat mutilated ({0x00, 0x20, 0xaf, 0xdc, 0x56, 0xd2} into {0x00, 0x04, 0xf5, 0x3b, 0x6a, 0x4b}). I don't get it.

I'm going to have to write my own calculator here soon...I just can't manipulate multiple bases as freely as I want/need to in every calc I've tried (the best I've found is bc, which has got to say something). Or maybe I should just use a piece of paper...

Anyway. Not much else happened yesterday. Mike's parents left for home. I fell asleep about 1600. Got up around 0030 today. Talked to Tara and Sara for a while. About 0400, I went outside and worked on AIMd code. Added some comments and thought about implementation for a while. Started on the 'master buddy list' structs. We'll see where that goes. Came in about 0515 when I ran out of batteries. It was a lovely morning...until the sun came up.

Well, not much going on here. I'll supply you with a few more bits of humor from linux-kernel...

It's [being totally wrong] happened before, it will happen again. However, you had better come up with a better argument before I believe it happened this time. ... Ehh.. In real operating systems, we call that event a "system call". No message necessary or implied, unless you want to call the notion of switching privilege domains "messages" (and some people do: they call them messages just to prove that messages are as fast as system calls. In logic, that's equivalent to proving that liver tastes as good as ice cream by calling ice cream liver, and is in real life called "lying"). -- Linus

It's not unethical to do something that you're permitted to do with free software, even if you're simply using it as a gun to shoot yourself in the foot with. -- David Parsons

Things are never so simple when technology and fanaticism mix. --Marcus Ranum

Someone should make a 'Quotable l-k' database full of all these hilarious and touching tidbits...

Okay, so I was going to shut up for the day. But I can't contain my gittiness now. I just booted my 8227!! Granted, it wasn't from my own rpld, but I did get it booted! [Imagine me dancing around the garage now with laughing like an overly deranged madman...] Aparently generic NE2000's can't be in RPL servers (probably broken promiscuous/multicast mode). But DEC21143's can. I dug up an NDIS driver for the thing and booted my dad's machine with it using the same configuration as I was using with the NE2k. Works great. The 8227 lights flicker for a bit while it transfers the 360k image, a few moments of silence, and then the thing comes flickering back to life. And it powers on the PCMCIA card too. Spectacular sight to see after you've been going mad from using WinGate for this many months. Okay, after I calm down, I'll go back and get tcpdumps and continue writing my own rpld. LANtastic is still the Ugly Solution. I need to reconfigure the disk image as well...I have no idea what Jeremie named his wireless cell...

I've been going through a bit of the classiccomp list. They're discussing Pirates. Very amusing. "...portrays Steve Jobs as an acid-dropping, drug-pushing, commune-living, personal-conviction-abandoning, illegitimate-daughter-fathering, beyond-workaholic, universally-disliked-but-adored, employee-abusing, god-complex-nurturing, business disrupting idiot whom one expects to have his employees drink hemlock-ridden Kool-Aid at any moment. Oh, and Wozniak is portrayed as a small, fuzzy bunny. ... Steve Ballmer, however, is played by Krusty the Klown from the Simpsons." (Kai Kaltenbach, yes a Microsoft employee)

Okay, back to work... I think there are microscopic knats eating me...I keep feeling little pricks, but now signs of anything awry... Maybe I'm just going insane...again...

It's a bit later now (something like 1430). No success with my frame building. But it still boots nicely off of LANtastic. I also reconfigured my laptop and got rid of all that proxy rubbish. I can pull 17.1K/sec from deepthought to pileus (in windows on wireless across the 8227). I don't really remember what I was getting before. But that's quite good. Though I've heard people say they can get 300Kbit/sec across these IBM wireless cells... which would be 338K/sec... Oh well. It was such a glorifying process to tear down that WinGate machine. Reminded me of the day I took down my last Windows machine... Anyhow. I haven't given up on that protocol yet. I do still plan on getting it work -- even having to use LANtastic is bad (in fact, running DOS on 8227 is pretty bad, but there's nothing I can do about that until we have linux drivers for the wireless cards). I was also poking around in the 8227 a bit and figured out most of the connections for the internal connectors. I'm guessing you can patch on any random ISA card, with a bit of wiring trickery. At least the first 10 ISA address and data lines are exposed. That would be enough to do a floppy controller or a text-only video card (which would be rather useful in examing boot up messages, which could help me figure out why I can't get any bootrom other than the RPL one to work...that is still an option, btw...just to get rid of RPL all together).

Anyway. One less 24x7 machine in my room. Down to two now (mediocris and harmless, the latter of which may get downgraded to a needs-only basis if it's hard drive keeps making all that racket). I'm off to bed.

Hrm. Not much has really been going on. Spent the second half of Friday in the hospital as well. Over a dozen people were there as well. It was rather crowded. Mike went and got pizza and it was quite an interesting gathering. Saturday was more family stuff. Last night was spent over here. I went to bed a bit early...I've been really tired lately. Got up about 0200 this morning (Sun). Haven't done much of anything. They all went to church to show off Annika.

You'll notice a new directory appeared over at ftp.auk.cx.../pub/aim/aimd. Yes. An AIM server daemon. I've been working on it in my spare time. It's not very useful. My goal is not a full AIM server. I'm not sure why anyone would want one of those anyway. All I want is something I can test AIM clients with without actually connecting to the AOL servers. There's various reasons this is disirable. The big two are: 1) AOL limits the rate at which you can login (which is really annoying) and 2) being able to work with AIM code without Internet connectivity (which I still do and will do until I get a Linux-supported wireless setup). I wouldn't bother trying to use it. If you do try, you'll need to change the IP numbers in bos.c to match your server's IP. If I were to write a real client, I would thread it better and allow the authorizer and BOS to be on different machines. This version is actually threaded, but not very well. Needs some garbage collection. And multiple simultaneous users is not tested. I think it works, but it may not. It has the potential to be at least a half-assed server. Full functionality would require a better infrastructure from the start I think. It doesn't do any user authentication, status notification, and does not relay IMs. That means no buddy lists either. It has enough login functionality to let naim log in. However, AIM 2.0 for win95 will still not login to it. It sends a 0x0001/0x0017 SNAC and stalls. But if that client logs into the AOL server, it doesn't stall, but continues sending more SNACs. Weird. That is my other goal...to get Win95 clients logging into it. That seems a bit far off at this point.

This has been an odd weekend. My mail flow is down dramatically, no one around. Weird. Oh well. I've run out of things to talk about.

Some of Linus' diatribe of the day... In short: message passing as the fundamental operation of the OS is just an excercise in computer science masturbation. It may feel good, but you don't actually get anything DONE. Nobody has ever shown that it made sense in the real world.

Well, it's over (or begun, depending on your view). Annika, borne of Sara and Mike, came at 1157 this morning. 8lbs 8.3oz, 22in. She's beautiful, of course.

It all happened rather quickly. We went in about 0530. Heavy contractions started about 1030 at 6cm. Head came out about 1150. And the rest came out shortly thereafter. Incredible thing to watch. Unexplainable. Anyway. I came home a few minutes ago. To go from no contractions to delivery in less than six hours. Beats the average. Oh, she did it with no medication. She handled it very well. My mom got a blood blister on one of her fingers from the pressure of Sara's grip. I was on the other hand and am unscaved. I am going to take a short nap and go back down. She's in for 48hrs (she doesn't really need it, but the insurance covers it, so she might as well).

Well, here I am at 0200 attempting to calm down a bit. In case you hadn't taken notice, it's been a bit stressful around these parts. Lots of stuff happening. Yesterday, Sara's baby's father's parents and him arrived (from their lookout point in Kansas). I hate relatives. Especially ones not related to me. We're all heading down to the hospital at 0600 (ie, in a few hours). Sara will be induced shortly thereafter. (Of course, that has no bearning on when the baby will come. Refering to that kid's belligerence towards its birth up to this point, it's probably going to be a while.) Anyway... I'll just sit here and eating string cheese and drinking flat citra...

Yesterday I got up around midnight I think. I don't right remember. I decided after doing much of nothingness that I should try and get the 8227 going. First I tried the LANtastic stuff that Jeremie sent. No good there. Does the same thing my LANtastic setup did....just sits there like nothings happening. I am beginning to think something is wrong with my 8227. I asked Jeremie to get me some tcpdumps. Maybe I can figure out whats wrong. Or if nothing is wrong, at least write something that can let me RPL the cursed thing. [I have no idea how to do LLC-layer stuff from userspace in linux...probably should look into that. I'm not sure how much of the IBM LLC Jay has layed down in his SNA project. Maybe it could help. Maybe not. I'm not an expert on these things...] So anyway, I gave up on DOS. So I decided to try Slowaris (2.6 for Intel). Used my dad's box. It doesn't like AGP video cards. Nor does it like DEC ethernet cards. Nor does it like ISA PnP sound or SCSI. So that left me with only two cards: a #9GXE64 PCI video card and a generic NE2000 nic. (It's worth noting that it took me several hours to figure all this out.) So I finally got it booted (oh, it's X server can't handle 15bit or higher color either, contrary to what it says -- it can only do 8bit). The actual bit with rpld only took a few minutes. A few minutes to find out it doesn't work either. Get no response at all. Just like nothing is going on. Meanwhile, the 8227 is sitting there spewing the same LLC frame over and over again, trying to get something to listen to its screaming....

So by the time I got that far, I decided it was time to get to bed. Took a variety of short naps between noon and 0100 today. Much interruption in between, of course. Oh well. So anyway, you know where I'll be. Thunderbird Hospital.

While adding the Quotable Homer fortune file to deepthought's fortune, I decided to just upgrade fortune as well (since I included the "potentially offending fortunes database" (aka, any fortune that has the word 'penis' or eludes to any word sounding like 'penis', rhyming or otherwise) when I installed it a few months ago, I figured I should use a version of fortune that has the -o option that goes with the offending lines). Anyway... I was reading through the original Berkeley release of fortune and its README is rather hilarious. In reference to my definition of offensive above, it expands to "...containing any explicit language (see George Carlin's recent updated list)... Political and religious opinions are often sequestered in the potentially offensive section as well. Anything that assumes a world view...should not be in either, since they are not really funny unless *you* are racist, mysogynist, or homophobic...people should have a reasonable expectation that, should they just run fortune, they will not be offended. We know that some people take offense at anything, but normal people do have opinions, too, and have a right not to have their sensibilities offended by running a program which is supposed to be entertaining. People who run 'fortune -o' or 'fortune -a' are saying, in effect, that they are willing to have their sensibilities tweaked. However, they should not have their personal worth seriously (ie, not in jest) assaulted. Jokes which depend for their humor on racist, mysogynist, or homophobis stereotypes *do* seriously assault individual personal worth, and in a general entertaining medium we should be able to get by without it." Erm. Linux struct dirent isn't the same as BSD struct dirent. Ugh. Maybe I won't use BSD fortune. I'll just use the data files. The debian guys are right...strfile is stupid. MMmm...Simpsons...

Whoever (from scandium.netdrive.cx) keeps attepting to overflow my NFSd, please stop. All that beeping in my syslogs is getting annoying.

It's been a rather unproductive evening. I'm not real sure what to do next. Probably go find some tums.

I did actually make it to bed yesterday. Just didn't stay there. After several interruptions (family oddities) and the fact that it was in the mid-90degF's in my room, I got up. And UPS came. That convinced me too.

I'd actually forgotten...a few weeks ago, I picked up what amounts to a random pile of X10 parts. [For those who don't know, X10 is a way of networking small devices across 60hz AC lines. Its used for home automation devices, such as to turn lights on and off, etc.] Part of the pile was a CM11A computer controller. Thats good. Also included some screw-in incadesent devices, a lamp dimmer, and an RF reciever. Yes, and RF reciever. Its for the RF+IR remote that I also got. Six-function remote. Its huge. It does RF for X10 control and the IR side can be used to control five other IR devices just like a generic replacement remote for TVs and things. Its neat. I think I'll switch to it for my Big project instead of the Logimedia remote.

After a small mtndew-related accident, I got it all installed and programmed the screw-in modules with the windows software. It all works rather nicely via the remote. I haven't worked with the computer controller much at all (just to program those screw-in modules so I could turn them on and off with the remote). There's a couple packages for linux to use it. I'm sure they work just fine. One more thing for the todo list.

Also new on the todo list is a few other things of little importance. Firstly, I need to move our "little" (its actually rather large for a home) network to a better IP scheme. Probably a 10.x.x.x. This whole idea of just picking a random number when I set up a new box is getting confusing. It also makes DNS quite interesting (and usually non existant, meaning I must type in IPs, which means I must remember them, which means I won't). Also added to the list is the Video Blaster driver. I spent most of this morning working on the logical layout of that. The beginnings are here. It's kind of ugly. I need to clean up the code that I pasted in.

So back to yesterday, I went to bed around 1900 and got up about 0200. About the only really productive thing I got done was take a shower. Which is rather odd since that was probably the lowest priority of things. Yes. Priority inversion. The next step is a mutual exclusion deadlock...

Hola... Well, I didn't get much done after that. Basically did some more stuff out in the garage with the Suns that got me nowhere. I did get harmless running. This was only after I put another 8mb of RAM into it. Must've been the swapping that was taking so long. Did install. Works nicely. Even have X running. Went to bed around noon and up by 1900.

I've gotten a serial cable (set of cables) working that let me get a console on the SPARCstation 1/1+/2 (aka, trillian). The debian CDs finished downloading. I'm installing it now. I am rather impressed. The first few times I booted it up, it either 'Data Access Exception'ed or NULL pointer dereferenced. After moving it into my room (in order to log the bootlog to send it to the list), it works. It's great. Debian is great. It started installing itself diskless. And thats actually a supported mode of installation and operation! I have never seen a dist that will actually install itself that way. Incredible. I actually am planning to put a disk in it, but I am going to try this first. Other than displaying quite slow (serial console is at 9600baud :), it is going great. It's done with the base system...now the dilema is how to boot off the nfs root. Hrm. Well. Issue "boot net nfsroot=/trillian" and use the first kernel I found from the CD.

Just booted nicely. I'm in dselect now. I think 9600bps is closer to the way UNIX was meant to be...gives you time to think about if you really did the right thing or not before you see the result. On the main install portion now. Should take forever...

I didn't let it going. I stopped in the middle. I want to use a hard drive. NFS is too slow.

Just got done reading netbsd-vax for today... Firstly, someone pointed out a neat thing that I didn't know existed since Archie died: Lycos FTPsearch. Works rather well. In fact, it helped me find a boot image for my DECserver 200/MC terminal server (which is rather useless to me because I would need to have a LAT server going all the time, which means VMS). It will be neat to look at anyway. The only thing I would find useful about it is if I could use it as a console server (for Sun and DEC serial consoles, or even PC serial consoles). I put a copy of the image on ftp.auk.cx. I doubt anyone else wants it. But its there so I don't lose it.

Anyway, the Jabber head mailed me. He's go an 8227. He sent me a copy of his lantastic setup. I hope I can get the cursed thing booting. He also mentioned that Solaris (2.6 at least) has an rpld that supports the protocol. Might be an alternative. I will need to check to see if my 2.4 SPARC version has it. If not, I do have a copy of 2.6 x86 (a freebie package for "developers"). Need a linux rpld in any case.

Meanwhile, his main point of the mail was somewhat of a plea to help write AIM transport for Jabber. It looks like an interesting project. I've subscribed to the mailing list and started reading through some code. Doesn't look too difficult. And if they have a decent console client, I may just use that instead of naim. It would give me a reason to have an ICQ account too (I'd have to really dig to find my number...its been a while).

I suppose everyone has heard about the Qwest bid for USWorst and Frontier (I think its on slashrot now). It's useful to point out that USWest is our RBOC here in AZ. And that Primenet, my ISP, was bought by GlobalCentre a couple years back, and shortly after that GlobalCentre was bought by....Frontier Corp. Yes. Monopolies are fun, no? So if the gov't actually allows the mess, Qwest will own our local phone service, internet access, several of the major AZ backbones, and with options for long distance phone service. The last one is the one I think will kill the deal. Ever since the forced AT&T divestiture in 1980, its been mostly illegal for one company to be both an RBOC and a long distance provider.

Oh. For the curious. Here's a bootlog from my SPARCstation 1/1+/2 (it seems to have somewhat stabilized as a 1, however). [I do have a 2.2 kernel that runs on that box, but it doesn't have the cute penguin, so I made the log with this one instead.] I should start a running house-wide BogoMIPS sum.

It's 0842. For future notice, Sara is getting induced on Friday morning, 0600 local. For all our sakes, I hope the baby comes shortly thereafter.

I'm going to clean up my local copy of libfaim (I'm not sure how borken it is) and start on figuring out jabber. Well, it doesn't look like I've changed it at all.

Haha. CNBC has the CEO of Beyond.com (who has those commercials with the naked entepreneur) on the show...and he's doing the interview naked. Where does advertising stop?

Bill Gates just officially stated that he is a "Computer Geek". Interesting. Is that a professional title now? Do I need certification? I guess it gives me license to take over the world then?

The jabber cvs pokes along... Wee...just got two cabbar jabber clients talking to each other across my own server. And then jabbertransport promptly segfaulted. Whoops. Time to make an empty transport to fill AIM bits into.

Grrr... need to figure out why setuid Eterm doesn't work whle running Eterm directly as root does. Hum. /dev/ttyp[a-f] were not of the group tty. Yet all of them before and after that range were. Oh well.

Well, I messed with writing the transport a bit. Didn't really do anythingn useful. I'm having a difficult time getting the AIM name connected to a jabber name. Jeremie says he'll write up some info later today. For now, I'm going to bed.

Hmmm...Lessee...

So firstly, the debian dowload slowed down a bit and in fact the site I was getting it from eventually died. So I downloaded it to primenet from another site (from denmark, all the us sites are dead and the UK one is terribly slow) and made my 8mb chunks there. I'm now on the 41st of 49 for the second CD and the 33rd of 69 for the first CD. So maybe monday afternoon for completion. Oh well.

I did indeed go to bed. Got up around 1800. Went to Fuddrucker's dinner. There was a guy there who looked exactly like Robert (X) Cringely. Came home and basically did nothing for a while.

Cleaned up the garage a bit (not much). Burned a Stampede 0.89 CD. (I made the image on mediocris and burned it on my dad's box -- but that required moving my mp3 collection (all 1.5gb of it) to deepthought (which still has six or seven GB free!).) About 0400, I booted up a Sun 3/60 to make sure my serial cables were in order. They aren't. Bizarre problems. I think one of my big straight-through db25's has got a short or two in it. I can't think of any other explanation.

Slept for a couple hours. Took a shower (hadn't done that in a while). That's about it. I'm going to start a Stampede install for harmless here soon. Oh, I booted up fogbow too, to use as a terminal server (it was originally my CUMULUS master node, but at this point, its the only node left at all...486's with two token rings and an ethernet attached aren't terribly useful outside of a CUMULUS-like project). (I need a terminal server to plug the serial consoles of the Sun's and VAXen into, local graphics consoles are often inconvenient.)

For those not following linux-kernel, it looks like we may finally get a non-contiguous RAM patch into the Linus kernel. That means using video blasters is possible without taking out all your RAM.

So I dug up a Seagate 500mb SCSI drive (looks like it came out of a Mac, probably did) to go in harmless (as oppossed to the big HP 400mb SCSI). Stampede is formatting that now. We're going to find out if this decade-old 1x Sony CD reader can grok a fresh CDR disc. I'm going to start the copy and then go to bed. It stands to take a while (1x). Ha. I think the CDR disc is a bit lightweight for it. Sounds like its shuttering in there... but it is reading okay.

harmless is still installing. I'm not real sure why its so slow. It's decompressing from the hard drive, so it should be rather quick. I should have put more than 8mb ram in that box...

Woke up around 1830. Watched Simpsons et al, while listening to the beginning of Wagner's Die Meistersinger Von Neurenburg (performed by the Lyric Opera of Chicago) which also started at 1900. That just finished (at 0030).

I threw together a cute little ROM image merger that combines n xbit into one (n*x)bit image with the files laid out alternatingly, just as the CPU sees them. That let me put one of the Pericom images together and run it through strings. Not very helpful. But its nice to look at. One of them is also only 8bits. Weird.

I'm not sure what I'll do now. I'm sure I'll find something.

I did get to bed there eventually, just maybe not as quickly as I'd said. Of course. I never do.

Not really much has happened since I got up (at around 2100). Watch Wrath of Kahn. While compiling CVS GNOME/Enlightenment (it was, in fact, in the order opposite to that) for my laptop (it was still running a CVS from early May). E went well. ORBit just had major problems. Didn't want to reliably interpret wchar_t as a valid data type...at times. Other times, it liked it just fine. I don't even want to explain how I got that to compile. I accidently did gnome-core before I did gnome-libs. Both had minor, but easily correctable, problems. I really should have done them in the right order. I'm not sure if its related or not, but panel doesn't work. On the whole, it took six hours to get to the point where I could say that panel doesn't work. I was not amused. Of course, I was compiling over NFS over 10mb ether (on my laptop). GNOME is huge. And frankly, it doesn't have enough functionality to make it worth it for much longer. I have on the whole spent more time compiling and fixing GNOME components than I have using them.

The word on the MIPSCo boxen for now is to wait to see what I can get (and afford to ship). The guy has a two car garage full of the stuff and needs a bit of time to get through it.

A newer kernel patch for supporting RAM holes was pointed out on l-k. It looks like its done very nicely. I don't know whether my BIOS supports ACPI or not. We'll see. I'll need to port the patch to 2.3.6 as well (its for 2.2.6). There's a patch to fix the 2.3.6 mmap() problems in my inbox. Though even with that, Alan's word is "2.3.6 right now is definitly for the bold". I haven't applied the mmap patch yet, so I don't know what other run-time problems it has.

I think I may go back and finish my driver for the Video Blaster. It's been nearly exactly a year since I started it. And still no one else has written one. So I guess I'll finish mine.

I'm reading through the 2.3.6ac1 patch now. He put in my AVEC fixes and my netlink compile fix. However, it does NOT include the mmap fix. Interesting. Maybe he doesn't agree with it. In any case, its a rather expansive patch... 950kb uncompressed, and touches 469 files. He went through and changed the way __initfunc()'s are done (it's now just __init). That required changing every single thing that uses __initfunc()s.

I'm throwing together a 'new' box (harmless) for VB work, and anything else I don't want to reboot mediocris to play with. The only decent dist I've got laying around is Stampede Europa 0.89 (the same that I have on mediocris and pileus). [Uhm...Brock...where is that debian CD?] It's not going very nicely. The only Stampede bootfloppy that has network support doesn't support the filesystem format of the only root floppy thats available. I'm substituting a Slack 4.0 boot floppy. Hopefully that will work. Other than that, I'm just hanging around linuxnet #linux waiting for something exciting to happen...

Well, I've had it for now. Nothing wants to cooperate. Well, I shouldn't say nothing. In fact, its just floppy disks. Yes, I would be much farther along right now if I didn't have to deal with the cursed things. I'm going to bed.

Today was an all-around mail-reading day. This included sending a total of three kernel patches out (two of them you already know about, the other was correcting one of Pavel's patches that said swapping onto RAID is a Bad Thing, which isn't always true, as deepthought (and ihpled (RIP)) can attest to).

I surprisingly went to bed at around 1130. Got up at 1900. Slept while I was there. Ate dinner. Watched the last half of a ST:TNG rerun (one of the "Let's kill Tasha again!" episodes).

NatSemi has announced that if their Cyrix subisdiary doesn't get bought by 30 Jun, they are KILLING IT. It's bad too see another anti-Intel business die, but Cyrix was dead long before NatSemi bought them. AMD is the future. (/me drools over K7 specs...)

Anyway. Then I decided to start working on mediocris. Took it down. Pulled the Genius sound card i had in there (never did get PnP to load it). Pulled the AHA-2940 (I don't really need it anymore, again). Put in a Kingston ethernet board (so I can have a dedictated 100mb FD connection to my laptop) and a Pro Audio 16 (with SCSI, made by the now defunct Media Vision, a lovely Singaporian company, popular among anti-CREAFers like myself). Modified my radio->cd cable again (PAS uses an oddball 5pin CD audio connector). Plugged the line-out of the PAS into the line-in of the GUS. Plugged the AVEC audio out into the line-in of the PAS. Radio into the CD-in of the PAS. And my two CD players into the CD-ins of the GUS. This lovely two-tier setup gained me extra stereo input that I needed (well, I could've kept it the way I had it, but it wasn't electricaly the Right Thing to do (plugging two outputs together)). Turned it on. Tried to boot 2.3.6. Ran into the increasingly infamous mmap() bug (init couldn't load its shared libs -- this kills booting rather early in the process). Went back to 2.3.5. I had not noticed that ALSA didn't support the PAS. But OSS does. And rather surprisingly (and amusingly), I could load all the ALSA drivers and the ALSA GUS driver (which I now know i need to patch for static record buffers as well) and then simply insmod the OSS PAS module and everything would work great. gmix shows two sets of mixers. And, best of all, I can now play mp3's on mediocris (by routing them through /dev/dsp1, which is the PAS and therefore using OSS). It all works rather nicely indeed. I can have four physical stereo inputs, two physical mono inputs, four digial stereo outputs, and two synthesized stereo outputs all on the same machine and all coming out one output to my speakers. It's lovely.

I've decided I need to work on my SPARCs here soon. I've scheduled my SPARCday for Saturday. My dowload of the Debian/sparc 2.1 CDs should be done by then (my calculations come out to roughly 44hours of download time, which should be sometime on Sat). [Oh, and yes, that is why you are probably getting painfully slow connections to read this, btw.] On SPARC32, the only two dists around are Red Hat and Debian. So that was an obvious choice (this is for a SPARCstation 2 clone, btw). My other SPARC (the 4/330) needs work as well.

That is dependant, of course, unless my sister goes into labor between now and then. Which is duely possible. I even decided to take a shower today (which I had casually forgotten to do in a few days) in case she did. Contractions started yesterday.

I need a few RJ45 connectors so I can make a crossover cable from that new kingston to my laptop (I've been in the mood for speed these days, and 10mb FD wasn't cutting it).

Oh, be warned. My dad freaked out today because he got in a crop of new Kingston boards (which we get for the specific reason that they use the excellent DECchip 21x4x controllers) which now have Intel chips on them. Remember that Intel bought DEC's fabs. It's the same great chip. Just a different name.

Thirty MB out of 800mb+ of debian done. I'm getting the 8mb chunks of the ISO CD images. Seemed to make more sense to me than grabbing every file individually, if I was going to get them all anyway. Unfortunatly, I decided to do this one day late. My dad doesn't work on Fridays and therefore can't suck up his work's T1.

These "Running the unstoppable Windows NT" commercials from HP are getting annoying. Actually, not getting. They are. The first one I saw annoyed me. How they can get away with such blatent lying is unknown to me. Probably because some idiot beleives them. Though they would have to be living in a hole. I don't know of anyone who hasn't seen a Windows crash.

I should have a VAXday as well. Where I install NetBSD/vax on my MicroVAX 2000 and power up OpenVMS on the "big" VAXstation 3100m40. I never did get very far learning VMS.

Today's NTK points us to this article, which explores the phrase "Jar Jar is Gay". It's a rather badly written article that is basically just the unconnected rantings of a disjointed individual. But hey, that's what modern journalism is all about. Is there any connection between being not-gay and hating Jar Jar? I think not. [I'm not sure if the author of that article has a really bad sense of sarcasm or just can't make up their mind about whether they can accept homosexuality or not.] I hate Jar Jar not because I'm heterosexual and it promotes a more effeminate male character, but because Jar Jar's character is disruptive to the flow of Star Wars.

I hope you caught the joke with the 'modern journalism' part. Because it was quite amusing.

Hmmm...gallium arsenide (GaAs) as an Si replacement? Can anyone else hear the sound of Windows users' hard drives getting formatted?

Here's a good laugh for the bored. "Tell NBC To Renew "seaQuest DSV" So We Can Petition Them To Cancel It Again!"

Now there's something unexpected... Hits from .ibm.com on the IBM8227 pages. Also, looks like there's another Linux/VAX effort, or at least an extention of the other. And, well, they actually have a workable toolchain and a fair amount of running code. And the status page was updated a mere couple of days ago. "Wow." They've even got a screenshot (no it doesn't run minicom, X, or icewm...its a serial console). They've even got a vilo.

I'm sure this log format is dreadfully confusing to read. But I assure...I like it. I leave it open all the time and write something whenever I think of it.

Just removed all relevent references to {delphid,ihpled}.ml.org from this entire site (it only took me five months to get around to doing!). Still need to get majordomo and all my old mailing lists going again. I need to be at the optomotrists at 1130. Which is rather soon. Oh, and I may be able to get my hands on a few old original MIPSCo MIPS boxes (ie, back when MIPS made machines AND CPUs, in their pre-SGI era).

Been to the optometrist (I dont understand how people can stand glasses...such a dead perception; they kill depth of field, not to mention extremely poor field continuity -- ive worn contacts for over five years and have never had these types problems). Anyway.

More about those MIPSen. I think I will get a MIPSCo RS2030, which is an oddball, and, according to Tohru, is very poorly designed and uses some very strange logic. But the guy I'm getting them from appears to be a former MIPSCo employee, and has docs for it. Might be a fun port indeed (I'm in NetBSD mode now). Also, maybe an Magnum R3k or an Magnum or Millenium R4k. What's a summer without new wacky hardware?

I've never been one to favor overclocking, and this whole new Palm V overclocking is just creepy. Software that tweaks your clock? On your PalmPilot? That's just weird. I don't really care. I don't have a Palm and I don't really have a desire to have one.

David will enjoy the Ebay stock charts today. For those not sitting around watching CNBC all day long, Ebay is on its 17th hour of deadness. Their stock is down 9% on the day. And I must say that I would have done nothing different if I were them. This stuff just happens. At least they were smart enough to pick their slow-time (mid-week) to do the upgrade. It's just too bad it failed. And for that, I laugh. And I'll probably laugh some more when they have to extend all those auctions or pay the sellers. And if they extend them, that means they'll have twice the average number of auctions running just after coming back online!. Poor, poor Ebay. And Kent David's question is... "Why does anyone care?! Ebay has never even made a profit!"

I'm going to be going to bed here in a bit. Getting tired. It's nearly 1300 here.

I can't say I remember much about yesterday. I know I didn't get much done.

This morning/evening has been rather productive, I suppose. Linus finally released a 2.3.6, squashing the conspiracy theories about its nonexistance. In the 313kb patch was included my AVEC, TEA6320, and Alps tuner code. Which was broken. I then spent a deal of time fixing that, preparing a tidy patch and sending it off to Alan and l-k. And then went back to compiling 2.3.6. Which stopped rather abruptly. Fixed that. Prepared tidy (one line) patch. Sent to l-k.

Oh, I know. Last night, after failing to aid Josh in his Digi multiport serial problems, I watched a production of Puccini's Turandot. It was a rather special production, however, since it was actually performed in its written setting, The Forbidden City (no, not that one; the one in China).

So some completely qualified gay guy gets nominated for Treasury Secretary and some wacko Republican (as if there's any other kind) from OK starts a coup to kill the nomination. Gee. This is a wonderful government.

Anyway, after that I pondered doing something useful. Then decided to do something useful instead. After adding a couple things to my todo list, I decided to try one of them. The Pericom one, in fact. No, I didn't get anywhere. It's still dead. And the board isn't cracked like thought. And power is getting to the CPUs okay. So I gave up and decided to make copies of the ROMs. Or I thought I would. From the looks of it, they've taken the liberty of starting to corrupt themselves already. Bit rot is certainly understandable. They probably haven't been read in over a decade (the system is 13years old). Unfortunate. That's probably also why it doesn't boot. I also put up a description of the box and a rather complete parts list (yes, I was quite bored at 0300). That can be found here. (The note in the TODO list about it being rare was only a guess; I didn't know that I had only the 263rd unit manufactured.) The 37pin connector is kind of odd. I didn't try plugging a null modem into the LINE connector. It may trying to do a serial console or something, considering I dont have a keyboard for it. It's obviously MC68000 based (in the original gigantic 64pin DIP -- over an inch wide and four inches long! U12). The other rather prominent chip is the Hitachi HD63484 (U86). I'm guessing its the video controller. But I really don't know. If anyone knows anything about it, let me know. The ROM images (or whats left of them) are available here. I haven't figured out which ROMs go together (they're x8 ROMs, 68k is 16bit, I'm assuming they're paired).

This morning, as I've said, I've been compiling 2.3.6 and doing everything I have to do to prepare to upgrade mediocris (recompile bttv, update my alsa cvs tree, patch the gus driver for static DMA, etc etc etc).

Note to David: From what I have heard, AMC is a deathpit. And its best not to work for those sort of operations.

TVGrid just reminded me that Star Trek II will be on TBS on Friday. That's not one of my favorites. But I'll watch it anyway. Because its classic ST.

Hmm...I'm now in the Lexington Who's Who directory ("an interesting compilation of exceptional people with their profiles in a special Millennium print edition plus the searchable Registry on the Internet"). I guess I'm excpetional now. Aparently they have my 'profile' already, since all I had to do was update my address and email address. I'm excited. Finally someone is going to tell me why I'm 'exceptional'. I can't wait. Aparently the acceptance process is tough. The mail they sent me said that "Appointment is by invitation only." I wonder what kind of web crawler they use to 'invite' people...

Oh well. I didn't write here for Monday because basically all I did was sleep and feel rather terrible. Yes. So lets skip that day.

Tuesday was only slightly more interesting. Woke up at 0200. That was rather odd. My schedule keeps shifting. I'd explain why, but I don't think I need to.

I started updating my CVS trees, and adding another to my collection (Mozilla). I even wrote myself a script that updates them for me. I would put it in cron, but I can't figure out how to keep CVS from asking for a password. It doesn't support $CVS_PASSWORD anymore. I guess I'm going to have to hack the source a bit. I then compiled a new E. Up to DR0.16 now. I didn't notice any big changes. I'm using gtk themes now too. Not sure why. Maybe I just figured I needed to waste some more RAM. And CPU cycles.

So I was sitting there watching the Baby Show with my sister, and Brock and David wandered in. After watching a water birth, we went to a credit union (complete with barbeque) and to Taco Bell (complete with overcomplicated gaming rules -- especially for Canadians without 'math skills'). Many interesting conversations and I came home.

This salami is being rather isolationist today.

I took a nap. Ate dinner. And went back to bed. Got up around 0230 (shifting shifting shifting). Talked to Sara for a while. Then she went to bed. Then David got on. He aparently wrote 'one of the funniest things of all time' in his log. I've yet to read it; we'll see.

Yousuf wrote to say he's in Albuquerque (I hate that word) and to add something else to the last entry's theme of inbreeding. He pointed out that Christian creationism is based on inbreeding. So the question becomes...when did everyone stop inbreeding and start calling it normal breeding?

I'm glad I wasn't the only one who immediatly thought 'pogs!' at Taco Bell.

I think they should do a remake of the 'Crying Indian' commercial (you know, the one about pollution and littering). They should include a floating casino in the river. And the Native should be dressed as the millionaire he is. Unfortunatly, they can't do this. Because the original Crying Indian died a few months ago. And it turns out he was an Italian.

I'm really not typing well this evening....It really doesn't seem like Mon. More like a Tue.

I went to bed around noon. Got up about 1800. Ate dinner. Watched Simpsons and Family Guy. Then watched the DS9 finale. That was disappointing. I never really watched the show, but I got most of it. And it just wasn't good. Talked to Brock and Tara for a while.

Decided to move my Big Project over to GDK (D, not T). GDK offers a few things that modernize Xlib constructs a bit. I was thinking about using Glib anyway, because of the GMainLoop things and GList stuff. I figured glib is the bottom, I might as well move up one more layer to GDK. I still so no reason to both with GTK, however. In fact, I mostly can't because I can't send arbitrary events to itself.

Spent a while starting to make that move. Its going to take a while, but I think it will be cleaner in the long run. And it will let me get rid of the blocking Xlib calls, so reduced CPU usage. I must rewrite every routine I have that touches Xlib, of course. It comes down to about a 1klines that need recoding. I need to rewrite it anyway. I'm changing my interface. Now that I'm going to integrate video into my code as well, I'll have even more versatility in on-screen data. I think I've changed my mind about clipping too. I need it.

Irrelevent comment of the day: Only four women survived the first winter in the Plymoth Colony, 1620. And yet a largish community emerged from that colony. I'll let you put together my point.

Well. I never did get that nap. I just sat in the dark all morning. And for about the last two hours, in meditation. Which was probably a good way to spend it. Went down to Centennial about 0800 to find about 500 people that I had never seen before. Actually, I saw two that I'd seen before, and that I actually knew. But that was it. All the rest were strangers. Took the SAT. Blah. My distate for all things Math returns. I have no competant geometry abilities. Verbal was rather easy. Anyway. Went to mcdonalds after that. About 1230. Ate then slept.

Got up around 2030. David and Brock dropped in for no apparent reason. They stayed for nearly an hour, and then went to 'ambush people coming out of AMC'. I've just been talking to Sara and Tara most of the mid-day. I might have to take that afternoon nap this time. I'm still tired.

Random Red Dwarf quote: "Oh yeah. This is reality alright. I'd recognize it anywhere."

Sat outside for a while after writing yesterday. It was a rather lovely morning. About 60degF plue a mild breeze. Fell asleep around 0900 as expected. Woke up around 1530. Took a shower. Wandered around for a bit. Read mail. Went to dinner at Cracker Barrel. Worst meal I've ever had there. The bread was tough, the meat tasteless, and the potatoes just disgusting. Oh well. Haven't really done much past that. Studied for the SAT a while. I had better get a decent score...because I don't really have the grades to get into the school(s) I want based on them alone. I have to be down at school at 0800. I think I'll take an afternoon nap now.

I was really hoping to get some more coding done today. Oh well. Sara's baby is at the -1 position (crowning happens at +5) and just over 2cm dialation. It's nearly time...

Hum. That means its Friday. Hum. That means I have to take the SAT tomorrow.

Today was fairly dedicated to my Big Project code. Started out implementing channel descriptions and allowing for out-of-channel-order scanning (so I can reorder my channels -- COX has a funny way of doing things...for instance they put C-SPAN1 on channel 28, yet C-SPAN2 isn't until channel 41!). I also made some more hacks to xawtv. This includes adding the 'msgtime' remote command for specifying exactly how long the msg should be displayed for (its not allthat useful in the end, because if the time is too short, the timer expires before it even shows the msg and the msg stays there forever). Also removed all of the automatic messages and added code to my panel code to do a 'msgtime' manually instead (more control!). Then I started getting more unpredictable problems. Traced most of them back to the way I was caching the xawtv Window ID (it takes a while to find, so I was caching it so I could send commands to xawtv faster). I have yet to find any way of caching that number and have it still work. I have given up for now. Of course, that means that changing the channel takes roughly 3seconds (sure, it doesn't /sound/ like a long time, but when you're trying to scan channels for something to watch, it can be rather annoying).

So I starting looking for optimizations. And I didn't find any. I don't see any way to get xawtv to react any faster. [After you spend the time looking up the window ID, send the command to xawtv, then wait for xawtv to update the message and rearrange the clipping boxes for the bttv, you have to wait for it to look up the ID again, and then send the 'setstation' command to actually do it (the first operation just displays the channel name). This all takes forever.] I've kept telling myself over and over that I can't be making my app do everything iteslf; that I've got to let myself rely on other code. But I've had it. I've read through every line of xawtv-2.45. There's just no way I can tweak it anymore. So, I officially gave up on using xawtv.

And then after that, I even gave up the idea that I could actually hack xawtv /into/ my system. That would just be too messy. The most annoying thing would be that I'd have to use Xt (I'm using bare Xlib now). My whole goal in using bare Xlib was so I didn't have to deal with any toolkits. I don't need them. Just too much crap. I have yet to see a perfectly designed toolkit. It's just not possible (I still like gtk, though...but you already know why I'm not using that for this project). So I gave up on using xawtv at all.

Looked at Nathan/diz's tvset. I can't get that to work at all. Just gets stuff in a refresh loop. Even after some slight debugging and code removal, it didn't work. Gave up on that too. Looked at the NMX and XTV that come with Ralph's bttv package. Those both use the TVscreen widget for Xt. I started with NMX. I got XTV to work (which is motif-based, but works great with lesstif). Then just started writing code into NMX (the first step was converting it back to C from C++...yuck). Got it to the point where it could turn on the overlay and set the tuner frequency (not channel...the kernel only accepts frequencies; the mapping between channels is done in userspace, which is a good thing, since there's a mere dozen different channel layouts across the world). A word on the way the freq is done. It must be sent to the kernel as the freq times 16 (as an unsigned long -- this is done in order to keep floats as far away from the kernel as possible (because Intel Floating Point Exceptions are not handlable when tripped from inside the kernel)). Do ignore everything that appears to say otherwise. For this reason, by the time I started writing my own code independent of all other tv apps, I had closed every other app's code that I had open and just coded right from Alan's API spec (and even it has a few mistakes). Cable channel 12 (205.25) is sent to the kernel as 205.25*16=3284.

Also discovered (all by myself) that the TVCard line that I sent to Alan for this AVEC card is wrong. I misinterpretted the names. The field labelled 'tuner' does not mean the tuner type, but the input number that the tuner is on. And the 'svhs' field means the input number that the S-Video connector is wired to, not the number of S-Video inputs it has (which can actually only be zero or one with bt848/878). "Oops."

So anyway, after I finally got mad at everyone's code (including my own), I just started messing with the Video4Linux API is actually rather clean and simple. (However, I can see why there's a need for a revision...the current v4l API is basically just a bt848 designed driver with dillusions of being something interoperable with other cards. Its not. It appears to work rather poorly for cards without the bt848's exact feature set.) In any case, it was quite simple to get the overlay going. I'm really not sure why all tv apps out there are so bulky about it. Out of my code, about 60% of it is just error handling. Setting up the card is basically just a matter of calling a bunch of ioctl()s and then doing a final call to start the overlay. The chip does everything else. You just have to provide the kernel with the {x,y,width,height} set. I still need to work out how you do the audio, but other than that, I have routines to do everything I need to do.

That is, except clipping. (Clipping is when you tell the bt that you want rectangular holes punched in the overlay image. It's nice for when you're doing real TV apps where there's the possibility that only part of your tv window will get covered up. But in my application, that's never going to happen. And if it does, then I don't want to see it. And I won't. Because the bt writes over everything in its rectangle of the frame buffer, which is just fine with me.) So yeah, I don't think I'll implement clipping. The other reason I won't implement it besides that I don't need it is because its slow. It takes between a quarter and a full second to update the clipping array, along with having to deal with a lot of flickering in the process (XFree is really not good at this whole video overlay thing -- I REALLY WANT XFree 4.0!!).

I learned a lot today. A whole lot. And it was fun. I think I got up at like 1330, so, yeah, I missed the first half of Rose. I worked on all the above stuff all afternoon, night, and morning. I'll probably fall asleep here somewhere between 0800 and 0900. I think my mom is starting to get annoyed by my sleeping habits (ie, going to bed when everyone else gets up). However, since she hasn't provided me with an acceptable reason as to why its a bad thing, especially after I point out that I have no reason to be up during daylight, she can't hold it against me.

I think Tara is getting annoyed by my non-presense. Maybe I shall end my vacation today. Or maybe tomorrow. Tomorrow I have the house to myself, or so I'm told. Most likely someone will be home anyway. We're actually hoping that Sara's baby will come tomorrow, or sometime in the next few days. So I may be at the hospital (somewhat amusingly, my mother and I are my sister's birthing coaches; odd combination indeed). I think I have to go out to dinner with my grandparents tonight. So that means I have to get in the shower before midnight.

I hate it when the 2nd of the month is on the 3rd day of the week! So confusing.

I know I implied that I was going to go right to bed after I wrote yesterdays, however, I never actually made it. Well, I did. Just six or seven hours after I'd planned on it. By the time I got done writing, it was sunrise. So I stayed up a while longer. And then it rained (or at least pretended to), so I stayed up a while longer after that. Before I knew it, it was nearly noon. And then I finally just got terribly bored and fell asleep. So yes, I have become completly nocturnal. And yes, it is kind of fun.

Woke up around 1600. That seems to be a rather oft' appearing number. Probably because its a power of 2....and I live by powers of two.

Read slashdot. Found out that Nvidia released an accellerated X server for the RIVA chips. Went and grabbed that. It is indeed faster. I like it. You never realize exactly where the bottleneck is until its gone. Many things got faster that I wouldn't have expected to get faster. Oh well. We all know I'm a bad forecaster.

I was really rather sad that I fell asleep at noon. I missed Charlie Rose.

Someone has mailed me who has successfully gotten their IBM 8227 to boot. He's going to send me the stuff he did it with. Maybe that projects not as dead as I thought. Once I get a clean packet dump of how the RPL protocol works, I can write a linux daemon. And then we can be completely free of a DOS boot server. This all comes at a grand time, for once again I am getting pissed off at the WinGate proxy I'm using now. It's too slow (can't stream mp3's as fast as I know the cards can) and, well, its windows. It fitzzes out every few minutes. I have to kick it a bit with the wingate admin software to get it going. (Needless to say, this random coma is detrimental to streaming anything, let alone audio!) I'd still like to find some way to stream TV across it. Even if it is a really small box. Just to say that I can watch cable without a cable. Yes. That would give me a laugh.

I must say that I really didn't get much done today. Wasted the entire evening and morning futzing with sound and trying to get mp3's to play at all. The best I can do is with ALSA drivers and OSS emulation using an mpg123 compiled for OSS. But even then, it just plays the first three seconds and repeats it over and over again. No use to me.

Patch-wise, I threw together a diff against 2.3.5 that contains many of my changes to get the AVEC InterCapture bt board to work (including tuner line). I sent that off to l-k and v4l-l and to Alan. I didn't actually test them first, though. I don't actually use the kernel bt848 drivers, so it doesn't make any difference to me (I use Ralph's). I just submitted the patch so that the information wouldn't get lost anywhere (ie, it would be preserved in the Linus kernel release). Anyway, the other kernel tweak of the day was reading through the ALSA gus code and figuring out how to do static DMA buffers.

The GUS is an ISA board. Yet it does DMA. ISA only has 24 address lines. Therefore it can only DMA into the lower 16mb of system RAM (same problem as the video blaster has). But, the default behavor is to allocate DMA buffers on open("/dev/dsp") and then release them on close(). The problem comes when your address space fragments to the point where there is no 128kb contiguous block below 16mb. And memory fragmentation, you can always assume, I have a problem with. With my modifications, it allocates the buffers when the driver is registered (on insmod) and will not give it up until its unregistered (rmmod). This reduces the likelyhood that the problem can occur.

Now that you know all about ISA DMA.... I think I'll leave it at that. These logs are getting rather instructive lately.

I did a lot today. Let's see what I can remember.

Slept til noon thirty or so. Read for a few hours. Fell back asleep til 1600.

Fixed POP3 on deepthought, finally. Qpopper went in nicely. I didn't know that it was Qualcomm that wrote qpopper (also the author of Eudora). It works with netscape for my dad, so it should be good. Also tried to fix less (and installed strace in the process). Failed. I have no idea whats going on. For some reason the pipe() call isn't connecting the preprocessor to the parent correctly and it blocks inside a read(). Dunno.

Hacked up xawtv to support some more things that I need for my Big Project (preproject to mammatus, which is too complex to 'just do'; focussing the way I have it now lets me do it piece by piece with much fewer dependencies to get stuck in). Added the 'setbrightness', 'setcolor', 'sethue', and 'setcontrast' properties (which can be set with xawtv-remote). I further hacked my own menu/panel code to modify xawtv window properties directly, without the overhead of system()'ing to call xawtv-remote first. Also implemented tv_off() and tv_on() that each do the following: 1) 'volume mute', 2) 'setbrightness 0' (or 32768 if tv_on()), 3) 'capture off' (or on -- turns video overlay on/off). I have to set the brightness because if I just shut the overlay off, xawtv leaves the last frame in the window, defeating my purpose of having these functions to begin with. If I turn the brightness all the way down first, it still preserves the picture, but its 95% black (almost invisible), creating the illusion that it's really off.

Also redid some more cabling and now have my radio card plugged back in. This works nicely. Shut the TV off via my remote and turn on the radio, or if I feel like it, have them all on at the same time. I can control most everything via the remote now, however, I cannot see the changes -- I have not implemented the menu system very well at all. I plan on using xawtv's window to position the menus and things. xawtv will be the most critical part of my whole system, and I will rely on it for more things than just video.

Speaking of xawtv... I found another bug in it. If I do the entire tv_off() function (which contains three direct xawtv calls) followed immediatly by the tv_on() function (which contains another three xawtv calls), xawtv segfaults and leaves the bttv going (ie, I have to reboot to get my video back). If I make all the calls using the old system() shelling method, it works without dying. I don't have the patience to debug things like this (it takes me several minutes to reboot and get my environment set up to debug again after every segfault).

Also played with the main loop of my menu/panel app. I had accidently designed it so that it sucks as much CPU time as it can get (this is not overly effecient when I have mp3 encoding going on it the background). In that loop, I have to watch for data on the serial port file descriptor (for the remote) and for XEvents. I was doing this with a non-blocking select and a non-blocking XCheckMaskEvent() (the equievelent of a non-blocking select for XEvents). And both these non-blocking calls were by themselves in a while(1). Bad news indeed. I still can't find a way to ask Xlib for the file desriptior that X uses (to add it to my own select()). XInternalConnectionNumbers() returns zero in count_return, and a NULL fd_return (this is an X11R6-only call). The ConnectionNumber() macro (the X11R5 equivelent of the former) returns what appears to be a reasonable fd number (4, in my case). However, select() never sets that fd in the fd_set (ie, the socket is dry). So I have no idea what to do. For now, I'm using a blocking select() on the serial and a non-blocking XCheckMaskEvent() for X. However, this means that it will only get XEvents immediatly after it gets a serial event. Meaning no interactive X use of the menu. Not a big temporary problem, but it could be a big permanent problem.

I thought I'd done more. Oh well. It was still a fair amount.

Smoke was asking about Mandrake (the linux dist, not the plant). I know nothing about it, past that its heavily based on Red Hat. It looks to me like they just take the latest RH and upgrade it to the to-the-day upgrades of every package and rerelease it with thier own name. Can't tell, though. It may be Something Completely Different (tm). In any case, he did get his AMD 380MHz IBM box up to 450MHz (and its still working). Aye, it surprised me too.

Good enough for now.

I figured I'd take the time to rotate now, while I felt like making the effort.

I'll be in the mood to make changes this month, since it is the anniversary month. I've started with converting even more of my heading into a more baroque language (if you can't figure it out, it says "part twelve, for the month June, 1999"). Also, I'm going to start adding anchor tags to each day, so that its possible to link directly to a specific entry. Tags will be of the form "#ddmmmyyyy_ee" (dd = day number with leading zero if necessary, mmm = three letter month appreviation (in moderne english), yyyy = you better know what that means, ee = entry number (will most likely always be 00)). We'll see if its worth it. If nothing else, its the Logical Thing To Do. Also, I've changed the file names to the volume/part number instead of using the name of the month. Firstly, because there already is a 'june.html'. Secondly because I seem to like to put more than one month in a file. I'm also going to make index.html a symlinkto the current month instead of renaming during the rotate operation. I added another line at the top too. Just to frame the heading a bit better. And to annoy even more Windows users with the enormous header size (that header takes up the entire screen in Windows at 640x480). The background color for this month is the same as this month of last year, just shifted down a few bits (darker).

Last month is here. Though you could have just gone to the index to get that.