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

Where have I been you ask? Well, now you'll know The Rest of the Story...

Now wasn't that exciting? After that, I went to the garage and checked everything's status. All running. ihpled has been up for over 41days. I think that's about all new for now. I need to do a lot of catching up before I can determine much more....

Not exactly the best of days, but workable I guess...

Unpacked two "new" Dell OptiPlex Gn's in the cisco lab this morning...this was much more stressful than might be immediatly apparent, as the current conditions were about 90degF @ ~75% humidity IN THE ROOM! I sure wish they'd turn on the air conditioning before noon... Oh, and once again, they ordered machines without ethernet cards of any sort. I just told Unger if they didn't come in by next week, I'll stick a few extra ISA cards into them. These machines aren't permanent; they belong to the bizlab but we're going to use them until the permanent ones come in (rather, IF the permanent ones come in).

As for the Willow scan converter (vga->tv), it appears that yesterday's thoughts about it using the wrong power brick were not completely correct (it shipped with a 12v brick but the input is ~5v). Now, what happens when you plug it into any 5v supply (or bench supply) here, it starts drawing 1.5+ amps and the supply goes into current limiting and stops it, limiting it to ~3.0-3.5v. We're suspecting what appears to be a DC-DC converter on the main board, but there's no trace of the company that made it, making finding pinouts kind of difficult. ModuPower,Inc, anyone? At that low voltage, the thing works, but not very well (a 1/100th volt change makes it comes up with a new vert refresh for the conversion!).

Herbert sent over a stack of 11 (eleven) "dead" motherboards. He says we can half of what we "fix". They're all Asus boards of various vintage. All of them are one of three models: PVI-486SP3 (PCI+VL+ISA), P/I-P55TVP4 (PCI+ISA+SIMM+DIMM), P/I-P55TP4N (PCI+ISA+SIMMonly). I took them all out of their boxes individually and checked them over visually (looking for fractures and such). Nothing interesting found, in fact, nothing at all. They all look completely normal (which is kind of a bad thing since that means the failure is probably in an unfixable IC somewhere). Since I don't currently have any spare Pentium CPUs, I could only really test the 486 boards, of which there were four: one bootable, three dead (no video at least). (As is normal with herbert, not all his "dead" boards are dead, as he often gets frustrated quickly and finds swapping motherboards faster than really debugging the situation.)

Since I doubt he really needs a 486 board, I'm betting we'll get to keep that one. The rest may be fixable too, it's too early to tell. It's a fairly nice board for a 486. It's got voltage regulators so the DX4-100 can run at the proper voltage (although I've found no ill effects from running them at a full 5v instead of the recommended 3.4v). It's also quite quick. It's got 2 PCIs, one PCI/VL/ISA shared, and three(?) ISA (nonVL) slots. If I can find a case with a quiet power supply, it might be an alternative to that Dell NetPlex case/mobo I was starting to setup for the secret project. I may also look into getting another Diamond DVC1000, since they're really nice (much better than the planned video blaster), and are quite cheap. But, let's stick to the base project for now...

Slackware 3.5 is all downloaded, so the connection's back to full speed again. Because of time restraints, I haven't actually gotten it installed yet...

quick joyously arose from the near-dead this morning. It took nearly an hour just to do simple surgery (power supply swap). One of the hard drives squeaks a little, but nothing too major. It all appears to be working just grand. Still need to set up the VPNs to work again, though (ihpled has never run a VPN before, the last time we did was when delphi(d) was still in service).

That's good for now.

Decided to filter all 3382 messages of the iname/primenet inbox. Exciting, isn't it? Also, the connection is terribly slow today because of the background grabbing of Slackware 3.5...

The scan converter arrived today. It's dead. Very dead. My dad says he'll take a look tomorrow (he's far more qualified than I). The power supply is dead, the question is how much it killed with it.

Nothing much done. Watched tv-in-a-window for a bit (not as exciting as it sounds). Oh, and segfault.org debuted today at 4:04 pm (get it?).

Maybe today will be more interesting....

Got back last night about 2330 (UTC) from a short jount up to Flagstaff for most of the weekend (left Sat morning). Nothing terribly exciting. We had to "help" Amy (one of my sisters) move into her dorm room at NAU. Well, that also meant getting cinder blocks so that we could prop her bed up onto leveraging between the counter/desk and the blocks on the floor (two heights and one width upwards...32inches from the floor). About the only other things that happened were 1) ate alot of food, 2) made way to many trips to Target and Wal-Mart, 3) ate a lot of food. Way, way too much food...no more meals for a week... A word of reminder: never eat at the Cracker Barrel more than once a day...

Did homework from the time we arrived to just about a few hours ago, at which time, I decided to read mail. But, much to my disbelief, there had been absolutly no traffic on linux-kernel or any of the vger lists for that matter. I immediatly resubscribed, yet I've seen no messages from there. This is strange, very strange.

It looks like the scan converter (VGA/TV) got shipped sooner than I'd thought. Aparently he shipped it before I even sent the money. I guess there really are some trusing people left in the world... Should be here today, Tue at the latest (it's already at the Phoenix POP). Now, if I could only buy more time...or less homework...well, the latter is probably possible.

Speaking of school, I think I'm going to have to have a talk with Mrs Foster. I'm seriously thinking about auditing that class (ie, just taking it for the fun of it and not worrying about the grade). If she continues to give out this amount of workload (which is currently twice the load of both my AP classes combined per day), I'm not going to be able to stand it without naturally getting a not-so-pleasant mark. We'll see...may I can just convince her not to give as much work...

Since I already gave my word of advice in the first paragraph (!), I'm not sure what to close with... hmm... I know: You can easily get a 30% typing speed gain by typing with your eyelids closed...without the distraction of error and other stupid things, you can obtain near optimal efficiency and no longer need to fret over not articulating fast enough....tomorrow's topic: 'reading with your eyelids closed' (or: 'making the most of nonfiction literature'...or: 'creative writing 101'...or: 'how to avoid worrying about that arachnid-style creature crawling up your leg...ow!').

Pushed gtkFAIM 0.03pre7 out, probably before it was really ready. Included all of James' patches, fixed some things up, fixed a stupid typo that James was kind enough to point out, and that was pretty much it. But, it should now include all the features that I'd originally promised for 0.03, so 0.03 will hopefully become pre7+last batch of bugfixes. pre7 should fix the typo that was causing pre6 to be more unstable than pre5, but that's probably the only bug fix. The rest of them are all still a mystery.

Also started getting the tv/cdplayer/etc Dell-based box ready for Linux. Or rather, Linux ready for no physical filesystems. I'm using Slackware 3.5 as a base, and tweaking the installation until it works without a hard drive (only a raw kernel is loaded from a floppy, just like the old, antiquated delphi was). I'll have to download more of the new slack in order to continue (I have no CDs that new). I'm thinking I'm going to use ihpled as the boot server, but I may still revert back to eagle (though I'd really like to fully decommission eagle one of these days soon).

School was truely dreadful. No air conditioning all day. The rooms were terribly humid, not to mention way-too-warm. Terrible. Foster finnally got tired of it and moved last hour to the lecture hall, but it soon got just as warm as the rest. Speaking of Foster, she also somehow conned me into signing up for the Academic Decathalon. Shouldn't be terribly difficult, the main focus this year is the brain and philosophy. Also, I've ended up in the group doing the middle east for the big ancient civilization project. I was amazed that no one else wanted to do that region, as it's got to be one of the most interesting (and oldest) around. She was actually having to talk people into changing to that region so we'd have enough...

That handbook meeting this morning was not nice in the gym with no air conditioning. Most of it was q-and-a concerning the fascinating subject of parking. Some people are amazing in their concern with superficialities...

Oh, and who could forget chemistry this morning.... literally spent and hour and a half learning how to multiply (in the form of intrametric unit conversions). How so many people can get soooo confused over multiplying and dividing by 10 is incredible.

That's all my complaining for now.

This is probably the first really average day since school started (yes, it only took 3 nonaverages to reach averageness). Spent several hours after I got home watching CNN (concerning the missle strikes in Afgahnistan and the Sudan). Slept a bit. Did some humanities work (Siddhartha and Buddhism-related stuff), and generally prepared for this week to be over (not soon enough). Got into a discussion with brock on the topic of todays events (short, he was not really interested...more interested in his just-ended discussion on the afterlife or something or other). Throw in a couple hundred randomly placed idle minutes and you have today.

Nothing terribly exciting. Stayed in networking so I could hear the great networking guru herself speak about the wonders of using what she calls "firewalls". Although, her "firewalls" have really nothing to do with what the industry (or anyone else) actually calls a firewall. Listening to her, you could easily believe that firewalls will eliminate the IP shortage, kill world hunger, stop all war, and bring the world into complete utopia...

Although I couldn't stay for her full monologe, the parts I did hear were amusing. I'm not going to go into them here, as I really should get to bed. The part I did hear though, included her explanation of how PUSD got the Class A network of 10.x.x.x assigned by the InterNIC for thier private usage.... I'm not sure how much more blatently wrong you can get. Brock tells me that she later corrected herself (during more explanation of how wonderful "firewalls" are). And they let these people teach...

Anyway, yes, one-of-the-greatest-oses-of-all-time is still existant and bootable and in the same state we left it on the-machine-it-wasn't-suppose-to-be-on. Aparently, the district guys have no idea that "format != fdisk". Go figure. Well, anyway, it works and we're happy. Didn't get around to quick today, mainly because I couldn't find it. Brock, where is it???

It looks like I won an auction for a scan converter. The fun shall begin soon. The one I ended up with is quite nice. It's even got modulated RF outputs! Should be fun. I've stripped all the bridge-related peripherals from bridge, so that I can use it for my new automation box. It's a nice slender case and a very quiet power supply (it was originally a Dell NetPlex 425s/P chassis -- same as the OptiPlex chassis of the same model, but slenderer, and has real screws instead of thumb screws). I'll be using no hard drive, so the only noise emited will be from the power supply. I've retrieved the IDE 4x CD drive from fogbow (it was being used to increase that case's structural integrity). The reason that I haven't used it is because it can often go wacky and need a reboot to reset. I just plan to use it for audio, though, so hopefully it will work fine.

Think that's all for now. I seem to have entangled myself into another OOP(L) discussion with brock that I hope I can save myself from. I still seem to have some more calc homework to do... Oh, since brock took his ti85 back (temporarily he says), I've been using the ti92. Yes, the ti92 really does do real math! I was just as suprised as anyone... I still prefer the z80 calcs though... much smoother interface... the more cryptic the better IMO...

Had no time yesterday for this. It is currently Tuesday local, with Wed coming all too quickly. Explicitly boring day. Did some Win95 tricks this morning to get a bunch of Compaq Deskpro 2k's to detect thier sound card as something other than an "Other Device", and then eventually function. This was after an exilerating chemistry class (oh, yes, putting detgergent in milk is just SOOO enthrawling!). Followed by "lunch" (it was way too hot to be sitting outside), which preceded yet another adventure: calculus. The one thing I did learn: teachers don't teach you anything, even if it really seems like they do (they're probably teaching the wrong things, as was the case with the inspirational education provided to me by Orzekawski of year-ago relevence (GTA honors)). The day ended with Day 2 of the Mind vs (Brain vs Body) discussion in humanities. Always exciting, and usually boringly circular. Anyone ever heard of NEW IDEAS???

Well, I'm about done rambling for today (I must do some more school-related things before I even attempt to sleep). Brock tells me that 1) quick is dead (if you're not in the elite group who knows the significance of both the existance of that machine and the significance of it's death, then lets hope you have a reason for knowing). I don't know what we're going to do about that. I'm quite sure it's the power supply, but the arrangement of it's case doesn't allow for power supply swaps (well, without popping rivets anyway). Probably just move the drives to a new case/PS combo. 2) one-of-greatest-OSs-of-all-times is aparently still installed on a machine-it-wasn't-suppose-to-be-on. Boot floopies will aid us in the further discovery and exploitation of that fact. and 3) the "security" of denton's lab is minimal at best. Although this is not exactly new or revolutionary, I suppose it's worth mentioning. The geniuses at PUSD installed some stupid app entitled "Fortress" on them, which from what I can tell actually reduces security rather than increasing it. Aparently, it took David a whole couple-a-minutes to do, AND he did it with a PUSD drone in the same room. It is also come to my attention that all the passwords I put on those machines' BIOS' nearly a year ago are now gone, meaning the machines can now once again boot off floopies. Well, that of course reduces security to zero and defeats the purpose of Fortress, and any other security measure, completly. Oh, for those who don't know the Number One Way to get through Fortress: open up Internet Explorer or Netscape (any ver), and put in file://C|/command.com. Simple as that. Or if that doesn't work, you can do some tricks with MIME extension mapping, etc. But, that's something left to a larger essay on the subject at a later time....

Literally did nothing until about 1700 (local). Worked on gtkFAIM for a while. Did nothing somemore, and came back to gtkFAIM and preleased 0.03pre6. That's all I really have time to write. School unfortunatly has begun to consume my time already, and I must get to sleep.

AFAIK, the Earth is still slowly turning about.... Worked on the new FAIM backend today. It turns out that the reason I was not gettnig all the responses I should have been, was that the new queuing layer was SOOO effecient, it was creating timing problems with OSCAR. Well, not really. I was writing the FLAP header and the FLAP data in two seperate calls to Write() (the bulletproof wrapper for write()), and not all headers were getting matched up with thier data once they got to OSCAR. So, my fix was to do what I did in the old backend: allocate a buffer the size of the final packet, copy in both sections, then call Write() once. This requires an extra copy, but it does get rid of the problem.

That wasn't the only problem, though. The reason not everything else was working right was that I was not sending any of the required data for the Client Ready SNAC. After adding a copy of that data, it got farther and could receive IMs. BUT, the buddy list isn't getting sent properly, and the profile is unreliable. So, no buddy list notifications, which is kind of bad, especially since that's the second most important feature of the protocol! I've worked for probably 3 hours total on the buddy list problem. I can't understand why I'm not getting the notifications. I'm sending the Buddy Add command just as I have in everything else I've written. Nothing. I even tcpdumped it and compared the raw frames. Nothing is different! I just don't understand...

In any case, I did get callbacks done, and all the dispatchers should be laid out in the proper way.

I really should've gotten some other stuff done today, too, but I didn't. I'm doubting much will get done tomorrow either. It's all down hill from here...until over 200 days from now when I get my life back again...it's probably for the better -- I sure don't seem to take advantage of it...

Spent way too many hours trying to figure out why my new backend won't login properly. I've gone over all my old tcpdumps and I've found nothing that shouldn't work. According to those dumps, I should be fully logged in and processing live events by now (that is, three hours ago). I send the big block of commands (all 7 of them) all in one chunk, which it's suppose to respond to with the 7 responses. Well, the problem is that only 2 responses come back (and those two are the most irrevelent ones). I really wouldn't mind if it worked, but it doesn't. I'm even generating SNAC reqid's in order, yet it still won't work. But, if you let it continue, it'll act normal, and you'll show up on other's buddy lists, but you won't get any events, and you're buddy list isn't processed. Also, whenever it doesn't work, this strange SNAC (0001/0013) gets received. I can't find that subtype anywhere in my tables or dumps. Wierd. Also, when you send an IM to the new backend's SN, you get back a SN Not Logged In error.

Did some minor fixes on gtkFAIM, all of them having something to do with memory problems. This is all in an aparently futile attempt to hunt down the mysterious sigsegvs. No one knows... James and Kevin have both sent me patches, which I've yet to include. Both those patches are memory fixes. (I think I already have made most of Kevin's changes, though -- the main problem was I was (quite amusingly) using strlen() to get the (struct hostent->h_addr) length. After changing those to sizeof()s, electricfence was happier, though no where near content. It's main job tonight was to drive me crazy with stack dumps into itself. Ya, real fun...

Spent most of the morning 1) sleeping, 2) replying to AIM-related mail, 3) sleeping. At around 1230 (local), I drove down to the school to help out Unger and to yell at the guidance department some more (they just love seeing me walk in the door!). I'm still wondering why I went to help Unger. I guess it's mainly because it's me who has to deal with her for yet another semester. If I didn't have to do that, I would have told her off months ago (in case you didn't notice, I have no respect for her as a teacher...and my respect for her as a person is dwindling rapidly).

Finally got around to putting up a quick robots.txt file on ihpled. All the extra visitors are nice and all, but the spiders have really been eating my bandwidth. Especially this non-public (!) search engine at Inktomi. How they found me is beyond me...

Oh, yes, less go back to the guidance department. Well, it turns out that after TELLING them the changes that needed to be made ON THE PROPER DAY TO TELL THEM THAT, they STILL didn't change it! Now I can see if I told them on just any ol' day, but THIS WAS THE DAY I WAS SUPPOSE TO TELL THEM ON! Well, I was told to go in there this afternoon and pick up a "fixed schedule". Well, it was anything but fixed. In fact, it was still EXACTLY THE SAME! Neither of the two changes had been made. Now, I get to waste 90minutes on Monday in a class I'm not even suppose to be in because of their idiocy and lack of drive! YOu think people who were PAID to DO THEIR JOB would ACTUALLY DO THEIR JOB! Especially a guidance office the week before school starts! This is THIRD year in a row they've done stupid, but fatal, things to my schedule and every year I go in there and yell at them. And yet, they don't seem to care. I'm perterbed.

I did have a short chat with Mrs. Foster while I was there. General small talk, I guess. Nothing revolutionary. In case I haven't mentioned it before, Carol Foster is definitly one of the best (if not THE best) teacher I've ever had. You just have to be in her class to understand. (And don't believe what other people say about her -- although she conflicts with many, many people (well, most people), she is a very nice person, and a really good teacher.)

I suppose that's it for tonight. I don't believe anything else happened. Well, besides me not getting to Franks files today. Just remembered I shouldv'e been doing that. Oh, well. Maybe tomorrow.

Spent a while preparing gtkFAIM 0.03pre5. Most of that time was spent moving the link lists over to glib GList objects. That cleans a lot of things up, and aparently makes it far more stable. Also, finally got tired of slow interaction, and copied the block read code from the preliminary new FAIM backend into the old (currently used) backend (no, the new one isn't public yet).

Finally got my school schedule fixed. Finally. Only got one year left there after this one, and I'm sure they'll do something to mess it up too.

Got some S3 books today. A three book set: for the ViRGE, ViRGE/DX, and the ViRGE/MX. The last one I didn't even know existed before I had the book in my hand.

The tape arrived from Frank in the late afternoon. I was busy at the time and didn't get to it, and then got started on FAIM stuff and really got preoccupied.

I guess I figured I'd mentioned this already... gtkFAIM made it onto freshmeat today. ihpled has been getting between 1 and 3 individual hits per minute for most of the day since it was posted. I did get three people to mirror at least the latest tarball, and usually all of them. That should be nicer now. HTML can be compressed well enough by the modem that you don't really notice the speed problem. But for precompressed tar.gz's, you really, really notice. This is why I haven't taken the time to mirror the HTML stuff (yet). At this point, it would be more trouble than it's worth. If all this activity doesn't die down, I will mirror it. But a few days of this is tolerable.

I must say that all the comments I've been getting have been very inspiring. Thank you. I've gotten no negative and/or rude comments so far, but have gotten nearly two dozen positive comments.

Thats it for now. I really need to be getting to bed. Oh, btw, zeta has been up for 3 days, 22hours, 23minutes. That's got to be it's personal record.

Quite the boring of days. No tape from Frank. Probably didn't make it out till the next morning. Nothing terribly exciting happened today. Slept for way too long. Spent most of the evening attempting to fix gtkFAIM enough to let it at least login on RedHat 5.1 systems. This is quite the difficult thing since I don't have an RH5.1 system! I finally just disabled the gtk code that closes the login window. For some reason, that code segfaults, but aparently only on RH5.1 systems. I'd have to face the blame at either egcs or glibc2.0.7 (i doubt the latter, though since i'm using 2.0.6 just fine.). Hopefully someone with a problematic system can take a deeper look at it. In any case, I did go ahead and release a gtkFAIM 0.03pre4 once I found the problematic code. Also, I mirrored the FTP FAIM archive onto my primenet anon ftp account. For more info, see the main group of FAIM information.

Talked to Allison a bit in the morning. The topic: outlandish collector prices and DEC hardware (this IS Allison, after all!). She confirmed my ideas about the pricing of that DS5500. Although she didn't know much about it and wouldn't even take a guess at a reasonable price, her wording leads me to believe her estimate would be clearly in line with mine (if anything, I was high). Her expertise is not in the newer DEC hardware (ie, the MIPS machines), but I still take her opinion as near fact. She laughed almost as hard as I did about comparing a DS5500 with modern systems! She also recommended I spend AT MOST $25 on DECmate, though they should be free, especially if its only a I or II. Stupid factor was running high: I should've immediatly recognized that when he was talking about DS200's that he meant DECserver 200s (the 68k-based LAT terminal servers). I did do a little reading about PDP-8s after that (ie, the DECmates, which are CMOS PDP-8s). Nifty stuff. I did also go to the DS5500 sellers web site and took a look at some pictures of the system. I figured anything with a Qbus was large, but, well, I would certainly not guess it to be a MIPS-based system by looking at it! It's about the size of a half-dozen DS5ks.

I wish I'd gotten more done on the IR remote stuff, but I haven't. I'm serously thinking about integrating the interpretation code for my TV remote into the Logimedia code...

I've been having some strange sleeping troubles lately and have ended up with a completly wacky schedule. I did go to sleep yesterday, but not until around 0700 (local). Slept till about 1100. Was woken up by a call from Frank Galeno (IS Manager, KSAZ Television [Phoenix's FOX affiliate, locals know it as channel 10]). Aparently, the news division stores their scripts on some bizare OS/2-based system. Well, one of the files has gotten itself corrupt, and he needs to repair it, but aparently doesn't have the knowledge (he's outsourced projects to me on a couple occasions before), so he called me. I told him to send copies over of it over, along with a known-good sample. Well, he called me back about 6hours later to inform me that the only thing they will fit on is 4mm DAT. Luckily, I have a 4mm DAT drive (it's on ihpled). He's sending it overnight, so it should be here in the morning. And then, the fun begins.

While I was reading my mail this morning, I found someone here in town that's selling a DECsystem 5500. Being such a rare beast, I immediatly sent him my offer (a modest few hundred dollars). He did everything but laugh at that. Stating a bunch of crap about Moore's law and things, and eventually, and I quote: "What you've offered is 3 orders of magnitude below list cost in a time frame that Moore's Law indicates that ~2 orders of magnitude lower than list is more appropriate. This equipment is still very serviceable, and should be priced (at least!) according to some largish fraction of the minimum PC hardware that would be required in order to replace it. ...the DS5500 is still quite serviceable and in many ways is superior to current offerings at 1/10th the cost."

If this guy thinks he's going to sell this on a competitive pricing model he's mistaken. I must assume that he's using an unmodified Moore's table which is suppose to be used for new merchandise. I just can't believe he's trying to peddle a Qbus (!) machine, with only DSSI mass storage interfaces, into a market that's populated with PC-based low-cost hardware. To give reasonable comparison, I brought up the fact that you can go right down the street from him (to AZDigital) and pick up a very usable AlphaStation 200 4/233 for less than $500 that, according to Digital's OWN NUMBERS, is nearly 10 times faster (21TPS vs 198TPS)! As far as I can gather, he's looking to sell this monstrosity at a figure somewhere between US$700 and $1k.

Anyway, he does have some other old DEC stuff that sounds interesting. Mainly a dual-RD54 MV2k, a DECmate, and some DS200s. MV2ks are just neat to look at and marvel at thier heavily packed construction. DECmates I've never seen, just heard about from Allison. DS200s I've never even heard about. Oh, and he's got some Rainbows, but I'm not sure whether I really want a true micro or not (I usually try to stay away from collecting micros, but since they're DEC, I may reconsider). I REALLY WISH SOMEONE IN PHX WAS GETTING RID OF VAXEN!

While I was sleephacking early yesterday morning, I got a basic client-server going for IR remote control support. During the testing progress, I not only decoded the Logitech Logimedia(tm) protocol, but the protocol to the remote to my Samsung 19" TV as well. I would have done my VCR's too, if it had living batteries, but that it does not, and to get new ones, I'd have to actually stand up, which of course I only do under dire circumstances (like, lately, to catch lizards, of which we found another one in the bathroom tonight! our entire house has become infested with these incredibly addorable microlizards (with a few preadult translucent geckos mixed in for variety)). Oh, and in case you hadn't figured it out yet, the Logitech IR reciever demodulates signals other than it's own, which is why my TV remote works. What I'm currently thinking about is wiring up a couple IR transmiters on one of these machine's serial ports, and get single-remote automation for my room (maybe that's what I'll do for the extra EISA ALR board that's on it's way...hmmm....). In case you hadn't "caught my drift", I mean placing a transmiter over the reciever on the TV and VCR and controlling them via the serial ports, which brings up endless possibilites of functuality. Oh, and going back to the beginning of the story, the ircntld server merely passes the raw output of the IR reciever on to each of the attatched clients in a broadcast (one-to-all) fashion. I've been thinking about setting it up so clients can tell the daemon when they connect which device/buttons they want to hear events from, and the daemon would only forward the appropriate events to each client. This would be neat. I take that back, this WILL be neat. Also, the current test client (currently called tclient), interprets and recognizes all the keycodes from the Logimedia remote, except one. I don't understand WHY it occurs, but it happens during times where theres lots of continuous events (like you put your thumb on the pad and just start twirling around). I havent't put it into the tables because I don't know how to classify it. Oh, and the three different pad keycode sets are mapped and recognized appropriately (normal, accel1, accel2 depending on how long you hold it down for). It's over 80 different keycodes just for the 32button+pad remote. Oh, and I found out that the pad is not just a mere 8 segments like it was advertised when I got it: it's 16! (keycodes 0x40-0x4f normal, accel1: 0x50-0x5f, accel2: 0x60-0x6f). I may come up with a generalized protocol for the daemon output, though, so the client doesn't have to know every protocol out there. But, seeing as though this code will probably never get used outside of my own use, it's probably not worth doing it (except for the fact that it makes my life easier in the "long run"). Maybe later. First, I want to get stuff going with this remote, and get some progress done on my Personal Information Console that I'm working on to run on the Wyse. What makes this whole thing really neat: the IR reciever doesn't have to be on the same host as the client is running on (ie, you can control any machine on the network via an IR remote). This will become interesting once I get rid of the TV remote's dependency on the TV, and the VCR remote's dependency on the VCR (I'll be able to remap them to make full effective use of their layout). So many ideas, not enough time.

My schedule for school (you know, that thing that starts in 5 days) has not yet been corrected. These people seem to just find some way to screw up my life EVERY SINGLE YEAR! Freshmen year I had two english classes, sophomore year, I was in senior-level classes that I was not exactly qualified to be in record-wise, and this year, I'm once again in senior classes that I KNOW I don't have the prequisites to meet. Oh, and it says I have "RELEASE TIME" second hour, so I guess I get to have a 90minute break in the middle of the morning too!!! And if they don't let me keep that, I have to spend that time with Unger and her merry group of Cisco routers that she has no real idea how to operate, yet she claims to "teach" a class concerning the subject. Oh well, it should be a fairly amusing semester...again...

Oh, I've been leaving zeta on at "night" because I've been sleeping so strangely that I often get the urge to check my email or something in the middle of a sleep-phase of the wholistic sleep/eat/do nothing but say you are/sleep cyclical procedure. Last check it'd been up for 46hrs or so, which is the longest I've had it up in probably a couple years, and it still isn't making any abnoxious noises yet (but statistical evaluation tells me the "YET" is important in that last phrase).

Well, I stayed up till around noon (local), then slept for a few hours. Not much else done. I did get some neat things done before I went to sleep, though. Mainly: learned ncurses. The (as yet unreleased) current gmarket has ncurses as a compile-time option, and is dual-paned. In the bottom pane, theres a status message (says "Updating..." every few seconds). In the top pane (most of the screen), there's the stock listings. One symbol per line if more than 130 columns are available (ie, my wyse 50 in 132x25 mode), or the old format otherwise. It's very professional looking when running on the Wyse (it looks like the trading floor consoles in the 1980s movies (green only, slightly illegible,etc)).

Also spent an hour or so a few hours ago trying to figure out this DECserver 200/MC I'm getting tired of just looking at. Mainly, I cleaned it out (kleanex's and q-tips). There was some pretty bizzare dust in there. I think it was either in a fire or it's last administrator smoked at an exceptionally unhealthy rate. Very black and very fine (not to mention very plentiful). It's cleaned out (and my hands are quite black). It boots up to the point where it sends MOP boot requests onto the ether. I can't go any farther because I don't have the image it's looking for. I have thought about the interesting project that might come from writing a TCP/IP image for it (it's got a 10MHz 68k). It would have large hack value anyway.

Oh, and the biggest part of today: I got the SPARCstation 2 case. So, I put in the Opus board (fit nearly perfectly!) and powered it up. It seems to be doing fine. I don't have the brackets to mount a hard drive inside the case, but I'll figure something out. I don't have a video card either, which really isn't that big of a deal since I don't have a 13W3 cable for it anyway. I'm running it via the diagnostic-mode console off ttya (serial port) connected to zeta. The case looks BEAUTIFUL! Much nicer and cleaner than the Opus case the board is made for. It's even got the side-facing floppy drive. It's basically the same outside appearance as the Sun3/50/60 chassis', but it's a lot smaller and has the cute purple "feet" on it (they're not really feet: the bulk of the case is supported by large metal struts running the width of the case). And it's got a nice, tasteful "SPARCstation 2" label next to the recursive Sun logo (the "SPARC" even has that nifty special "A").

I haven't decided whether I'm going to sleep tonight or not. Seeing as how it's nearly 0500 (local), I doubt it.

Hey! I just logged in to zeta with the Wyse and TABS WORK NOW! I guess the terminal just needed a hard reset to work. Oh, and decided it wasn't worth it to sleep. Maybe later. Oh, I just figured out what fixed it! Doing a reset fixes it all. I'll just put that in my .profile and forget about it... And that explains why it worked in BSD: NetBSD executes stty in .profile. Ok, I looked in the .profile template that came with my original Slackware (really old--3.0 or somewhere). It has a line in there (`eval tset -sQ "$TERM"`) that's commented out, with the remark: "With the right /etc/termcap, this should work great.". So, I've uncommented that, and Patrick's right, it works great.

Ok, I'm here. Now what... You know it's only the 10th of the month, yet this file is already over half the size of a single month file? It's 50kb already, the last two months around about 90kb each...

Attempted to take a stab at reading bytes off this nifty IR remote control. I don't know whether I've metioned it here before or not. It's got over 30 buttons plus an 8segment, well, pad? I guess that's a name for it. It's a circular rubber pad that you move around. I'm trying to think of a consumer product thats got this, but I can't seem to think of one. Anyway, I'd forgotten most of my knowledge about raw TTY control from the last time I learned it (that was way back a year or so ago when I startd my bestupsd project (RIP)). So, I quickly started reading through my old sources. I'd forgotten that the VMIN/VTIME parameters were more magic than logic, and initally left them out. But, when it started acting strangely and illogically, I put them in. Now I remember. Anyway, the remote's "protocol" is fairly simple. It's got PRESS and RELEASE codes, with the actual keycode in between. What are suppose to be the mouse buttons behave a little differently: they have two seperate keycodes, one for PRESS and one for RELEASE. So, when you press one of them, you got on a total: MAINPRESS, BUTTONPRESS, MAINRELEASE, MAINPRESS, BUTTONRELEASE, MAINRELEASE. That's a little odd to handle, but it's nothing too bizare. Also, the pad has some interesting things. There's what seems to be accelleration codes (the keycode changes the longer you hold down the segment). The tricky part is that this makes the keycode change without sending a PRESS/RELEASE sequence (it changes interim). Should be interesting to handle, but cirtainly ignorable if need be.

I've also been working on getting this Wyse WY50 to function right. I've gotten everything working except tabs, which I think is a problem with something on my system (zeta). I think it's termcap thats screwy. If I login to eagle (NetBSD 1.3), it works fine. If I log into ahplaxob (RedHat 5.0) it works fine. If I login to any of the Slackware boxen, it doesn't work right: some tabs get expanded, others don't. ls is the wierdest looking one, although it's improving -- I updated the fileutils package on zeta to 3.16 which cured several of the ls-related problems (and fixed my colors in rxvt to be the same as they are on a standard linux console!).

I'm planning a bigger project, which will probably be for my use only, since it would be useless to generalize. A general real-time information console. Not just stocks, but weather warnings (this is monsoon season: these are very important), and anything else like that. But, the only requirement is that it's got to be text-only (I'll be all ncurses-based since it will be running on the Wyse terminal). Should be amusing (of course, it's going to be able to be controlled from across the room via my remote control (which will be accessed through the tcp-based multiplexer I'm working on)!).

Off to watch the Jetsons! (By far one of the more humorous machinations of futuristic utopia!) And then, off to bed...hopefully in that order...

Put a 2minute bug fix into gmarket, and released version 0.02. There was an interpretation bug that made "-1/8" look like "+65535/8". Although this may be fine for those who know what it means, it's not aesthetically pleasing to people who don't. Also, although I haven't gotten a GTK version out yet, I am thinking of making an extended console version. Since I moved my Wyse terminal to a place that's visible from anywhere in the room, it would be really nifty if I could have a scrolling stock ticker displayed on it. It can do up to 132 columns (yikes!), but only 25 lines, so using ncurses and kind of doing windowing might help here. Interesting possibilities...

Spent a few minutes attempting to figure out why zeta's /dev/ttyS1 wasn't working. I did find another really neat thing in 2.1 kernel's /proc: /proc/tty/driver/serial! It lists out all the flags and everything it knows about the serial ports. It was enough to tell me that the reason my second serial port didn't work was that it wasn't detected. So, instead of taking the lid off like I was planning on, I rebooted and checked BIOS. It was set for "Auto" for both of the UARTS. I statically set both of them to 3f8/4 and 2f8/3, respec, and now they both get detected just fine (and I can now use both my remote control and the terminal at the same time). I don't know whether it's a BIOS bug or just the Plug-and-Play resource contention problem that usually happens whenever the word "Auto" is uttered...

Speaking of the terminal, it's now sitting atop my VCR, next to my TV (which does dance, btw). While I was moving that, though, I ended up doing far more, as is usually the case (because of the stack/heap-based nature of the contents of my room, moving a single item always moves at least a dozen others, usually in a downward motion). In this case, it was several dozen 3.5in floppy disks that kissed the floor. I'm quite sure I've discussed my loathing of floppy disks in this area before, but I'll say it again: I HATE 3.5IN FLOPPY DISKS!!!! They're too small to stack properly and too large to, well, stack properly! They're just ridiculous. And, there's just too damn many of them (and that's not my fault!). I haven't used a floppy disk for anything other than booting in probably 4 or 5 years (back before we figured out a null serial cable works just as well as ethernet, and then switched to ethernet). And the only reason I use them for booting is because most machines can't nativly boot off the network (which I think would be a handy addition to most PC BIOSes). Well, and since I implemented my non-virtual-enough stacks recursively, after the floppies fell from the terminal movement, a couple nibbles of CDs fell as well. So, I took a half-hour or so and lined them up in an extra nearby drawer. It's very heavy, but it's tightly packed and nothing will move (or if it does, it can't go anywhere).

During my short cleaning spree, I decided to rid myself of the RCA cable that's been spanning my mouse pad for weeks now (connecting the VCR to the Bt848 in zeta). Suprisingly enough, I found a connector that fits the Diamond DVC1000's "camera" port just fine! It turns out the pin layout is exactly the same as that used for the Macintosh serial/localtalk cable! Absolutly no modification and it connects all the pins (even the ones I don't need). So, there's now a cable that goes directly from that to an RCA connector (with 12ft of 75ohm coax in between). Still no audio yet (they'll probably be another cable that goes from an RCA to a phono connector for the GUS' line in), but at least video works without the fragility I had before (two clip-on jumper cables and a resistor lead).

Discussed distributed operating systems with Brock early this morning. Nothing revolutionary.

Finally fixed the parse_incoming_im() bug in gtkFAIM v0.03pre1. Well, not fixed, but just about 99% reliable. It can't truely be fixed in The Right Way(tm) until the new FAIM backend is done. Also, I think I've finally learned how to get auto-scroll-down to work on the conversation text box. It doesn't work 100% correct, actually, it only scrolls down to about 98%, but 99.99% seems as close as you can get without segfaulting in the GTK code. I'd like to at least fix the rc file parsing a little bit before a real v0.03 (which will probably also include sound support since that will be trivial after the rc file gets fixed).

I actually slept till about 1400 (localtime) on Sat. I don't know why. It's not like I didn't have anything to do. I'm to the point where I'm completely bored even though there's dozens of things to get done while I still have the time. Oh, which reminds me, school starts a week from tomorrow...

Someone on the ti-hardware mailing list said he was interested in porting a UNIX varient to the z80-based calcs. I laughed and posted that it would be stupid, as by the time you cut it down enough to fit in the TI-8x (2 <= x <= 6)calcs, there wouldn't be anything useful left. You must note that I tried my hardest to keep emotion out of that post. If you need a good hourly laugh, subscribe to that list! It's full of such idiots that the statements they make are hilarious. I've completely tried avoiding posting before, but since I didn't want them spreading rumors that the smallest Linux was ELKS, I just had to post. I'll never do it again if I can avoid it. They're probably wondering who the hell I am and why I posted to their list... let them wonder...

Read a little bit on the GGI Project. I'm dissappointed. I figured a project that had such emotional members would have more up-to-date documentation on the product they advocate so heavily. Yet, you go to that web site and everything still talks about EvStacks (which is suppose to be deprecated), and there's absolutly nothing on KGICon. Everything I know about kgicon comes from reading linux-kernel and the limited about of commented ggi code. I did find enough to tell me, though, that my #9 Imagine 128s2r2 is not supported. But, the S3 ViRGE/DX/GX is, of course. It doesn't seem they have any "generic vga" driver, so it would appear that I won't be able to take advantage of any of GGIs multi-head features. Oh well.

I know I probably did something else today, but I just can't remember. Still haven't caught all the lizards yet. It appears there's at least one left. He's too fast for me. I'll just wait for all the storms to be over and he'll probably let himself out. Or, he'll just get real big eating the collection of spiders I've worked so hard to collect (if you call not throwing anything away and letting it all pile up on the floor "working", that is). Then, he can be a new pet and give the cat something to play with (she seems to be getting bored these days, and she could use the excersize). Oh, well, I guess we'll find out in the Spring when there's a dozen more microlizards scurring around my pillows...

Started out the morning waaay to early. Walked down to the school, stood in line for 10minutes outside, another 20minutes inside. It turns out that someone got the bright clue that you don't NEED a new ID card every year. So, this year, they just take your prior year's card and put a '1998-99' sticker over the '1997-98' writing. Gee, some taxpayer must of complained or something... Wait, no, a taxpayer complaint wouldn't have done a damn thing...

Got back HOME about 0800 (localtime) and decided I should do some hacking while I was still tired (it's always easier to code when you're half asleep). This meant working on the new FAIM backend some more. No REAL progress. It seems like everytime I sit down and start in on it, I have a different idea about how it should be done. I really just need to write one down then code it no matter how terrible it is, just so it will get done and finally fix gtkFAIM.

After I'd just gotten started confusing myself, my dad called and wanted me to go install a un*x on a Mac LC520. So, I drove down there. I really didn't get anything done down there. I finally just ended up bringing a drive home to load off an already installed machine (I think NetBSD/mac68k will be easiest -- esp since ftp.mac.linux-m68k.org is dead/down/empty). He also gave me a monitor to take home. So, zeta is halfway to having two heads (you do the math). All I'm waiting on now is for doug to get ripped apart (which is waiting on him getting the new GigaByte Super7 board and K6-2 that he wants), so I can have the S3 ViRGE/DX board he's currently using. Speaking of S3, I got a ViRGE DX/GX data book today. Little did I know when I ordered it that it was only a "what's changed since ViRGE" book. I'm going to have to order the base ViRGE book now...

Anyway, also while I was down at PUSD's Cholla Annex, I examined some old PS/2 servers. The first I opened up was a Model 60 (I didn't even know there was a 60). It was quite neat. The chassis is padded with foam in a very strategic way in order to promote a clear, efficient circulation path. Some IBM engineer had fun with that chassis design. That box also had the largest MFM drive I've ever seen. It was mounted virtically (ie, spindles perpendicular to the floor) and was quite huge. He said it was only 110mb, and that's about what it felt like (I just had to take it out to see what was behind it on the motherboard). Well, what was back there was about what I'd suspected: a true i80386SX/20. The RAM for this model was put onto MCA-based expansion slots (the two bottom slots are double length and can accomodate both standard MCA cards or RAM cards). Non-standard features for this box included an MCA 16/4 token ring card (the same kind that are in the CUMULUS nodes), an MCA ethernet card from a company I didn't recognize, and an Adaptec AHA-1640 (the MCA-version of the ISA AHA-1540, except that since the 1640 is based on MCA and not restrictive ISA, it can do real DMA with more than 16mb of RAM). I also peeled open a Model 80 for comparison. Roughly the same. It appeared to be nearly the exact same chassis. I don't quite understand what the differences between the 60 and the 80 are. There were no model 95s around for comparison, which is strange for that place -- they usually have lots of extra PS/2s, but it's registration week and many of the out-of-commission servers get put back into service for this one week of the year. My dad also showed me the true-blue IBM "Blue Lightning" 486-class processors (suprisingly, I'd never seen one before; IBMs direct-to-market chipmaking days didn't last very long and this was one of their final lines). This one in particular was a 386->486 upgrade they had bought for a model 80, but then stripped back out when the machine came out of service a year ago. Interesting chip. IMO, it would be neat to be able to but one of the model 80s back together as a complete system (but not with the MFM drive -- SCSI would be adequate) and use it as the master node for CUMULUS: a true-blue IBM Beowulf! That would be neat and would certainly be a great thing to advertise it as... Wow, the length of this paragraph truly shows where my interest lies....

Ok, after all the fun down at Cholla, I drove back home. Laughed at the stock market for a while. Fell asleep shortly after the bell. Did the whole sleep thing for a couple hours, woke up and decided I should get something done. So, I cleared off my desk (mostly, just lifed up the Wyse and dumped everything that fell off the desk in the process into the trash can) and setup the "new" monitor in place of the old terminal (I don't know what I'm going to do with it yet -- it will probably end up on top of my VCR (on top of ahplaxob) so I can watch my TV picture dance while the terminal warms up). So after I got the monitor plugged in, I figured I might as well power up ahplaxob and see how fast the monitor could go.

If you remember back to the last time I powered up ahplaxob, it kept freezing at arbitrary places in X windows. Well, so I can finally narrow it down to the video card. Or rather, the X server's video driver for it. I don't really think it's Alpha-ready. The system works great as long as you don't start X. Since gimp on zeta broke when I install glibc2, I needed to find another place to use gimp. ahplaxob is loaded with RedHat 5.0 (sickening) and therefore came with a gimp (0.99.12 I think). So, I just set DISPLAY and used zeta's X server. Worked great. Check out the new delphid/ihpled front page for some of tonight's output. I was getting bored of the old one and decided it needed a little modification. I especially like my Linux/ia32SMP emblem at the bottom (a 10minute creation -- that's how long it took to figure out why converting to indexed drops the drop shadows).

Oh, and I skimmed through the beginning of the Python tutorial. Neat language. I also installed a python interpreter on ihpled, for those interested. For some reason tonight I ended up at ESR's (Eric S Raymond's) site which lured me into the INTERCAL stuff. Now THAT is cool language. Only five operators, none of which are arithmatical: three binary and two unary (and they all operate only at the binary level). The INTERCAL manual is hilarious, as are most of the INTERCAL-related sites. The names they have for the various operators/specifiers are especially amusing. Go to the delphid/ihpled front page and follow the link in today's news section. Also, I installed C-INTERCAL on ihpled for everyone to enjoy. It converts INTERCAL into C, and then compiles it with gcc. Bascially: INTERCAL is neat. If you're not insane yet (and you're a rare one if you're reading this and that's true), learn INTERCAL. You won't have to deal with that whole issue of being a sane human anymore....

There was also a big storm that rolled through about midnight (local). Not all that much rain (the storm last Thurs was far more productive moisture-wise). This one was neat, though. Mostly electrical. And your life just can't be complete unless you've stood in the middle of a big electrical storm in awe. So, yes, I did spend some time doing that. At about 0030 (local) there was several small, but very intestive storm cells showing up on the doppler. But by 0130 (local), they'd all mellowed out and unified into a single monolithic cell that spanned all the way from Gila Bend to Williams (north of Prescott). And it was nearly as wide as it was tall. That was about the time that the rain all but stopped and the sky began to glow with constant lightening. Suprisingly, the power only had one major blib (which was what reminded me to go move the switch to the UPS -- the note had done no good, it actually took the network to stop again before I remembered!). Hopefully someone was recording oxygenation and ozone levels -- the strange lightening would prove to be fascinating on paper. Anyway, enough about that. (I still hold by my statement that babbling is impossible!)

And remember: your day is not complete until you've shouted "42!" at least once... 42! there. Good morning.

Threw together a quick little app that retrieves live stock quotes from Lycos Investing. I learned a quick way to get specific stocks from looking at their (PapriCom-provided) Stock Watch app for Windows. There's just a http URL you go fetch for each stock, and you get a short near-live summary for that stock (last sale, day high/low, opening price, last sale time, trading volume, net change for the day). It's delayed about 20minutes, but it's useful anyway. I'm going to slap on a GTK+ interface in the next few days so it can be as "full-featured" as the Windows PapriCom client. (Right now, it just downloads the updates for your list of stocks every 5 seconds and displays the summary in text form.) You can grab a copy of the current source here.

Also worked on the new FAIM backend some more. I've done something I've always wanted to do: have the fd_list for the main select() loop be dynamic, getting updated with the current list of open connections. This means unified queues for all connections, which means ABSTRACTION and SIMPLIFICATION, to the extreme. Also, it means that connecting to chat servers and to (shutter) advertisements servers extremely simple, and no more complicated than adding more command interpretations. I've limited the need for client-side functionality to a point where the client can implement as much or as little as it wants, or at least I've done that in my head -- my main codebase is not to that point yet.

What it does right now: call aim_auth_start(), which begins the login process (opens a connection to the authorizer, send the Authorization Request command) and returns control back to the client, which should start it's main select() loop at that point. This will then catch all the rest of the commands on all presently known and unknown connections. When a handler wants to open a new connection (though I haven't done this more than once), it calls aim_getnextconn() to retrieve a pointer to the next free connection struct, then opens the connection and sets the .fd element of it's new connection struct, sends what it wants to, and returns. The new connection fd will automatically get put into the main selects() fd_list, and all commands on the new connection (along with all others) will get put into the rxqueue, and dispatched in a timely manner. Neat, eh?

The 4 4mb 30pin 80ns Parity SIMMs that I hope will work in the new SPARCstation arrived. Yet, still no SPARC case. It SHOULD be here tomorrow if UPS doesn't mess up yet again. Gee...I remember back when you could get stuff from SoCal in less than 3 days...gee...that was just two months ago! Well, seeing as how it's still sitting in San Fernando, unless they forgot to scan it when they put it on the truck, it's not going to come until Monday...

Well, after that brutal storm earlier yesterday, I did awake to find the world still in one piece (though the decision of whether that's a good thing or not is left to the reader). I still haven't gotten the switch moved to an UPS yet -- the note is sitting right in front of me, too!

I forgot to mention yesterday that my dad decided that computer collecting was not such a bad thing after all, and he brought home an old Non-Linear Systems Kaypro 10 and a vintage Brother printer yesterday. The Kaypro 10, of course, has a 10mb MFM hard drive (the Kaypro we had (a II) had no hard drive, just the dual 5.25in 360kb floppies). The printer was something that was expensive for it's time. Not just is it a daisy-wheel, but it has the sheet feeder attachement too! (That means it uses standard copier paper, not the bifold stuff that most of the cheap printers used at the time -- what made that worse was that laser-perferation hadn't come into heavy usage yet.)

It also seems that several pre-adult lizards taking residents in my room, somewhere near my bed. I found one crawling on my pillows and another down lower. They both got away before I could do anything. They're quick before they get old...

Tomorrow is the beginning of the beginning: school registration. This is one of the days where it's painful to have a last name that starts with 'F' or higher. Too early to get up, I tell you...

Well, SRP has been having fun tonight, I'm sure. A VERY LARGE storm rolled through here about 0800 (0100 local time), just a few hours ago. As usual, I was running around frantically deeply hoping the power would return before all my batteries ran out (this house never stops being exciting). It did, and it was only really out for less than 2minutes. David was also on and he was having troubles too, though none appeared to be as severe as the ones over here (he reported a few mild sags, only one of which was enough to shut off his monitor). Basically, the storm was very noisy and VERY wet. (BTW, the power actually went out around 0812 or so (0112 local), and as I'll explain in a minute, it was a service outage over here.) I'm guessing there's about 3.5cm of water in the flowerbeds out front from just the mere 30-40minutes it actually did anything (it was a total of about two hours of massive thundering and lightning). Anyway, yes, it was a service outage over here because even though I said I was going to move it, the ethernet switch was STILL plugged directly into the wall! I've written myself a note to remind me in the morning (it requires digging through arachnids and other assorted things with mandibles, so I'd prefer to do it during daylight hours). Luckily, we're all safe.

The AZPE finally got back to me. They took my suggestion and opened an account with Llama Communications, the premier Linux-based web hosting company. For $20/mnth you get 150MB of space, unlimited transfers, they deal with the InterNIC for you, and you get other assorted things (all Apache-related features, as well as MySQL). And, unlike most other "modern" hosting companies, Llamacom gives you TELNET and FTP access!!! I mean you can get a real bash prompt! It's great. I put what I could find of my stuff for them up there, and more to come soon, as well as some cleanups (esp the first page).

Worked on my new FAIM backend for a while. Most of the routines I want seem to be there and working. IMHO, the abstraction is done quite well. For one thing, I've kept the difference of whether a command is incoming or outgoing down to a very minimum (and they're infact the same exact struct until you hit the lowest level (the scheduler for tx or the command fetcher for rx). Above the transmit scheduler, there lies a routine to translate the abstract data types (SNACs, TLVs, etc) into raw byte data, and above that is the top most level of the backend (the routines to add/remove/create abstract data types). This top layer is exactly the same. The translator just goes in the other direction for receiving, and is then called the interpreter, which sits above the command fetcher, which actually reads the socket. I've tried to make everything thread-safe, although I'm sure it's broken somewhere. Actually, I haven't even started on sockets yet. The queues are populated artificially with programmed static data. But, everything is working properly according to that. I still need to decided how to lay out a main event loop that does what I need and still works with other libraries such as GTK+ (which handles the main event loop competely internally). The design of the main event loop will then affect the way the rest of the backend lays out. Everything seems to be looking good, though.

I messed around with some token ring stuff. I can't seem to figure out what the wiring on a "crossover" cable should be like (by "crossover cable", I mean the cable that connects the RI and RO of two seperate MAUs together). I'm truely stumped. I've got one cable that's a straight thru and it works when I simply connect the RI and RO of the same MAU together. And then I've got a crossed TX/RX cable, which also works!! That is just confusing me. I'm going to have to find that TR FAQ and see if it's got the wirings. (The reason I need this is to connect the two Andrews 8228-clones together.)

I also put together a short blurb about my current other interest, that being making money with microcap stocks. See here for my choices that make up a small sample portfolio of both microcaps and larger stocks. I've had to cut my microcaps list short because of some problems with the normal ones I monitor, problems which don't make them worthwhile stocks to deal with. So, there's only two left, and that's it. The rest are larger stocks, and all of them are tech stocks. We'll see how we do with this set. A few of the things I should have included but didn't because I wanted to keep it around US$300: Lucent, MCI, Texas Inst, and 3COM. I use Lycos Investing to keep the stats for me. My board is much bigger than the few listed here (it's contains several dozen stocks that I watch on a regular basis).

In a battle to waste time, I decided to go check out what Yahoo has added to their OS list recently. Not much. I did find one interesting one though: AIOS. It's a project aimed at create an operating system that integrates artificial intelligence at all levels. Neat.

It's getting late and I should get going. !eyB

Finally decided to waste an afternoon at the DMV getting a driver's license. ZZZZzzzz.... Not too bad: just under an hour. And it's about the most amusing segment of PeopleWatching I've had in a while. There's some real loons out there, I just can't seem to find them as much as I could before... BTW, if you don't understand what I'm talking about, just go to some public place and just start watching people. Most of them are unbelievably amusing! Luckily, no one informed USWest I was there today and they didn't get a chance to kill a T1 or anything before I got out... yahoo!

That wasted most of my day, as my mother seem to have lost the keys to the safety deposit box, and we spent an hour looking for them. Yes, I did ask here whether or not they give you two keys... she said "Yes, but I keep them together because I never loose them.". Arg. What an answer....

Thought about the new FAIM backend somemore. That required too much thought, so I stopped. Actually, before I started that, I was doing some minor work on gtkFAIM. I added some routines to use instead of the normal str*() ANSI routines that take into account the rules of AIM Screen Names (case insensitivity, space irrelevance, etc). The most important two: snlen() and sncmp(). The reason I made these two was to deal with spaces. You see, AIM specifies spaces in SNs as irrelevent and useless. So, it looks like they can be added and removed at will. snlen() is the same as strlen() except it doesn't count spaces. sncmp() is like strcasecmp() except that spaces get skipped. Before I made these, if you sent a message to someone with a AIM-registered space in their SN without specifying the space, then when they responded, it would open up another box for the same person (one with a space and one without). This fixes that mess. I also registered a new screen name for myself ("diputs 81" -- "me stupid" spelled backwards and with the 'm' and 'e' changed to their numerical look-alikes (can you believe that "diputs em" was already taken!!!!)). This was so I could test the bug fix without waiting for a buddy with a space in their SN to come on...

Oh, and since someone asked for my latest tree, I decided I'd fixed enough bugs to merit another release. gtkFAIM v0.03pre1 is open for the taking. I still haven't fixed the rc file parse, or added sound (that one is really just dependant on the rc file, since it's technically simple to implement, but getting the names of the files into it is another story). Someday. (Probably the day right after I finally get parse_incoming_im() to work reliably.) Speaking of that function, it's broken again. Anyone using v0.03pre1 will have to fix it themselves, as I'm not going to fix it reliably until I figure out exactly why/how it broke again (and little fury code gremlins are NOT a valid reason!).

Talked to David (TGaArdvark) for a bit. I derived the "Tale of Bill Gates, JC++, and JZBorland" from a simple question he asked me that was suppose to be a joke. I don't quite remember what the question was, or what most of the story was about. But I remember it was quite amusing to write and chronicled the history of a M$' developement products division in a sort of methaphoric, possibly alligoric, way.... He and Brock only found it mildly amusing, but, who cares... Brock also got a lecture on VLANs: what they are and why you should use them. I really don't mean to give a lecture for such a simple question ("what's a VLAN?"), but usually ends up that way...

Brock wanted to see what Gnome looked like, so I took a picture. (It's over 200kb, so be carefull -- it's a full 1280x1024x16bit shot.) I also included the VNC box I was using to talk to him with. I figured that provided slightly more amusement than the mere sight of the Gnome panel (mine is quite boreful at the moment, since I rarely use it because of it's bad interactions with fvwm's virtual desktop panning). (In case you can't figure it out, we were discussing the best late night snack that didn't involve the microwave (which is very annoying to operate at this time of morning), although you can learn most of my favorite combinations in this log, his were still unknown to me, which I suppose was the point of the conversation, if there actually was one.) That little dodad jetting out from the top foot of the panel contains the buttons for my remote connections (in this order: usr01.primenet.com, netscape, ihpled, and eagle). And the two below the foot are local rxvt and a local superuser rxvt. Real exciting isn't it? Oh, and that's Stonehenge in the back (picture courtesy of a very old Micrographix Picture Publisher CD -- about 2k large TIFs containing images from around the world -- I've slightely edited this one with GIMP just so I could remove the Windows-based tarnish put upon it in years past).

Oh, and for a short while this morning, I watched (and laughed at) CNBC. It's simply amazing how many Dow theorists have a different interpretation on a 299.43pt loss in a single day! In the larger picture, it's what, down 800pts in the past month? And just a few months ago, the Dow hit its high above the 9k mark, and now they're complaining that it went DOWN!?!? Imagine that! Something that went up and then came back down! Can anyone say "CORRECTION"?? Anyway... I agree with guy who predicted a 50% correction off the Dow's highs before year's end. That's very realistic. Anyway, many of the stocks I care about went up, so it doesn't really matter too much microeconomically. Macroeconically, yes, I will probably start actually caring soon...quite soon... bonds, anyone?

Ended the evening with a short round of gtkFAIM bug smashing. It turns out that the new parse_incoming_im() routine that I wrote probably two weeks ago didn't work correctly! The parsing of IMs from non-AOL members (ie, Free (AIM) members) was off by two bytes! That means it got everything correct but the message, which was 0+1 bytes long (two bytes before the real message start, theres {0x00, 0x00}). Just changing a += 2 to a += 4 fixed it, but created a new bug. Aparently there's a Law of Conservation of TLVs, too. Adding a -= 2 later on did the trick. It's now working fine, seemingly. Every once and a while it will just segfault for no apparent reason (usually while it's doing nothing). Because I can't even limit it down to a function or even have it be reproducable, I can't really debug it. Also, for some reason glib (which is required for linking with gtk) suppresses the generation of core files and I don't know how to turn it back on, so gdb is of no real use to me in this case. The one thing I can usually say: it doesn't happen while I'm watching it!

Before I got annoyed with not being able to talk to AIM users (btw, my quick solution was to start up VNC and steal my dad's Win95 desktop, which allowed me to look at the new AIM 2.0beta Chat/chatnav features!), I was tossing around (rather, coding around) ideas on the new AIM backend I'm going to write. I think I've got a decently-abstracted API. It's got a few oddities that I should probably change around, but nothing real big. And, yes, James was working on a new backend and, yes, I was going to use it, but since he's stopped activily working on this project, I just decided it wouldn't be THAT hard to code my own (I could use the experience). And, it's always easier to write frontends for a backend API thats coded in your own style. I've actively started on my new backend. At this point, it can go from abstract TLVs and RAWs (using functions sort of similar to James' early backend, but still vastly different) and convert them into a raw block of bytes, which is whats important. From there, it should be simplistic to write a quick and easy scheduler/queuing sublayer, and then the transmit side will be complete. Recieving is messy. Very messy. I haven't started on my recieving functions yet. Anyway, the TLV/RAW ENcoding functions are far enough along to generate a full "Authorization Request" command with less than a dozen lines of code for the frontend (which are all very simple), and spit out a raw version that looks identical to the one thats used in the current FAIM backend. And, IMHO, it's very clean.

The only part that's messy about this new backend is the backend itself. I've attempted to make a decent effort at an object-oriented (shutter) like aproach. For each command generated, theres a list stored of generic structs, each of which contains a union that contains any one of the three currently supported types (TLV, RAW_CHR, RAW_SHORT). This makes it clean at the upper levels (ie, the front end), but can get kind of messy at the mid and lower levels and you must know exactly what type you're currently looking at to avoid corrupting the union (specifications like flapdata->data.tlvdata->data[0] are not uncommon (that particular one accesses the first byte of the TLV value section of the first node in the flapdata->data list); being so abstracted, these things can get long and difficult to type, but I think it's worth it because it's simpler in the long run to understand and maintain).

Back to the real world... The bickering on the PLUG (Phoenix Linux Users' Group) list about splitting up and/or more formalized organization continues. This may be the longest and most heated argument yet on that list. It's a mess. Everyone seems to be expressing their opinions (even people no ones ever heard from before). I've stayed out of it. It's not worth it. The only comment I made was correcting someone who said that Linus was the "overlord" of Linux. Statements like that just don't deserve to be ignored. Anyone who thinks that Linus has any more power over Linux or even just the kernel itself than anyone else is badly mistaken! He has nearly none, and, it's happened before that someone splits off a seperate kernel tree because of disagreements with Linus (David Miller (with is vger tree) and Alan (although his are usually just stepping stones before he merges with Linus for the next release) both have seperate releases, and Linus doesn't care (as he shouldn't). Anyway, I'm sure you all knew that stuff and it wasn't worth talking about. I just hope the PLUG arguments die down a bit so that everyone can figure out who gives a damn. Ok, you want my opinion? I say that splitting up is a bad thing. Having sub-LUGs is okay, if managed correctly. But having two totally disseperate LUGs in the same area just seems pointless to me. If it doesn't seem so pointless to you, you should hear the reasons this guy decided to make a new group. (He did it because PLUG was too unorganized and he couldn't change that! Yet in the four months he's been with PLUG, he's made NO attempts to help organize or even make suggestions! This just seems very insulting to me (and aparently a lot of other people as well).) Some people...

The flow of free stuff continued today. Cisco sent me the book Switching Technology in the Local Network. Nice, big, expensive book, for free. It covers all sorts of material that I would have NEVER BOUGHT a book about, yet I now have a book anyway. One more item to add to the "Pile of Things I Should Read Someday". Sadly, that pile gets taller at a rate faster than it gets shorter. Someday, someday...

I find it interesting that completely by accident, I began writing this log exactly 24hrs, 11.1seconds after I began to write yesterdays. Conincidence, I suppose...

FYI, icebergs are heavier than they look, and "may appear closer than they actually are"...

Messed around with all sorts of Java implementations in a nearly fatal attempt to get AIM Java and ICQJava running again. Of course, nothing. I've ended up with Kaffe 1.0b. It's come the farthest: it can execute HelloWorld without crashing. Believe me, that's the most success I've had tonight. I finally just gave up on the blackdown (Sun-based) JDK because of it's endless number of incompatibilities with glibc's dyna loader (read: this is what I spent most of tonight attempting to fix). Kaffe was the best choice, as not only does it run HelloWorld, but it's also Open Source, has a custom AWT implementation, and contains absolutly none of the Sun-sourced baggage that the blackdown JDK does. Kaffe does need some help though. For instance, to get jar to work, I had to include the class library from the blackdown JDK 1.1.6. And for ICQJava and AIMJava, I've also had to include the stub classes for sun.audio that I use for Netscape. But, even with those, both apps still die in the same exact spot (java/awt/MediaTrackerEntry.imageUpdate, line 286). I've given up for now.

Well, before I started that, I did some relativly major bug smashing on gtkFAIM. I think you can get it through a decent-length conversation without it crashing now (hopefully, anyway -- no one was on AIM tonight that could stand to talk to me long enough for it to crash -- I guess that means it suits my needs good enough). Also, you can actually close IM windows now without it getting stuck in a loop. Now THAT was a stupid bug...

Other than that, it was a quite relaxing day. We ventured off to the Olive Garden for dinner (my grandparents have a show this week, so as according to tradition, they take us out to dinner the Sunday before). Superb dining there, as always.

This was just one of those days... you know...the ones where the Sun comes up and then several hours later goes back down! How maddening that fact be! What's even worse is, that it's going to come back here in a few hours and make me mad again! "Like your mother always says:" If you think you won't be able to do it for very long, don't bother trying... If only the Sun had enough intelligence to head those words... Well, I'm off to a horizontal slab to await the return of our dear idiotic closest stellar neighbor...

Well, here I be, sipping some strage bagged liquid entitled "CapriSun" and munching on strings of cheese funnily enough called "string cheese"... And I'm thinking...you know? this really doesn't taste as bad as it sounds...

Ok, this was probably one of the more successful days. Just fixing the linker problem this morning was enough to call it a day. Oh, yes, I fixed that problem that linker problem that was causing GUILE not to build correctly (or rather, link correctly -- remember, it was linking to both libc5 and libc6/glibc2). What I thought was a quite amusing cause of the problem: /sbin/libdl.so was still symlinked to /sbin/libdl.so.1.8.5! Not only did ldconfig not relink it to libdl.so.2, but it didn't even catch the slip-stream upgrade to 1.9.9! It was irritating and amusing all at the same time... In case you're wondering how I finally figured to look there... bash did it. I'd given up on bash for a few days until this morning, when I figured I might as well watch something fail for a while... But, watching the warnings it spit out during linkage (which, luckily for me, said "libc5" all over them), I was able to narrow it down to the "-ldl" operand to gcc. Then, of course, I noticed the problem, fixed it, and submited my appology to the glibc-linux list for disrupting them with a stupid problem...

Since then, I've had no linker-related troubles. But, I seem to have accumulated every other kind of trouble. libg++ was interesting. On a lot of packages today (for example, most of GNOME), I just went and grabbed the RPMs for them off ftp.redhat.com and forgot about it. There's just some tweaking thats not worth the trouble, and since RedHat was kind of enough to do it for me, I used it. Of course, that required me to install the rpm app, and then add a few place-holder scripts where Slackware doesn't have cutsie automated procedures. (Most of them just echo'd: "REMEMBER: This isn't RedHat...".)

The one thing that was working BEFORE today and is NOT now, is enlightenment. It compiles and links and all that ok, but when you run it, the background is some putred vomit-shaded orange, all the places where the PNG images are suppose to be are just gray boxes. Because of those two facts, I'm guessing it has something to do with libpng. (Good guess, right?!) I've put in several different copies of libpng to no avail, have recompiled and used prebuilt rpms of everything E depends on, and it's still broke. I must assume that something that got installed with the GNOME RPMs is conflicting (probably imlib or libpng or something). I don't know how to fix it. I gave up for the day.

Well, tonight I realized just how stubborn and nostalgic I can be. I just CAN'T get rid of fvwm. Nothing works as well as it does (even when its broken). Nothing acts the way I like, except fvwm. And, esp this one, fvwm is the simplest of them all. And finally, I've used it for what? 3? 4 years? I know quite alot about configuring it by now... But, it's not going to fit my needs for much longer (allthough I've said that many times now). Some things you just can't let go.... (Of course, you could probably guess my opinions on getting rid of things just by looking at my room and the garage of junk (aka, The Pile)....

Speaking of The Pile, there's a new pile, but this ones only got CUMULUS nodes on it. Eight of them, to be exact. Very heavy. I spent about 15minutes this afternoon running more CAT5 up to the MAUs, and attempting to find an orderly way to run the AC into this thing. I think on power distribution, we'll be ok. The main component in the PS/2 55sx that drew the most current was the hard drive, which we've completely unplugged (for more reason than just power (eg, heat, noise, etc)). I think we're going to split it off the 220v that goes to the clothes dryer... of course, that means we'll have to multiplex our time if we want to dry any clothes... Better yet...leave the lids off the PS/2s and let THEM dry the clothes...

I'm now downloading a new jdk, since I've thouroly destroyed all my other copies (well, one of them was an accident...the other...well...yesterday was maddening...). Hopefully I can get AIM/Java running again, so I can take the time to fix all the bugs in gtkFAIM (gee, haven't worked on that in a week or so).

I'm not sure what else earthshaking happened today. I'm sure you can think of something. Be creative. It's due tomorrow... 0800 UTC. See you then.

We're moving to UTC time this month, also using a more explanative notation: "dd mmm hhmmss.uu". Oh, and a note for those who DON'T know where I live (which I hope is the majority), the times on previous volumes has been in MST (really, it's not MST either...the proper name would be AZST or AST or something...stupid DST...). (MST=UTC-7h, btw)

Since Brock could offer me no helpful suggestions on this month's color, I picked my own. Of course, then he said he didn't like it, but... Anyway, it's a variation on the chroma normally entitled "Green" (though I really would like to know where that idiotic name came from!). Brock's compaint was about it's "happiness" (!). Well, whatever.

Anyway, stuff IS still happening "today while you weren't watching", although it may not appear that way....

After today (yesterday), I should probably be lucky I'm alive. You see, glibc2 really DOES bite. And it hurts. It makes your machine crash. Although, it was fun for a while... when I was attempting to recompile bash using glibc, so I downloaded a new termcap. Little did I know that there's actually TWO termcaps to choose from on Linux, and it looks like I chose the WRONG one. But, the really stupid part was replacing the old termcap with this one before I tested it (I did make a backup though). Then I even more stupidly exited the su I was in. I then tried to su back. That was a bad idea. You see, bash was my shell, and bash is dynamically linked with the old termcap. This is where the interesting part happened... since I couldn't change the passwd file so I could use a different shell, and I couldn't su or rlogin or anything into root, I ftp'd in. Yes, that security feature I removed so many years ago came in handy. I ftp into my machine as root, uploaded a modified /etc/passwd with root's shell set to /usr/X11/bin/xfm (a really old X-based file manager, this is probably the first time I've ever actually used it), and then su'ed to root. Well, xfm started, and I moved back the backup copies of termcap and put back up the original /etc/passwd. That "fixed" the problem, but it cirtainly was not the best way....

Also, I've had not much of anything else besides failure. Installing this GNOME thing has got to be a capitalist ploy to waste internet bandwidth or something. Everything it says it requires requires ten other things for itself. I've probably downloaded 50mb just today to get this even halfway installed (and most of that was just for imlib!). The most annoying part of today was GUILE, the GNU SCHEME interpreter. It must be a problem with my developement environment, because I downloaded every version of that thing I could find and still ended up with an oddly-linked binary. Get this: 1) it linked to the wrong libdl (the dynamic loader lib), 2) it linked itself TWICE to libc.so.6, 3) it also decided it needed to be linked to the old libc.so.5 which it couldn't even find for itself! This just bothered me. And it was ultimitly the end of the line for tonight. Actually, I could have gone farther if I'd installed the libg++ lib. That would have gotten me mico and gnome-core. But would have also sucked up another two hours of the day, so I stopped. I'd like to get Enlightenment installed, mainly because my faithful (and quite ancient) fvwm 1.24r is starteing to go wacky (namely, it refuses to start the clock by itself). And, I've been meaning to go to E for almost a year now....

Also spent about 10 minutes moving PS/2s around. It relieved alot of pressure off the VAXen and Suns. And now all 8 CUMULUS nodes are on the same side of the room. It's HOT out there (for reference, at 2200 yesterday, the heat index was STILL a 105degF!).

Right before I wrote this, I was amusing myself by watching Brock attempt to have a serious discussion about religion and social orginisation with a bunch of (well, IMO, anyway) idiots on IRC. Then, of course, I came over here and started a discussion with him about color (I figured he'd need to start at that low level before he was able to speak sensibly again).

Oh, I guess I should mention that "vol two: jul 1998" of this document is available at http://delphid.ml.org/log/july.html.