In 2038, I will be 64 years old. I design and write all of my software explictly so that it will break (badly) in 2038. I hope all of my peers do so as well. Everybody who works for me does.
The plan is, about 2033, people will start going insane over the Y2038 problem. I will be able to leverage my experience as a Senior UNIX Systems Programmer with a core strength in C to grab all kinds of consulting money. Then, in 2037, I'll do some hard-code hacking (i.e. enter deep hack mode for about 6 months) for some really high-end clients (whoever has the most cash on hand), and throw a bunch of money in the bank.
This is really great, because I don't have a retirement plan, and I'm sure the old age pension will be bust by the time I'm 65. So, after having watched a bunch of COBOL/CICS/etc guys get rich in the late 90s, I want to do the same thing in my early sixties. The best part is, I watched the Y2K crap roll out, and I know how to play the management types that get stuck with clock problems... so I can suck them into weeks and weeks of meetings at huge consulting rates. Maybe I'll be able to bill $1000/hr by then!
Most programmers older than I will be long gone. Most programmers younger thank I won't be able to understand the problem, due to brain infestations of the of the Microsoft and Java variety.
There will be few old-sk00l UNIX hats running around. I will be one of them. Hopefully, by then, I will be able to grow a grey beard, so I can really look the part. My skills will be in supreme demand. I'll get rich off the problems I helped to create, and retire in comfort.
> You use your damn pinky finger! The weakest digit possible! > Try hitting control-v or ctl-b with your left hand, and you'll find your > hand stretching very unnaturally, even on an ergonomic keyboard.
That's why the control key belongs next to the A, not below the shift key. And why it's useful to let press-escape substitute for hold-meta.
Control-X, Shift-S! Save all unsaved files, baby, WITHOUT using your pinky!
Control-K, Control-Y! Kill! Yank! No pinky action, either!
Meta-Control-Q! Re-indent your code!
What other key combinations do you need?
> Windows?
Oh, right, that toy OS for the toy computers. I think I'll stick with emacs, thanks. It's almost big enough to even BE an OS.
> Finally, as for "real programmers don't care" > about interfaces and paradigms, I think they do.
Pffft. Real programmers don't give a flying fuck about the user interface, as long as it meets a minimum standard of useability.
What do you we care about? We care about ass-hats not breaking our software because some idiot pulled his head out of his ass and decided to break libc, or move a bunch of important configurations around. We care about writing code, not about learning new ways to cut and paste or drag porn icons onto the media player. We don't mind ADDITIONAL functionality that comes for free, but god DAMN it, don't change what we already use. We're already busy enough without having to remember what the friggin' short cut key for switching virtual desktops is.
For me, FVWM 1.24 did everything I needed for fast and efficient hacking in early 1997. That's great. You know what? I haven't upgraded my GUI since then. That's right. I haven't wasted a freakin' minute in eight years on GUI config or learning, because I found a solution that worked well. Time for a new box? Great, FTP over the FVWM binary and config files, get to work. Yes, that's right, my BINARY from 1997 still works -- I built it on Solaris 2.5.1, and the freakin' ABI is still backwards compatible that far.
Try going back eight MONTHS with a Linux binary and see what happens.
And THAT, my friend is the problem with Linux. If people would just stop moving shit around and breaking libraries, headers, dependencies, blah blah, maybe more programmers would use it. Most of us don't really want to learn eight hundred different distros only to have our apps break when libc7 comes out -- we're already too friggin' busy writing code. Just give us a platform that works today, tommorow, and a decade from now.. exactly the same way.
And when I say "exactly the same way" -- I mean as far as the software and documentation are concerned. Go ahead, add shit, just don't change what's already there. How hard is that?
>> You don't have to copy the packages. "pkgadd -d..." works fine.
> I noticed this option, but pkgadd seemed to consistently ignore it...
Works fine on other versions. Did you specify the directory where your packages are stored [right], or did you specify the directory where your package is stored [wrong]?
Oh, and as for the installer. Again, on other versions, if you install from the "Install" CD with a supported video card, you get an X installer (actually, I think as of 7 or 8 it's a nasty java->X installer). If you toss that disc and just shove in Software 1/1, you get the good (curses) installer.
> It CAN take days to verify that the logic is > correct and that the record set you are getting > back is in fact what you think it is.......and then writing three other ways to do the same query, and shooting them through the optimizer to see which one will run better on the same set of data..
Uh huh, I hear you.
I think *anybody* who doesn't at least pause to think for a couple of hours on a query returning millions of rows is asking for trouble.
Oh, look, this cartesian join works fine for ten records in the lab -- let's ship it!
This is not an MS-specific thing.. certain contests here require that you solve a skill-testing question.. Nothing difficult, usually it's like 47 * 8 + 2 / 4... I forget the legal reason for this, but I've always suspected it was to keep Americans from winning our prizes.
> If you are talking 5G of data or less on a single box, > realize that Sybase ASE 12.5 for Linux is free
Hmm, that might be enough reason to switch to Linux, assuming I can find x86 hardware I like. I'm sure it must exist by now.. all I want is high I/O bandwidth, absolute reliability, hardware fault reporting, and serial console access. Didn't exist last time I looked (half a decade ago). Oh, and it has to be price-competitive with SPARC boxen; I doubt that will be an issue.
5G is pushing it, but we could make app changes to allow that.. Core DB is around the 2GB mark, then transaction logs start to grow... every 6 months or so, they get flushed.. but our retention mandate is only 40 days.
Thanks for the tip! I'm familiar with the SyBase of yesteryear. It was adequate then, I'm sure it will be fine now. Like most organizations, we don't need much in the way of an RDBMS -- just atomic operations, good backup/recovery plan, and ability to hire high-priced help if the shit hits the fan. That third category is one reason why FOSS DBs are a little weak in my eyes -- we'd have to have good support in house.
Is there really an advantage to running Hibernate on the java side of things when the JVM is only running to run JDBC, on the same box as the RDBMS? One would think that the RDBMS could make better caching decisions about data [locality of reference] than the application layer -- and you'd be looking at two caches now, instead of one. My usual goal is to have enough RAM around for the RDBMS to rarely need to hit the disk under normal application load. Hibernate has one concrete disadvantage from where I sit -- reports run against the RDBMS engine (i.e. SQL*Plus command line) are not always up-to-date, which also makes me nervous about crash recovery.
On the other hand, hibernate has the advantage that I have developers somewhat familiar with it (it is used in another application suite we have), meaning no ramp time as would be the case with ODBC.
I also have strong abstraction from the database layer. We wrote an abstraction layer in 1997 when we were using a small-time RDBMS which subsequently went under. The switch to Oracle in 1999 was a piece of cake; write a new server, badda-bing, badda-boom: clients can talk oracle now, all apps work.
Fortunately, Oracle's still doing the job for us, but new installs are getting pricey. We are seriously looking at either PostgresSQL or MySQL for future work. Either will work, as long as they support the SELECT.. FOR UPDATE; UPDATE; ROLLBACK|COMMIT; paradigm. And easily support hot backup snapshots. Oracle almost does this for you with archive logs, although sanity requires coldsnapping it every six months or so.
I wish jdbc/odbc had been around in 1997 (on BSDI and Solaris). I'm thinking that maybe I should replace the client interface guts with an odbc abstraction.. Hmm.. I wonder what the performance hit would be?
[...] thought it was going to be a hard drama about the pain and suffering of the early days of spaceflight [...] I would love to see a BBC style 'remake' of the series
It's cold outside, there's no kind of atmosphere...
Slashdot Mirror
Haaaaa ahaaaaa!
You just made my point!
(see this post)
Damn, dude, you're 1023? Wow, that's pretty freakin' old!
> I was wondering what happens in 2038?
In 2038, I will be 64 years old. I design and write all of my software explictly so that it will break (badly) in 2038. I hope all of my peers do so as well. Everybody who works for me does.
The plan is, about 2033, people will start going insane over the Y2038 problem. I will be able to leverage my experience as a Senior UNIX Systems Programmer with a core strength in C to grab all kinds of consulting money. Then, in 2037, I'll do some hard-code hacking (i.e. enter deep hack mode for about 6 months) for some really high-end clients (whoever has the most cash on hand), and throw a bunch of money in the bank.
This is really great, because I don't have a retirement plan, and I'm sure the old age pension will be bust by the time I'm 65. So, after having watched a bunch of COBOL/CICS/etc guys get rich in the late 90s, I want to do the same thing in my early sixties. The best part is, I watched the Y2K crap roll out, and I know how to play the management types that get stuck with clock problems... so I can suck them into weeks and weeks of meetings at huge consulting rates. Maybe I'll be able to bill $1000/hr by then!
Most programmers older than I will be long gone. Most programmers younger thank I won't be able to understand the problem, due to brain infestations of the of the Microsoft and Java variety.
There will be few old-sk00l UNIX hats running around. I will be one of them. Hopefully, by then, I will be able to grow a grey beard, so I can really look the part. My skills will be in supreme demand. I'll get rich off the problems I helped to create, and retire in comfort.
Actually, the line from the song is "Went the distance [, now I'm back on my feet]"
No. Costumers are people who supply costumes.
> You use your damn pinky finger! The weakest digit possible!
> Try hitting control-v or ctl-b with your left hand, and you'll find your
> hand stretching very unnaturally, even on an
ergonomic keyboard.
That's why the control key belongs next to the A, not below the shift key. And why it's useful to let press-escape substitute for hold-meta.
Control-X, Shift-S! Save all unsaved files, baby, WITHOUT using your pinky!
Control-K, Control-Y! Kill! Yank! No pinky action, either!
Meta-Control-Q! Re-indent your code!
What other key combinations do you need?
> Windows?
Oh, right, that toy OS for the toy computers. I think I'll stick with emacs, thanks. It's almost big enough to even BE an OS.
> in most cases, the user machine was clean.
Yeah, right. Where the hell did they get clean windoze boxen from??
> Finally, as for "real programmers don't care"
> about interfaces and paradigms, I think they do.
Pffft. Real programmers don't give a flying fuck about the user interface, as long as it meets a minimum standard of useability.
What do you we care about? We care about ass-hats not breaking our software because some idiot pulled his head out of his ass and decided to break libc, or move a bunch of important configurations around. We care about writing code, not about learning new ways to cut and paste or drag porn icons onto the media player. We don't mind ADDITIONAL functionality that comes for free, but god DAMN it, don't change what we already use. We're already busy enough without having to remember what the friggin' short cut key for switching virtual desktops is.
For me, FVWM 1.24 did everything I needed for fast and efficient hacking in early 1997. That's great. You know what? I haven't upgraded my GUI since then. That's right. I haven't wasted a freakin' minute in eight years on GUI config or learning, because I found a solution that worked well. Time for a new box? Great, FTP over the FVWM binary and config files, get to work. Yes, that's right, my BINARY from 1997 still works -- I built it on Solaris 2.5.1, and the freakin' ABI is still backwards compatible that far.
Try going back eight MONTHS with a Linux binary and see what happens.
And THAT, my friend is the problem with Linux. If people would just stop moving shit around and breaking libraries, headers, dependencies, blah blah, maybe more programmers would use it. Most of us don't really want to learn eight hundred different distros only to have our apps break when libc7 comes out -- we're already too friggin' busy writing code. Just give us a platform that works today, tommorow, and a decade from now.. exactly the same way.
And when I say "exactly the same way" -- I mean as far as the software and documentation are concerned. Go ahead, add shit, just don't change what's already there. How hard is that?
Christ.
> This discovery also marks the possibility of
> stars that look strikingly like planets.
We already have those.. examples include John Goodman and Roseanne Barr.
I don't think SunOS ran on PCs, but I'm pretty sure it ran on the Sun/386 workstations.
> I do wonder if they've closed off the syslog UDP port, though.
FWIW, the default configuration is NOT to listen on the UDP port. If you want that, you have to explicitly launch syslogd -r.
>> You don't have to copy the packages. "pkgadd -d ..." works fine.
> I noticed this option, but pkgadd seemed to consistently ignore it...
Works fine on other versions. Did you specify the directory where your packages are stored [right], or did you specify the directory where your package is stored [wrong]?
Oh, and as for the installer. Again, on other versions, if you install from the "Install" CD with a supported video card, you get an X installer (actually, I think as of 7 or 8 it's a nasty java->X installer). If you toss that disc and just shove in Software 1/1, you get the good (curses) installer.
> It CAN take days to verify that the logic is ...and then writing three other ways to do the same query, and shooting them through the optimizer to see which one will run better on the same set of data..
> correct and that the record set you are getting
> back is in fact what you think it is....
Uh huh, I hear you.
I think *anybody* who doesn't at least pause to think for a couple of hours on a query returning millions of rows is asking for trouble.
Oh, look, this cartesian join works fine for ten records in the lab -- let's ship it!
This is not an MS-specific thing.. certain contests here require that you solve a skill-testing question.. Nothing difficult, usually it's like 47 * 8 + 2 / 4... I forget the legal reason for this, but I've always suspected it was to keep Americans from winning our prizes.
> I remember back in high school, [...]
> [...] an HP Presario.
God DAMN, and you have a 6-digit UID?
Jesus Christ.
> I think we should have some sort of "SlashDot
> old-timers" thread somewhere on this site.
Maybe Signal 11 could swing by, that would be entertaining.
> I've found that Unicode values do not always work as URL parameters.
HTTP POST?
Never tried, myself, living in locale LANG=C
> how does he know what ear wax tastes like in the first place?
Inadvertant consumption of Bertie Bott's Every Flavour Beans.
> Then you're shit out of luck
;)
Man-o-man. STILL? A little birdy told me that "all that transaction stuff" had been taken care of in MySQL 4. Glad I didn't believe him.
Any kind of locking yet, which doesn't involve locking the whole damned table just to atomically update a row?
> If you are talking 5G of data or less on a single box,
> realize that Sybase ASE 12.5 for Linux is free
Hmm, that might be enough reason to switch to Linux, assuming I can find x86 hardware I like. I'm sure it must exist by now.. all I want is high I/O bandwidth, absolute reliability, hardware fault reporting, and serial console access. Didn't exist last time I looked (half a decade ago). Oh, and it has to be price-competitive with SPARC boxen; I doubt that will be an issue.
5G is pushing it, but we could make app changes to allow that.. Core DB is around the 2GB mark, then transaction logs start to grow... every 6 months or so, they get flushed.. but our retention mandate is only 40 days.
Thanks for the tip! I'm familiar with the SyBase of yesteryear. It was adequate then, I'm sure it will be fine now. Like most organizations, we don't need much in the way of an RDBMS -- just atomic operations, good backup/recovery plan, and ability to hire high-priced help if the shit hits the fan. That third category is one reason why FOSS DBs are a little weak in my eyes -- we'd have to have good support in house.
Hey, thanks for the feedback!
Is there really an advantage to running Hibernate on the java side of things when the JVM is only running to run JDBC, on the same box as the RDBMS? One would think that the RDBMS could make better caching decisions about data [locality of reference] than the application layer -- and you'd be looking at two caches now, instead of one. My usual goal is to have enough RAM around for the RDBMS to rarely need to hit the disk under normal application load. Hibernate has one concrete disadvantage from where I sit -- reports run against the RDBMS engine (i.e. SQL*Plus command line) are not always up-to-date, which also makes me nervous about crash recovery.
On the other hand, hibernate has the advantage that I have developers somewhat familiar with it (it is used in another application suite we have), meaning no ramp time as would be the case with ODBC.
I also have strong abstraction from the database layer. We wrote an abstraction layer in 1997 when we were using a small-time RDBMS which subsequently went under. The switch to Oracle in 1999 was a piece of cake; write a new server, badda-bing, badda-boom: clients can talk oracle now, all apps work.
.. FOR UPDATE; UPDATE; ROLLBACK|COMMIT; paradigm. And easily support hot backup snapshots. Oracle almost does this for you with archive logs, although sanity requires coldsnapping it every six months or so.
Fortunately, Oracle's still doing the job for us, but new installs are getting pricey. We are seriously looking at either PostgresSQL or MySQL for future work. Either will work, as long as they support the SELECT
I wish jdbc/odbc had been around in 1997 (on BSDI and Solaris). I'm thinking that maybe I should replace the client interface guts with an odbc abstraction.. Hmm.. I wonder what the performance hit would be?
You could always use the anonymization service at anon.penet.fi.
It's cold outside, there's no kind of atmosphere...
Didn't they already do that one?
The answer is obvious... if I'm listening to tunes on my Palm, I can't surf the 'net.