Given that clients were suing Scient for delivering crap, and all the engineers I know laughed at Scient's pathetic salary offers, and the employees were obsessed with The Process rather than actually shipping things;
Aolserver has been out since early 1995/late 1994, when it was called Naviserver. AOL bought the entire company (Navisoft) just to get server (and in-house support). They wrote it from the ground up; I should also point out that in 1994, Apache didn't exist yet.
The original focus, which remains the same today, was pooled database connectivity in a threaded webserver. What this means is if you interact heavily with the database, the overhead of db client creation is removed. Single process architecture makes caching simple.
Apache only recently gained db pooling, and caching is still done out-of-process, eg, Squid. Naviserver had this back in 1995. Given that Apache is process-based, not threaded, Apache's pooling is still inferior.
Don't switch just for language wanking; that means nothing for user experience. But for a site with heavy database interaction, aolserver can show an order of magnitude performance improvement over Apache.
Visit philg's intro for more background. (Oh, and if you can't stand Tcl, Perl & Phython versions are being developed.)
Philip Greenspun saw this server, liked it (he really likes Tcl for some reason),
You're a dork. No one 'really likes Tcl'. Philg likes pooled database connections and a single-process (threaded) architecture for easy caching. Apache still doesn't have that.
Allow me to repeat - the issue is not language wanking, it is strong DB tools.
The best quote is from their table of contents, under the "Development Tools" section:
DO LINUX DEVELOPERS UNDERSTAND.NET CHANGES TO VISUAL STUDIO?
WHEN DO THEY PLAN TO CHANGE TO.NET VISUAL STUDIO?
I just fell over laughing. Are linux developers aware of significant changes to development platforms they don't use? When do they plan to throw out all their existing tools for a system that is totally incompatible with their existing systems? Are they salivating to replace their setups with a system that is still vaporware?
God, I might buy this report just for the humor value. Rock on, EvansData! You might have a career in comedy after all....
When the Allies captured Heisenberg & the nuclear team, they placed them under house arrest, for several months, and (secretly) taped all their conversations - which were many, loud, and angry. The logs have since been published; I forget the title, but if you can find it, do. It's a really illustrative example of the many conflicting interests they worked under, and why -- a more concrete analysis of the subject of this thread.
I would really like an 'update comment' function, so I could come back and add the title...
I'm not a security expert but i have some experience implementing distributed systems that use HTTP, and you can easily use SSL (or any stronger PKI based authentication system) to integrate security into your services.
The point is not protecting the content of the data stream from tampering, the point is that random, [untrusted] apps are making procedure calls on your system -- and you can't block them at the firewall. Is BackOrfice running around on your NT network? Block the port it communicates over at the router/firewall. Legal objections to P2P servers? Block the port. Running a closed-source package that 'reports back to the mothership' - block the port.
Now run all those over SOAP -- all of a sudden, your ability to control these actions over your network are trashed. Not impaired, but totally fucked. The amount of processing you need to throw at inspecting/analyzing/blocking every packet over port 80 will be financially inaccessible to all but large companies.
Again, the problem is not that programmers are incompetent, or that connections are unencrypted; just that maintaining a secure network, in the face of hostile applications (both inside and out) demands filtering rules for TCP/IP. Tunneling everything over HTTP threatens to damage such efforts in a really, really, bad way.
I was actually just hoping for a flame(just in the mood), but a good discussion is better...
Fortran90/95/2000 does not have 72-char line limits, has derived types, pointers, dynamic memory allocation, overloading... most of the stuff you've spent 30 years developing. What it doesn't have is a lot of crap that is unnecessary for scientific programing.
But the only free compiler only handles F77, which was one of my points. In response to your question, GCC came out in the mid-80's. Before that, of course, you could use the Bell Labs C compiler that came with Unix. A key to the prevalence of open source, and presumably software innovation, on computers before PCs was that compilers were accessible to, if not all, a great many (And pre-mid-80's, real work was done on time-shared machines).
No insult here - basically you're working with a dead language -- The vast majority can't afford to read/execute it. Standard specialist situation though - a small elite of people with access to forgotten tongues. The advice, your advice, "Use the best tool for the job." rings true. (I get most work done in LISP, SQL, C, & Perl. Lately Ruby.)
Language disputes are only suited for flamewars; serious discussions are wasted (design is more important anyway).
How much does a good commercial C++ compiler cost?
From a few hundred to a couple tho. Keep in mind, however, that all of them keep to modern flavors, however -- so you can still use g++, but it'll take longer to compile, and execute slower; but you won't be restricted to a 24 year old language spec...
Almost every single major computational chemistry code is written in a dialect of Fortran
Interesting; I didn't know that. My Chem grad friends tell me they get all their work done on packages written for Macs, but don't specify any languages(they're in organometallic catalysts).
When I pursued my Physics degree we wrote in C. (the also-programmer element, like myself, wrote components in C and tied them together with scripting languages. Incidentally, this approach is best in the real world, too.) The F77 limitations were very real - none of us had F90 at home, limited access to research machines, etc. Plus, the wealth of published physical algorithms written in C continues to creep up on those written in FORTRAN. And learning the syntax of a language is different from learning to program well in a language, as I am sure you know.
When I worked in a bioinformatics lab, the code there was written in C, and they were trying to move to Java...but then again, they were a little crazy.
So I don't think FORTRAN has a lock on sci computing.
My only real complaints are that FORTRAN is infeasible to most people and thus a burden to teach, esp. when other languages offer comparable speeds & more features. I only flamed since the previous poster presented me with such a wide target. Use what you use. -- oh, but avoid C++.
As a point of curiousity, where are you applying your chem skillz at the moment?
The Fortarn code runs faster and the development time was 1/10 of what it would have been to write in C or C++. Your point is?
Well gee, I suppose once you get past the 72-char line limits, the dearth of types, the pathetic flow control, and the complete lack of objects, higher order functions, lazy eval, complex data structures, & everything we've spent 30 years developing... you're left with code that almost no one can or will read. Nor can they get compilers. Writing in a flavor past F77?? Cheapest compiler - $1000. Is that what you might call student priced?
Anyway, your magical '10 times faster' FORTRAN skillz don't really help you since you have to work with a team. A team, mind you, who are now all using Matlab, LISP, or Perl-with-C-mods. Oh, and they have a more pre-debugged algorithms available on the web, and better code profilers for tuning. So how's your amazing, faster than assembly, execution speed now??
FORTRAN was cool if you were an engineer when your choices were FORTRAN, COBOL, or LISP. Times have changed; and you aren't writing Apollo control systems in 128k. So fuck off and get a real language. Your undergrads will thank you, or at least forgive you.
I thought the point was that Enoch *did* die. I mean, he gets shot; coughing up blood he's babbling about getting to the church to get his gear; R.von.H. shows up with his old cigar box - a box related to his status in this ancient secret 'Societas Eruditoium'. Lods of foreshadowing on this box -- Shaftoe even has smack dreams about the contents of the box. Anyway, R.von.H shows, uses the mystical contents of the box, revives Enoch, drives him away.
Secondary effects: the woman inherits Enoch's military death benefits & takes his son to England. Enoch is legally dead -- no one is going to come looking for him while he does his secret shit in the Phillipines. So, even if I'm wrong, and it was just a fake death, there are your reasons.
30's w/ wife + kids? That horrifying. If anything, this guy presents an image of "when you reach this age, this is what you *don't* want to be."
*shudder*
Re:Rampant homophobia? Not necessarily.
on
eFront From Inside
·
· Score: 1
I totally agree with you that we don't need to be so hypersensitive, but there is a point of difference here. I don't think this has to do with 'degrading the gay community' or any such; nor any kind of PC whining. This is just a criticism of the gamer community (and a knock on their age) -- why would the gamer kids reserve 'gay' as their greatest insult, even as it has become a non-insult (in the population @large) over the last decade?
Eckel tried to insult PA by calling them 'gay'. These days, that's not much of an insult. In my neighborhood in San Francisco, the response would be "yes, and??". Most other places outside of the Bible Belt seem to feel the same.
So, are the warez doodz just under socialized? Do we need to take them to hot, hot, gay raves? How do we break through to them that the 80's are over?
Of course I'd much prefer operating systems via console, but saying that you can't remotely operate it isn't true at all.
I wasn't saying that it's *impossible* to use Windows remotely. I've worked with a few packages for that, and considering what they're up against (MS), they're pretty nice. But it's just not in the same ballpark as Unix, which is designed for remote users; futhermore, going down this side-path pulls us away from my question:
Does anyone else here think that unix (as compared to win/mac) provides a feeling of a virtual space, which induces an emotional attachment that makes us better (or more involved) users?
Not that our Un*x boxen are inherently any better. We just seem to "care" more about knowing what our servers are actually doing.
All your other points are correct; I just noted something with the above. We *do* care more about what our unices are *doing* than the MS machines. I venture that there is a psychological reason for this - since we can ssh/telnet into a un*x box, use X over the network, etc, easily process jobs across multiple machines (PMV, etc), heavily script; there is more a sense of *location* associated with a Unix machine than a MS machine, which goes for a TV-style interface, and you can't operate on in without physically *going* to it. Additionally, the increased use of the keyboard on unices gives extra tactile interface. Adding it all up, a unix box, even remotely, gives a sense of 'place' to the user/admin.
Consequently, of course we care more about what it's doing! Those other processes/users are our neighbors! And even recreational admins spend lots of time tuning their boxen; is this analagous to perfecting your living space?
I'm straying off topic. In short, unices give a virtual sense of place, which then ties into all sorts of hardwired emotional responses for dealing with our physical environment.
Am I nuts, or is this good advice for a UI designer??
Unfortunately, everyone uses the term 'consumers' these days. It seems encoded in our legal spirit -- citizen's rights are void, public action is undertaken to defend 'rights of consumers'. WTF??
I have recently been correcting everyone who uses 'consumers' when they should mean 'citizens', but this approach doesn't sound like it will filter up to the Feds. How can we attack them with the correct meme??
I thought we were talking about high-stability, high-performance, high expense DB products. Most OODBMS's are 1.x iterations of what is still, essentially, a research idea.
Are you really going to write your own text indexing system if Oracle's ConText/InterMedia is available?
Hell, Yes!! Use PLS!!
Doesn't Oracle claim to give you enormous performance benefits for the web if you use their products end-to-end?
If by 'end-to-end', you mean all the way to port 80, then I would have to say that they're full of crap. Oracle's silly java-webserver-inside-the-DB can't touch a well-tuned Apache or Aolserver instance (or even an untuned one...). Are Oracle engineers at the forefront of web clustering? Do they offer premier environments for whichever language in their servers? etc, etc, etc..
Generally, tech wisdom holds that most companies have one good idea; everything else they make is mediocre. Oracle makes probably the best all-around DB out there; most of the rest just sucks.
Aside from that, you're completely correct. 'Keeping it portable' is a practical impossibility for a high perf DB; and the investments in time & people should really keep you stuck to your vendor for 5-10 years.
America: Everything is permitted, except that which is forbidden.
Germany: Everything is forbidden, except that which is permitted.
Soviet Union: Everything is forbidden, including that which is permitted.
Italy: Everything is permitted, especially that which is forbidden.
The only Linux kernel panic I had was from bad hardware too. Most notably a loose SIMM slot, not suprising it panicked when half its memory disappeared..
I wouldn't call that a kernel 'panic', so much as a kernel 'full-on vertigo terror'.
I think I'll go add that to my 2.4 source tree. Just a shame I won't see it...
You're right that a utilized Microsoft machine has a fraction of the stability of a new install. (But remember the time counter bug last year - the machine would crash every 48 days or so, used or not...) The other poster gave some specifics about his crashes - namely, that he's never seen a NT2000 BSOD. I haven't either; I've hung it half a dozen times, but no BSOD. I assume it's the windows equiv of an app calling 'XGetPointer' and then crashing. Of course, on a Unix I can ssh in and fix it...
I'm forced to dual boot b between Linux and Windows, and I have *never* had the Linux kernel die on me.
I've been keeping track of kernel panics I've had on Linux. Since 1.4, I'm up to 5. And I think 3 of them (on 2.1) are attributable to a bad motherboard. (It had a tiny crack in it, I found later. A few rare times, thermal changes opened it while my machine was running.)
I have not bothered to track BSODs -- I don't think I need explain why.
Slashdot Mirror
I wonder how such a company might have a future?
I need to clean up some parts of Cole Valley.
The original focus, which remains the same today, was pooled database connectivity in a threaded webserver. What this means is if you interact heavily with the database, the overhead of db client creation is removed. Single process architecture makes caching simple.
Apache only recently gained db pooling, and caching is still done out-of-process, eg, Squid. Naviserver had this back in 1995. Given that Apache is process-based, not threaded, Apache's pooling is still inferior.
Don't switch just for language wanking; that means nothing for user experience. But for a site with heavy database interaction, aolserver can show an order of magnitude performance improvement over Apache.
Visit philg's intro for more background. (Oh, and if you can't stand Tcl, Perl & Phython versions are being developed.)
Allow me to repeat - the issue is not language wanking, it is strong DB tools.
God, I might buy this report just for the humor value. Rock on, EvansData! You might have a career in comedy after all....
I would really like an 'update comment' function, so I could come back and add the title...
Now run all those over SOAP -- all of a sudden, your ability to control these actions over your network are trashed. Not impaired, but totally fucked. The amount of processing you need to throw at inspecting/analyzing/blocking every packet over port 80 will be financially inaccessible to all but large companies.
Again, the problem is not that programmers are incompetent, or that connections are unencrypted; just that maintaining a secure network, in the face of hostile applications (both inside and out) demands filtering rules for TCP/IP. Tunneling everything over HTTP threatens to damage such efforts in a really, really, bad way.
No insult here - basically you're working with a dead language -- The vast majority can't afford to read/execute it. Standard specialist situation though - a small elite of people with access to forgotten tongues. The advice, your advice, "Use the best tool for the job." rings true. (I get most work done in LISP, SQL, C, & Perl. Lately Ruby.)
Language disputes are only suited for flamewars; serious discussions are wasted (design is more important anyway). From a few hundred to a couple tho. Keep in mind, however, that all of them keep to modern flavors, however -- so you can still use g++, but it'll take longer to compile, and execute slower; but you won't be restricted to a 24 year old language spec... Interesting; I didn't know that. My Chem grad friends tell me they get all their work done on packages written for Macs, but don't specify any languages(they're in organometallic catalysts).
When I pursued my Physics degree we wrote in C. (the also-programmer element, like myself, wrote components in C and tied them together with scripting languages. Incidentally, this approach is best in the real world, too.) The F77 limitations were very real - none of us had F90 at home, limited access to research machines, etc. Plus, the wealth of published physical algorithms written in C continues to creep up on those written in FORTRAN. And learning the syntax of a language is different from learning to program well in a language, as I am sure you know.
When I worked in a bioinformatics lab, the code there was written in C, and they were trying to move to Java...but then again, they were a little crazy.
So I don't think FORTRAN has a lock on sci computing.
My only real complaints are that FORTRAN is infeasible to most people and thus a burden to teach, esp. when other languages offer comparable speeds & more features. I only flamed since the previous poster presented me with such a wide target. Use what you use. -- oh, but avoid C++.
As a point of curiousity, where are you applying your chem skillz at the moment?
Anyway, your magical '10 times faster' FORTRAN skillz don't really help you since you have to work with a team. A team, mind you, who are now all using Matlab, LISP, or Perl-with-C-mods. Oh, and they have a more pre-debugged algorithms available on the web, and better code profilers for tuning. So how's your amazing, faster than assembly, execution speed now??
FORTRAN was cool if you were an engineer when your choices were FORTRAN, COBOL, or LISP. Times have changed; and you aren't writing Apollo control systems in 128k. So fuck off and get a real language. Your undergrads will thank you, or at least forgive you.
Secondary effects: the woman inherits Enoch's military death benefits & takes his son to England. Enoch is legally dead -- no one is going to come looking for him while he does his secret shit in the Phillipines. So, even if I'm wrong, and it was just a fake death, there are your reasons.
*shudder*
Eckel tried to insult PA by calling them 'gay'. These days, that's not much of an insult. In my neighborhood in San Francisco, the response would be "yes, and??". Most other places outside of the Bible Belt seem to feel the same.
So, are the warez doodz just under socialized? Do we need to take them to hot, hot, gay raves? How do we break through to them that the 80's are over?
Okay, enough /.ting. Time to play some gamez!!
argue about our lifestyles
no proof about skills
move along, silly person.
Does anyone else here think that unix (as compared to win/mac) provides a feeling of a virtual space, which induces an emotional attachment that makes us better (or more involved) users?
Consequently, of course we care more about what it's doing! Those other processes/users are our neighbors! And even recreational admins spend lots of time tuning their boxen; is this analagous to perfecting your living space?
I'm straying off topic. In short, unices give a virtual sense of place, which then ties into all sorts of hardwired emotional responses for dealing with our physical environment.
Am I nuts, or is this good advice for a UI designer??
Seriously. Visualize 10TB DBs in GemStone. Do that, then we'll talk.
Oh, and see if you can get those OODBMS people to grasp querying rather than navigation.
I have recently been correcting everyone who uses 'consumers' when they should mean 'citizens', but this approach doesn't sound like it will filter up to the Feds. How can we attack them with the correct meme??
Different thread. Move along.
Generally, tech wisdom holds that most companies have one good idea; everything else they make is mediocre. Oracle makes probably the best all-around DB out there; most of the rest just sucks.
Aside from that, you're completely correct. 'Keeping it portable' is a practical impossibility for a high perf DB; and the investments in time & people should really keep you stuck to your vendor for 5-10 years.
America: Everything is permitted, except that which is forbidden.
Germany: Everything is forbidden, except that which is permitted.
Soviet Union: Everything is forbidden, including that which is permitted.
Italy: Everything is permitted, especially that which is forbidden.
When you crave a burrito, come to El Balazo in the Haight. Or the grocery in Mountain View at the intersection of Rengstroff & Old Middlefield.
But what I *really* hurt for is a Philly from Delaware Sub shops in Austin.
What did you leave behind?
I think I'll go add that to my 2.4 source tree. Just a shame I won't see it...
You're right that a utilized Microsoft machine has a fraction of the stability of a new install. (But remember the time counter bug last year - the machine would crash every 48 days or so, used or not...) The other poster gave some specifics about his crashes - namely, that he's never seen a NT2000 BSOD. I haven't either; I've hung it half a dozen times, but no BSOD. I assume it's the windows equiv of an app calling 'XGetPointer' and then crashing. Of course, on a Unix I can ssh in and fix it...
I have not bothered to track BSODs -- I don't think I need explain why.
I'll keep you posted.