Convenience and security are often at opposite ends of the computing spectrum.
The UNIX security model is not mass market friendly. The end user cares about security, but is not going to the work required to maintain it. The average joe can just barely be encouraged to lock his door at night, but only after decades of mass media has convinced him that if he doesn't, he'll allow a deranged lunatic to come into his home and rape and kill his wife and kids. Nothing that can happen on your computer compares to that, so people just won't be bothered to take the same precautions.
Need proof? Fine. Install Windows XP for someone; give them an administrative account and a restricted account. Tell them to only use the administrative account to install stuff, and use the restricted account for every day stuff. Come back in a few months.
What do you find? They hit one "access denied" message on the restricted account and have used the administrative account ever since.
Forcing a security model that the customer clearly rejects does no one a service.
This is unfortunately a case where treating the endless symptoms of a disease, while far from optimal, is better than trying to cure the disease.
They care about applications. Killer applications.
Windows has >90% marketshare and it costs a non-trivial amount of money to develop an application that works across the board.
Adding MacOS and Linux versions is often difficult enough to justify.
Even when a number of competing hardware platforms implement the same API, the per-hardware platform bugs/test scenarios can be cost prohibitive. Java's a good example of this.
What application is killer enough that someone can stand to rewrite/QA to support.1% marketshare platforms?
Windows has no technical advantage over alternative platforms; few would argue with that point. The windows monopoly comes from people not being comfortable administering and tinkering with their computers.
I'll agree with you, Windows offers no technical advantage over other platforms, but it does offer one really really compelling business advantage.
By simply targetting Windows, I can generate one binary that will run on approximately 500 million PCs, give or take a couple of million with weird configurations.
Target Linux? Well, sure, but the Linux version won't be anywhere near as user friendly as the Windows version since I'll have to put the burden on the user install the dependencies, patch their kernel (the big 10 distributions don't have out of the box support for what the app does), script it to start up the way they like it, etc. Would you trust a third party application to patch your kernel? I wouldn't. Will I ever recoup the porting costs? Probably not, which means I'm not making a good business decision by making the port, but doing it anyway since I want to use the app on my workstation at home.
Those things one does through the registry are things Microsoft does not want the average user to understand.
I just finished writing an application that saves state to the registry. It's actually really handy, since I don't need to manage config file loading, parsing, saving, etc. The fact that it's not stored in a file makes it less likely to be swept up into a different directory. It's always in the place that I expect it to be.
Can a regular joe user figure out what my application's keys in the registry do? Well, sort of. It's a pretty simple application. Do they gain anything by tweaking registry values?
Not much, unless they want to see the post-install introduction screen again.
I'd say I understand it pretty well.
Additionally, I can take any running application and trace it and see what keys it is modifying, using free tools. The fact that every application does this the same way means it's pretty easy to isolate the library calls that update a configuration.
Did you know that Explorer saves the position of almost every single window in the registry, every time you move it? No doubt a sinister plot to remember where you like to keep your windows so that they can be in the same place, next time.
There's this architect I know of who maintains a 10 year old SGI workstation running some ancient CAD program. I asked him why he goes through the trouble, and he became LIVID. It'll be a cold day in HELL before he installs Windows and that Autocad garbage, apparantly.
Then there's also this interactive media artist I met once. He hand compiled his entire system from scratch, modified the video4linux driver to get better performance, and claims that he hasn't touched an Adobe product in years. I asked him why he wasn't using a Mac, and he went into some tirade about how some program was discontinued once and he couldn't find anything that would read his old saved work, and he swears he'll never put himself in that position again, open source all the way (I didn't ask for the details).
Both of these people appeared very serious about their craft.
So when someone tells me the eccentric outside of the box thinking individuals install MacOS X, I don't take it too seriously.
Must every word we write be played back in monotone, completely out of context, character, to see if it can be used against us?
Anyone who understands hacker culture, or Bram's personality, would read that and laugh.
God, I probably have something just as ridiculous saved somewhere. In fact, I know I do, and I know someone else got ahold of it and spruced it up with Photoshop and made it look like a piece of communist propaganda that I'm just about ready to print out a million copies of. I come across it every so often and laugh, and anyone who knows me would laugh if they saw it.
And a court would say that clearly this is the mind of a terrorist.
Microsoft is successful because they understand how to open their platform.
Seriously.
Every other platform developer is too proprietary -- they're so afraid of leaving money on the table that they try to control EVERYTHING. They can't see the larger picture. They don't know how to monetize open platforms, they capture some modest marketshare and charge everyone to death to do anything on them, but ultimately they never become a standard, and the developer spends resources targetting yet another platform instead of making their products better.
Deep down, despite all of their shortcomings, Microsoft's a lot better than the alternatives. Read technical literate back when Microsoft was starting out -- there was a clear division between those resisting standardization on a single platform (for entirely valid reasons, I'll admit) and those who were frustrated by the high costs of targetting multiple moving (open?) platforms, or locking themselves into a platform that was way too proprietary to ever let anyone gamble on risky applications.
They defined the software product industry by striking the sweet spot, a balance between overly restrictive and unsupportedly open. Maybe it's because they have techies at the top of the company? Whatever. No one understands how to do this better than them. You can target Microsoft's platform and have a customer base of, what, 500-750 million PCs?
Monopolies aren't so great, but we could be stuck with far worse. Imagine someone took a cut of every e-commerce transaction?
I saw someone pick up a Tablet PC the other day. He played with it gleefully, and showed me the coolest app: Microsoft Journal.
Behold: it looks just like a marble notebook page that you can write on with a "pen" thanks to the touch screen! It allows you to save notes and even try to digitize the text! I remarked that it was quite indeed an impressive piece of technology for only $3000.
When am I going to get one? Probably not anytime soon, since I've discovered a much better solution for myself: a small attache journal and a fountain pen. Total cost: about $50 (I splurged on the pen).
Check it out, with the journal I can:
Quickly jot down a phone number
Scribble a username/password that some shitty download site made me invent (sun/java) so I can come back later in case I download the wrong file
Make up to-do lists
Schedule items on said to-do list
Intra-office messaging (scribble on a page, rip it out, leave it on someone's desk)
Keep track of confirmation numbers after you make a stock sale or order cable service or whatever
Note what exactly you FedExed (you note the tracking number, too) to who so you can really say with conviction that you sent Schedule C to your accountant
Scribble down thoughts as they relate to a business deal, or personal matter, or whatever. Helps you keep an eye on the goal, and maintain a consistent position.
Draw little pictures to kill time
It's always on me, requires no external power source, and is quite pleasing to use. No spyware, viruses, hackers, hardware failures, or bugs to worry about.
Plus, since the pen is a indispensible part of the equation, you always carry it on you, and can sign your name with a flourish on a credit card receipt, or otherwise display the quality of your pedigree to impressionable onlookers.
I've also found that, at least when I speak now, I slightly better organize my thoughts before I open my mouth since it's a pain in the ass to re-word a paragraph that you've half-written. This journal dialogue transfers to my vocal dialogue, for some reason.
Oh, also, cursive handwriting is the shit. I forgot how much I loved it.
Yes, it's the ultimate personal assistant. Between you and me, I think these things are going to be BIG!
"Here we were, working 90 hours a week to build something to help Linux, working long hours, giving up a lot of money and time, and all the time we were doing that, at every step of the way, we had people abusing us, sending us nasty e-mail, telling us we're jerks. It's been constant. It gets tiresome. After five years of that you just say: It doesn't make business sense and it isn't that fun."
Translation:
"For years we enjoyed a lot of low-cost publicity for BitKeeper because the Linux kernel team was using it. We could go to IBM and HP and say `Look, these guys use it, and they're really discriminating programmers. So, that'll be $2000 per seat.' It costs us relatively little to donate a server to the kernel team, and the occasional nasty email is easy to delete. All in all, it was a successful campaign until TRIDGE CAME ALONG and almost made us obsolete -- at that point we'd be crazy to give him any chance of actually polishing it off, so we yanked the plug."
Standing on the shoulders of giants...
on
McVoy Strikes Back
·
· Score: 1
Depending on how you define innovation, one can argue that it's rare to find it anywhere. Everything is built on everything else. Is BitTorrent innovative? I'd say yes, it's the first independent app that offers a truly powerful P2P-accelerated download. Someone may say it is not so revolutionary, it is just an incremental step from concepts found in Kazaa, or eDonkey. And you know what? They're right!
Is BitKeeper innovative? Someone might swear by whatever cool feature it has, and proclaim that it's revolutionary (I've never used it), but I could just say that it builds on concepts in CVS/RCS/SCCS. Without those predecessors, BitKeeper might not even exist.
I digress, lets attack the real misconception: shrinkwrap software products are not universally viable business models! There's Microsoft, which is the industry big-player, Adobe, and then thousands of little boutique companies that make peanuts on selling boxed software and make their real profits through consulting and support.
The model is hardly viable for proprietary software companies, so it shouldn't suprise anyone if it doesn't make sense for open source software companies. Red Hat's product model is based on convenience, or as Bob Young would say, the "Ketchup Business Model". Any idiot can make ketchup at home -- it's really easy, easier than downloading and burning an ISO. But no one does.
Red Hat isn't so lucky to be like Heinz in this regard, so they make up the bulk of their income through service and support. And so does every single other shrinkwrap software company not named Microsoft.
Go out and buy a journal. A really nice one costs $15. Buy yourself a fountain pen. A halfway decent one costs $25. Dust off those memories on cursive handwriting. The writing flows smoothly and I find it to be quite pleasant, relaxing even. Ink cartridges last about a week apiece.
I must take two or three pages of notes a day, that have nothing to do with work. Really useful for looking up what you were thinking about last month, errant phone numbers, confirmation codes, and the like.
Fits in a jacket pocket. Always on. No software to manage.
I feel vulnerable if I leave the house without it.
It's a time tested tactic. People who were considering building a system on top of Google Maps suddenly put those plans on hold if they know that Microsoft is going to wipe them out soon enough.:D
Cause you're just going to be rewriting it every month anyway.
I've noticed that a lot of people are absolutely terrified of writing code. They don't actually want to think about the problem, they just want to put a black box in its place and say mission accomplished.
It's a nice ideal, but I'm not sure the developer wouldn't have been better off solving the problem uniquely, exactly, and effectively without involving some other beast of a module that has its own release cycles, maintenance hassles, etc.
It may take me longer to solve the problem myself, but it feels like time better spent because I know what's there -- it's not a black box.
I can take ownership of the problem..... bingo! Just answered my own question.;)
What it really comes down to is using your time effectively. If you and I were to write the same application, but I used a proper RDBMS and you used MySQL, my application would be done sooner because I wouldn't have to reinvent the wheel (or, your application would be less stable because you didn't bother with referential integrity checks, transactioning your data updates, etc). You app might be a bit faster (doubtful if we're talking about non-trivial datasets, and especially not if you've reimplemented many RDBMS features in your application layer), but it's highly unlikely the end user will notice that you're doing 5000 transactions per second and I'm only doing 4500.
Not installing and managing an RDBMS is my idea of using my time effectively.
Customer wants a small CRM? CGI under Apache that writes DBM files. Takes a day or three.
Customer wants a large CRM with 10,000 users? Build around an async I/O library that speaks web, maintains an in-core database, flushed to disk every 30 seconds. Takes a week or three.
Customer wants a really large CRM with 100,000 daily users? Build multiple sites in the previous fashion, one to handle accounts, one to handle contacts, one to handle queries, etc. Tie them all together using the magic of cookies and internal or external redirects. Takes a month or three.
You, my friend, have crappy hardware. Or choose your kernels very badly.
Blah blaah blaah.
Would I have offended less people if I said that our servers running OpenBSD were crashing?
How long was I supposed to make the original post? Let me go on...
In the ENTERPRISE, you get to learn when ${CURRENT_DISTRO_VENDOR} says STABLE, they really mean STABLE-ish.
In the ENTERPRISE, fucking idiots come out of the woodwork and cause some political battle over choosing ${CURRENT_DISTRO_VENDOR}, and that ${HIS_DISTRO_OF_CHOICE} servers have been running for years without any downtime doing the same exact job. Sort of!
In the ENTERPRISE, you end up compiling your own kernels from scratch because you learned that going with the distro's kernel should work in theory, but just doesn't.
In the ENTERPRISE, you get to learn that if you hit BIND and syslog with a thousand messages a second, they deadlock, and good luck managing to ssh in when syslog() calls blocks forever because ssh is trying to record your login!
In the ENTERPRISE, you learn that the syslog daemon you picked, of all fucking things to worry about, doesn't have a -n option to not resolve hostnames. So you use LD_PRELOAD to make gethostbyname() always fail.
In the ENTERPRISE, you get to learn that almost every shipping company will cause subtle but fatal damage to your servers if you ship them across the country!
Oh my, you're right! If that happens, it means I might have to... USE MY BRAIN AGAIN! *GASP!*
Why does anyone expect something that worked fine with 10 users to work fine with 100,000? It'd be swell if it did work out that way, but do you always write a quicksort if a bubble sort will do just fine, knowing you can switch to a quicksort later on if it becomes a problem?
"Enterprise" projects are so full of unique challenges, I don't see why something like the way you store your data makes a fuckola of a difference in the grand scheme of things. Everything gets rewritten a few times anyway. Even eBay.
It's much more harmful when you get into the mindset that application A automatically means data storage scheme Y, or specifically, Brand O's data storage scheme Y.
It's not that simple in the real world. Think about things like concurrency. If you have 200 clients running your app at the same time and 10 of them are trying to update a complex set of related tables, how do they avoid stepping on each others toes? Are you going to have your client apps communicate with each other and coordinate the inserts? Or just lock all the tables and kill performance? How do you ensure that everyone who is reading those tables sees a consistent data set?
Check thisout: you _ s e r i a l i z e _ the requests!
Why, yes, serializing the requests when they each take 30 seconds would blow concurrency out of the water, but call me strange -- when I say to myself "hey, this thing takes 30 seconds to complete", I don't say "if my RDBMS had all of this insane complexity built into it, I wouldn't have to care that it takes 30 seconds!! DAMNIT!"
Everyone's so used to running things in their own virtual process with unlimited memory and unlimited wall clock time and having your write() block because of some guy's slowassed modem, that if you spend an hour thinking of a different world, everything that implies an RDBMS suddenly disappears.
Alas, the solution is a really hard pill to swallow: figuring problems out for yourself!
Far be it from me to fault anyone that needs an RDBMS to keep their paycheck coming.
That's all true, up until the point where your database craps out on you (you've seen it happen to Slashdot frequently, Penny Arcade, and pretty much every MySQL-backed website out there). Then it's painfully obvious to the user. However, this isn't about the end user. It's about the programmer, and using the proper tool for the job. There's absolutely no reason you should have to replicate the functions of an RDBMS in each and every one of your applications, and if you find yourself doing so it's pretty clear that you're using the wrong tool.
Blaaah blaah blah. It occurred to me recently that the only reason I even feel the need to use an RDBMS in a web application is because under Apache, it was easier to store data in an RDBMS than gain permission to store it in the filesystem in some concurrency safe way on a shared hosting environment.
That environment is so dangerous that I took these assumptions with me when I had 10 servers with root access. :/
The best change I ever made for my productivity and sanity was writing a tiny little module that reduces the entire RDBMS into a circa 1985-style dBase file. It's great! The entire module only has a handful of SQL calls, and the application is like SQL WHO?! The solution to my concurrency problems was to only allow one process to hit the db module at a time.
What? It won't scale, you say? I guess when the site starts having a hundred hits a second, I can worry about allowing some processes to have shared read only access and block writers until they have exclusive access. Who knows, maybe when there are a thousand hits a second, I'll have something like commit/rollback! Or I could just rewrite the fucker from scratch!
At least, I would have said that until I discovered that I didn't even need to use Apache, that I didn't need to keep around a flock of threads to handle requests -- then my productivity really skyrocketed -- everything stays in memory and once in awhile I hit a button that commits all of the data structures to disk! One day the data got sooo big that it locked the whole site for a single second to do the commit, so I axed the problem by only committing the modified data every 30 seconds instead! The system's never more than 30 seconds out of date, and it's never in an inconsistent state! And there's no RDBMS to maintain, no SQL to care about, and it's lightning mother-fucking fast, you know, the way a 3GHz computer with 2 gigs of RAM is meant to be! ROFL!
Sure, management bitched that they couldn't peak at the site data through Microsoft Access, so 15 minutes later they had a spigot that jizzed CSV all over their faces. That shut them
up fast.
But seriously, please keep debating the merits of one SQL database versus the other SQL databases, it will have such an impact on your life.
My retort to a mailing list flamewar over the
enterprise
readiness of MySQL vs. Oracle vs. PostgreSQL vs. tin cans and string .
Once you start getting into "serious" work, or "enterprise"
level computing, which is all anyone argues about, every single
assumption gets tossed out the window.
Thought your OS was stable? Yeah, it's pretty stable, but
when it gets hit all day every day at 100%, it crashes for
some reason every few months. I'd love to say that Linux is the
exception here, but well, it isn't.
Maybe you bought the highest quality disks? And avoided the
"bad" vendor? Wrong! This year the bad vendor is the one
you bought plenty of! Looks like your recovery plan didn't
consider that 25% of your disks would fail in the first year!
Thought you had enough RAM? You don't! And you can't add more
because you're on a 32-bit platform. Sucker! Start migrating
to 64-bit and learn a whole new bunch of gotchas the hard way.
Hey! This RAID adapter has an awfully funny glitch! When you
pop a brand new disk in, if you reboot, it treats it as a whole
new array, and the funniest part is that it renumbers all of the
other arrays! Kernel panic: can't mount root device! What a
laugh! Good thing we have RAID here to give us added reliability!
Thought you'd never max out that fridge computer? Well, you
just did. It looks like your developers decide to get sloppy
when they think they have infinite capacity. A couple of weeks
of performance analysis and retuning the algorithms instead of
doing real new work!
Thought that replication setup would scale infinitely? Well,
infinitely actually means 10,000 queries/sec. Yup, that's the ceiling.
No choice now but to re-architect the whole system into a decentralized
dataset. Hey, since it's all so decentralized, lets just store CSV files!
Added bonus: management types love it!
Six months of re-engineering to decentralize the whole system,
and another six months to phase it in. And it sure will require
downtime!
For all of the talk of mission critical feature this and enterprise
functionality that, in the end, these "real work" loads are handled
by the resourcefulness of your people, because no platform is going
to even come close to solving all of your problems.
Package X vs. package Y does not make a difference in the big picture.
If only. MySQL, PostgreSQL, Oracle, BerkeleyDB, or peach fuzz? The answer
is obvious: pick the one your team is most capable and most comfortable
with. Got it? Great! You've just solved the easiest problem you're
ever going to have.
What, you don't think a 14 year old was right to be outraged that someone was adding yet another restriction to their lives? That was literally my first collision with any kind of politics, and it completely affected me. There were forces out there already working hard against my interests, instead of leaving me alone. Quite an effective impression to make on an otherwise apolitical youngster.
But you're right, I guess I was a stupid kid incapable of making any decisions for myself. I should've just accepted my innocence and youth and used my un-responsibility to go vandalize some property and knock up some girls with low self-esteem.
...was when I was 14 years old. He was taking some position against video games which lead the industry to shit its pants and make the ESRB. I wasn't old enough to vote at the time, but I was old enough to save up the dough to buy Mortal Kombat 2 for the SNES. But that was foiled when I was asked to show identification.
Eventually, I returned with my mother who approved the sale. All was well.
Except for Senator Lieberman.
In 2000, I was old enough to vote, and voted my revenge. It wasn't Bush vs. Gore to me. It was Lieberman and Tipper "Explicit Lyrics" Gore vs. anyone else.
I voted for Hillary Clinton as Senator, but if I have to decide who to vote for in 2008 the same way as I did in 2000, so be it. And I'm sure a lot of people who will be voting for the first time will see it that way, too.
When I sit down in front of a Linux server I can whip out super professional tools like strace and ltrace and lsof and dig through/proc, whip out gdb, etc. and follow everything step by step through the source code. Everything's usually simple enough that I can fit the entire system in my head and see where the broken piece is (except for PAM). The system invites me to do this. I can diagnose problems scientifically, professionally, and quickly.
When I sit in front of a Windows box, with some exceptions, all I can do is push the same set of buttons that the user has been pushing, and see if I can find a combination that works.
PCs force me to become an amateur. Reason: bad tools available.
Slashdot Mirror
Convenience and security are often at opposite ends of the computing spectrum.
The UNIX security model is not mass market friendly. The end user cares about security, but is not going to the work required to maintain it. The average joe can just barely be encouraged to lock his door at night, but only after decades of mass media has convinced him that if he doesn't, he'll allow a deranged lunatic to come into his home and rape and kill his wife and kids. Nothing that can happen on your computer compares to that, so people just won't be bothered to take the same precautions.
Need proof? Fine. Install Windows XP for someone; give them an administrative account and a restricted account. Tell them to only use the administrative account to install stuff, and use the restricted account for every day stuff. Come back in a few months. What do you find? They hit one "access denied" message on the restricted account and have used the administrative account ever since.
Forcing a security model that the customer clearly rejects does no one a service.
This is unfortunately a case where treating the endless symptoms of a disease, while far from optimal, is better than trying to cure the disease.
They care about applications. Killer applications.
Windows has >90% marketshare and it costs a non-trivial amount of money to develop an application that works across the board.
Adding MacOS and Linux versions is often difficult enough to justify.
Even when a number of competing hardware platforms implement the same API, the per-hardware platform bugs/test scenarios can be cost prohibitive. Java's a good example of this.
What application is killer enough that someone can stand to rewrite/QA to support .1% marketshare platforms?
Even if a free Linux version was dropped in my lap (which it won't) it's still going to be more expensive to support.
Windows has no technical advantage over alternative platforms; few would argue with that point. The windows monopoly comes from people not being comfortable administering and tinkering with their computers.
I'll agree with you, Windows offers no technical advantage over other platforms, but it does offer one really really compelling business advantage.
By simply targetting Windows, I can generate one binary that will run on approximately 500 million PCs, give or take a couple of million with weird configurations.
Target Linux? Well, sure, but the Linux version won't be anywhere near as user friendly as the Windows version since I'll have to put the burden on the user install the dependencies, patch their kernel (the big 10 distributions don't have out of the box support for what the app does), script it to start up the way they like it, etc. Would you trust a third party application to patch your kernel? I wouldn't. Will I ever recoup the porting costs? Probably not, which means I'm not making a good business decision by making the port, but doing it anyway since I want to use the app on my workstation at home.
Those things one does through the registry are things Microsoft does not want the average user to understand.
I just finished writing an application that saves state to the registry. It's actually really handy, since I don't need to manage config file loading, parsing, saving, etc. The fact that it's not stored in a file makes it less likely to be swept up into a different directory. It's always in the place that I expect it to be.
Can a regular joe user figure out what my application's keys in the registry do? Well, sort of. It's a pretty simple application. Do they gain anything by tweaking registry values? Not much, unless they want to see the post-install introduction screen again.
I'd say I understand it pretty well.
Additionally, I can take any running application and trace it and see what keys it is modifying, using free tools. The fact that every application does this the same way means it's pretty easy to isolate the library calls that update a configuration.
Did you know that Explorer saves the position of almost every single window in the registry, every time you move it? No doubt a sinister plot to remember where you like to keep your windows so that they can be in the same place, next time.
Bastards.
There's this architect I know of who maintains a 10 year old SGI workstation running some ancient CAD program. I asked him why he goes through the trouble, and he became LIVID. It'll be a cold day in HELL before he installs Windows and that Autocad garbage, apparantly.
Then there's also this interactive media artist I met once. He hand compiled his entire system from scratch, modified the video4linux driver to get better performance, and claims that he hasn't touched an Adobe product in years. I asked him why he wasn't using a Mac, and he went into some tirade about how some program was discontinued once and he couldn't find anything that would read his old saved work, and he swears he'll never put himself in that position again, open source all the way (I didn't ask for the details).
Both of these people appeared very serious about their craft.
So when someone tells me the eccentric outside of the box thinking individuals install MacOS X, I don't take it too seriously.
No one has a sense of humor?
No one can get subtlety?
Must every word we write be played back in monotone, completely out of context, character, to see if it can be used against us?
Anyone who understands hacker culture, or Bram's personality, would read that and laugh.
God, I probably have something just as ridiculous saved somewhere. In fact, I know I do, and I know someone else got ahold of it and spruced it up with Photoshop and made it look like a piece of communist propaganda that I'm just about ready to print out a million copies of. I come across it every so often and laugh, and anyone who knows me would laugh if they saw it.
And a court would say that clearly this is the mind of a terrorist.
Fuck.
Microsoft is successful because they understand how to open their platform.
Seriously.
Every other platform developer is too proprietary -- they're so afraid of leaving money on the table that they try to control EVERYTHING. They can't see the larger picture. They don't know how to monetize open platforms, they capture some modest marketshare and charge everyone to death to do anything on them, but ultimately they never become a standard, and the developer spends resources targetting yet another platform instead of making their products better.
Deep down, despite all of their shortcomings, Microsoft's a lot better than the alternatives. Read technical literate back when Microsoft was starting out -- there was a clear division between those resisting standardization on a single platform (for entirely valid reasons, I'll admit) and those who were frustrated by the high costs of targetting multiple moving (open?) platforms, or locking themselves into a platform that was way too proprietary to ever let anyone gamble on risky applications.
They defined the software product industry by striking the sweet spot, a balance between overly restrictive and unsupportedly open. Maybe it's because they have techies at the top of the company? Whatever. No one understands how to do this better than them. You can target Microsoft's platform and have a customer base of, what, 500-750 million PCs?
Monopolies aren't so great, but we could be stuck with far worse. Imagine someone took a cut of every e-commerce transaction?
I saw someone pick up a Tablet PC the other day. He played with it gleefully, and showed me the coolest app: Microsoft Journal.
Behold: it looks just like a marble notebook page that you can write on with a "pen" thanks to the touch screen! It allows you to save notes and even try to digitize the text! I remarked that it was quite indeed an impressive piece of technology for only $3000.
When am I going to get one? Probably not anytime soon, since I've discovered a much better solution for myself: a small attache journal and a fountain pen. Total cost: about $50 (I splurged on the pen).
Check it out, with the journal I can:
It's always on me, requires no external power source, and is quite pleasing to use. No spyware, viruses, hackers, hardware failures, or bugs to worry about.
Plus, since the pen is a indispensible part of the equation, you always carry it on you, and can sign your name with a flourish on a credit card receipt, or otherwise display the quality of your pedigree to impressionable onlookers.
I've also found that, at least when I speak now, I slightly better organize my thoughts before I open my mouth since it's a pain in the ass to re-word a paragraph that you've half-written. This journal dialogue transfers to my vocal dialogue, for some reason.
Oh, also, cursive handwriting is the shit. I forgot how much I loved it.
Yes, it's the ultimate personal assistant. Between you and me, I think these things are going to be BIG!
"Here we were, working 90 hours a week to build something to help Linux, working long hours, giving up a lot of money and time, and all the time we were doing that, at every step of the way, we had people abusing us, sending us nasty e-mail, telling us we're jerks. It's been constant. It gets tiresome. After five years of that you just say: It doesn't make business sense and it isn't that fun."
Translation:
"For years we enjoyed a lot of low-cost publicity for BitKeeper because the Linux kernel team was using it. We could go to IBM and HP and say `Look, these guys use it, and they're really discriminating programmers. So, that'll be $2000 per seat.' It costs us relatively little to donate a server to the kernel team, and the occasional nasty email is easy to delete. All in all, it was a successful campaign until TRIDGE CAME ALONG and almost made us obsolete -- at that point we'd be crazy to give him any chance of actually polishing it off, so we yanked the plug."
Depending on how you define innovation, one can argue that it's rare to find it anywhere. Everything is built on everything else. Is BitTorrent innovative? I'd say yes, it's the first independent app that offers a truly powerful P2P-accelerated download. Someone may say it is not so revolutionary, it is just an incremental step from concepts found in Kazaa, or eDonkey. And you know what? They're right!
Is BitKeeper innovative? Someone might swear by whatever cool feature it has, and proclaim that it's revolutionary (I've never used it), but I could just say that it builds on concepts in CVS/RCS/SCCS. Without those predecessors, BitKeeper might not even exist.
I digress, lets attack the real misconception: shrinkwrap software products are not universally viable business models! There's Microsoft, which is the industry big-player, Adobe, and then thousands of little boutique companies that make peanuts on selling boxed software and make their real profits through consulting and support.
The model is hardly viable for proprietary software companies, so it shouldn't suprise anyone if it doesn't make sense for open source software companies. Red Hat's product model is based on convenience, or as Bob Young would say, the "Ketchup Business Model". Any idiot can make ketchup at home -- it's really easy, easier than downloading and burning an ISO. But no one does. Red Hat isn't so lucky to be like Heinz in this regard, so they make up the bulk of their income through service and support. And so does every single other shrinkwrap software company not named Microsoft.
Go out and buy a journal. A really nice one costs $15. Buy yourself a fountain pen. A halfway decent one costs $25. Dust off those memories on cursive handwriting. The writing flows smoothly and I find it to be quite pleasant, relaxing even. Ink cartridges last about a week apiece.
I must take two or three pages of notes a day, that have nothing to do with work. Really useful for looking up what you were thinking about last month, errant phone numbers, confirmation codes, and the like.
Fits in a jacket pocket. Always on. No software to manage.
I feel vulnerable if I leave the house without it.
It's a time tested tactic. People who were considering building a system on top of Google Maps suddenly put those plans on hold if they know that Microsoft is going to wipe them out soon enough. :D
So was Google leading or following when they provided a map service with a few cool enhancements over the competition?
Is Microsoft leading or following when they provide a map service with a few cool enhancements over the competition?
Oh, I see. Because it's Google, they're INNOVATING, but because it's Microsoft, they're RIPOFF COPYCATTERS!
Cause you're just going to be rewriting it every month anyway.
I've noticed that a lot of people are absolutely terrified of writing code. They don't actually want to think about the problem, they just want to put a black box in its place and say mission accomplished.
It's a nice ideal, but I'm not sure the developer wouldn't have been better off solving the problem uniquely, exactly, and effectively without involving some other beast of a module that has its own release cycles, maintenance hassles, etc.
It may take me longer to solve the problem myself, but it feels like time better spent because I know what's there -- it's not a black box.
I can take ownership of the problem..... bingo! Just answered my own question. ;)
What it really comes down to is using your time effectively. If you and I were to write the same application, but I used a proper RDBMS and you used MySQL, my application would be done sooner because I wouldn't have to reinvent the wheel (or, your application would be less stable because you didn't bother with referential integrity checks, transactioning your data updates, etc). You app might be a bit faster (doubtful if we're talking about non-trivial datasets, and especially not if you've reimplemented many RDBMS features in your application layer), but it's highly unlikely the end user will notice that you're doing 5000 transactions per second and I'm only doing 4500.
Not installing and managing an RDBMS is my idea of using my time effectively.
Customer wants a small CRM? CGI under Apache that writes DBM files. Takes a day or three.
Customer wants a large CRM with 10,000 users? Build around an async I/O library that speaks web, maintains an in-core database, flushed to disk every 30 seconds. Takes a week or three.
Customer wants a really large CRM with 100,000 daily users? Build multiple sites in the previous fashion, one to handle accounts, one to handle contacts, one to handle queries, etc. Tie them all together using the magic of cookies and internal or external redirects. Takes a month or three.
No SQL? No RDBMS? Sounds like heaven to me.
You, my friend, have crappy hardware. Or choose your kernels very badly.
Blah blaah blaah.
Would I have offended less people if I said that our servers running OpenBSD were crashing?
How long was I supposed to make the original post? Let me go on...
In the ENTERPRISE, you get to learn when ${CURRENT_DISTRO_VENDOR} says STABLE, they really mean STABLE-ish.
In the ENTERPRISE, fucking idiots come out of the woodwork and cause some political battle over choosing ${CURRENT_DISTRO_VENDOR}, and that ${HIS_DISTRO_OF_CHOICE} servers have been running for years without any downtime doing the same exact job. Sort of!
In the ENTERPRISE, you end up compiling your own kernels from scratch because you learned that going with the distro's kernel should work in theory, but just doesn't.
In the ENTERPRISE, you get to learn that if you hit BIND and syslog with a thousand messages a second, they deadlock, and good luck managing to ssh in when syslog() calls blocks forever because ssh is trying to record your login!
In the ENTERPRISE, you learn that the syslog daemon you picked, of all fucking things to worry about, doesn't have a -n option to not resolve hostnames. So you use LD_PRELOAD to make gethostbyname() always fail.
In the ENTERPRISE, you get to learn that almost every shipping company will cause subtle but fatal damage to your servers if you ship them across the country!
In the ENTERPRISE, server's bitch is you!
If you have to write a sorting routine, bubble sort is the easiest to write, easiest to maintain, and will work fine on modest datasets.
Should these assumptions fail you, you can go back and write a better algorithm.
These lessons can carry over to data storage too.
Oh my, you're right! If that happens, it means I might have to... USE MY BRAIN AGAIN! *GASP!*
Why does anyone expect something that worked fine with 10 users to work fine with 100,000? It'd be swell if it did work out that way, but do you always write a quicksort if a bubble sort will do just fine, knowing you can switch to a quicksort later on if it becomes a problem?
"Enterprise" projects are so full of unique challenges, I don't see why something like the way you store your data makes a fuckola of a difference in the grand scheme of things. Everything gets rewritten a few times anyway. Even eBay.
It's much more harmful when you get into the mindset that application A automatically means data storage scheme Y, or specifically, Brand O's data storage scheme Y.
It's not that simple in the real world. Think about things like concurrency. If you have 200 clients running your app at the same time and 10 of them are trying to update a complex set of related tables, how do they avoid stepping on each others toes? Are you going to have your client apps communicate with each other and coordinate the inserts? Or just lock all the tables and kill performance? How do you ensure that everyone who is reading those tables sees a consistent data set?
Check this out: you _ s e r i a l i z e _ the requests!
Why, yes, serializing the requests when they each take 30 seconds would blow concurrency out of the water, but call me strange -- when I say to myself "hey, this thing takes 30 seconds to complete", I don't say "if my RDBMS had all of this insane complexity built into it, I wouldn't have to care that it takes 30 seconds!! DAMNIT!"
Everyone's so used to running things in their own virtual process with unlimited memory and unlimited wall clock time and having your write() block because of some guy's slowassed modem, that if you spend an hour thinking of a different world, everything that implies an RDBMS suddenly disappears.
Alas, the solution is a really hard pill to swallow: figuring problems out for yourself!
Far be it from me to fault anyone that needs an RDBMS to keep their paycheck coming.
That's all true, up until the point where your database craps out on you (you've seen it happen to Slashdot frequently, Penny Arcade, and pretty much every MySQL-backed website out there). Then it's painfully obvious to the user. However, this isn't about the end user. It's about the programmer, and using the proper tool for the job. There's absolutely no reason you should have to replicate the functions of an RDBMS in each and every one of your applications, and if you find yourself doing so it's pretty clear that you're using the wrong tool.
Blaaah blaah blah. It occurred to me recently that the only reason I even feel the need to use an RDBMS in a web application is because under Apache, it was easier to store data in an RDBMS than gain permission to store it in the filesystem in some concurrency safe way on a shared hosting environment.
That environment is so dangerous that I took these assumptions with me when I had 10 servers with root access. : /
The best change I ever made for my productivity and sanity was writing a tiny little module that reduces the entire RDBMS into a circa 1985-style dBase file. It's great! The entire module only has a handful of SQL calls, and the application is like SQL WHO?! The solution to my concurrency problems was to only allow one process to hit the db module at a time.
What? It won't scale, you say? I guess when the site starts having a hundred hits a second, I can worry about allowing some processes to have shared read only access and block writers until they have exclusive access. Who knows, maybe when there are a thousand hits a second, I'll have something like commit/rollback! Or I could just rewrite the fucker from scratch!
At least, I would have said that until I discovered that I didn't even need to use Apache, that I didn't need to keep around a flock of threads to handle requests -- then my productivity really skyrocketed -- everything stays in memory and once in awhile I hit a button that commits all of the data structures to disk! One day the data got sooo big that it locked the whole site for a single second to do the commit, so I axed the problem by only committing the modified data every 30 seconds instead! The system's never more than 30 seconds out of date, and it's never in an inconsistent state! And there's no RDBMS to maintain, no SQL to care about, and it's lightning mother-fucking fast, you know, the way a 3GHz computer with 2 gigs of RAM is meant to be! ROFL!
Sure, management bitched that they couldn't peak at the site data through Microsoft Access, so 15 minutes later they had a spigot that jizzed CSV all over their faces. That shut them up fast.
But seriously, please keep debating the merits of one SQL database versus the other SQL databases, it will have such an impact on your life.
Copied from my blog^Wweb based journal.
Once you start getting into "serious" work, or "enterprise" level computing, which is all anyone argues about, every single assumption gets tossed out the window.
Thought your OS was stable? Yeah, it's pretty stable, but when it gets hit all day every day at 100%, it crashes for some reason every few months. I'd love to say that Linux is the exception here, but well, it isn't.
Maybe you bought the highest quality disks? And avoided the "bad" vendor? Wrong! This year the bad vendor is the one you bought plenty of! Looks like your recovery plan didn't consider that 25% of your disks would fail in the first year!
Thought you had enough RAM? You don't! And you can't add more because you're on a 32-bit platform. Sucker! Start migrating to 64-bit and learn a whole new bunch of gotchas the hard way.
Hey! This RAID adapter has an awfully funny glitch! When you pop a brand new disk in, if you reboot, it treats it as a whole new array, and the funniest part is that it renumbers all of the other arrays! Kernel panic: can't mount root device! What a laugh! Good thing we have RAID here to give us added reliability!
Thought you'd never max out that fridge computer? Well, you just did. It looks like your developers decide to get sloppy when they think they have infinite capacity. A couple of weeks of performance analysis and retuning the algorithms instead of doing real new work!
Thought that replication setup would scale infinitely? Well, infinitely actually means 10,000 queries/sec. Yup, that's the ceiling. No choice now but to re-architect the whole system into a decentralized dataset. Hey, since it's all so decentralized, lets just store CSV files! Added bonus: management types love it!
Six months of re-engineering to decentralize the whole system, and another six months to phase it in. And it sure will require downtime!
For all of the talk of mission critical feature this and enterprise functionality that, in the end, these "real work" loads are handled by the resourcefulness of your people, because no platform is going to even come close to solving all of your problems.
Package X vs. package Y does not make a difference in the big picture. If only. MySQL, PostgreSQL, Oracle, BerkeleyDB, or peach fuzz? The answer is obvious: pick the one your team is most capable and most comfortable with. Got it? Great! You've just solved the easiest problem you're ever going to have.
Bah! Humbug.
What, you don't think a 14 year old was right to be outraged that someone was adding yet another restriction to their lives? That was literally my first collision with any kind of politics, and it completely affected me. There were forces out there already working hard against my interests, instead of leaving me alone. Quite an effective impression to make on an otherwise apolitical youngster.
But you're right, I guess I was a stupid kid incapable of making any decisions for myself. I should've just accepted my innocence and youth and used my un-responsibility to go vandalize some property and knock up some girls with low self-esteem.
Sigh, childhood lost, thanks to video games.
...was when I was 14 years old. He was taking some position against video games which lead the industry to shit its pants and make the ESRB. I wasn't old enough to vote at the time, but I was old enough to save up the dough to buy Mortal Kombat 2 for the SNES. But that was foiled when I was asked to show identification.
Eventually, I returned with my mother who approved the sale. All was well.
Except for Senator Lieberman.
In 2000, I was old enough to vote, and voted my revenge. It wasn't Bush vs. Gore to me. It was Lieberman and Tipper "Explicit Lyrics" Gore vs. anyone else.
I voted for Hillary Clinton as Senator, but if I have to decide who to vote for in 2008 the same way as I did in 2000, so be it. And I'm sure a lot of people who will be voting for the first time will see it that way, too.
When I sit down in front of a Linux server I can whip out super professional tools like strace and ltrace and lsof and dig through /proc, whip out gdb, etc. and follow everything step by step through the source code. Everything's usually simple enough that I can fit the entire system in my head and see where the broken piece is (except for PAM). The system invites me to do this. I can diagnose problems scientifically, professionally, and quickly.
When I sit in front of a Windows box, with some exceptions, all I can do is push the same set of buttons that the user has been pushing, and see if I can find a combination that works.
PCs force me to become an amateur. Reason: bad tools available.