As a "big system" developer I know most of the most interesting and hairy bugs are only found on live systems under live loads. In my own little test patch using every tool at my disposal I find many bugs. But often not the important ones. The ones that really bite the customer.
Thus the open source world can be way ahead of the proprietary guys if when a "live load" bug bites, the user can look at the source and poke in the memory with a debugger.
So the user doesn't have a clue. Well, not surprising is it? We hide everything. You will be shocked at how fast a smart but clueless user can pick up a clue or three if he has the source and a friendly debugger in front of him.
And yes he can download the source (and twenty libraries, recompile them (after working out how to switch debugging on), but he won't. No time, insufficient clue. The debug symbols and sources must always be present. Disk drives are way big enough now.
What I propose is more than just a powerful way of finding and fixing bugs. Its a way of generating clue.
Yes, we have moved far from the Open Source roots.
If a package crashes or does the wrong thing? What do we do now?
Like a Windows Luser we shrug and go off and do something else.
In the bad old download, compile and install your own s/ware days you had the debug symbols and the source, if it crashed, you fixed it and sent the patch back to the developer.
Am I saying we should go back to the pre-dpkg pre rpm days?
No, but given the size of the disks we have the distros should include the source and the debug symbols and if a package crashes it should by default load up the debugger and point you at the problem.
I willing to bet hard cash that the pace of open source development would pick up by an order of magnitude if the distros did that.
Why?
The million monkeys effect means the software gets hammered in ways never imagined by the developer. Thus the user is by far in the best position to trigger and find and often get good clues as to where the bugs are.
Having the source around and a debugger is an introductions to programming for every user. Instead of millions of luser's, we have create millions of guru's in training.
"Described by some as Africa's first World War, the conflict in the DRC (formerly known as Zaire) has involved seven nations. There have been a number of complex reasons, including conflicts over basic resources such as water, access and control over rich minerals and other resources as well as various political agendas. This has been fueled and supported by various national and international corporations and other regimes which have an interest in the outcome of the conflict. There have been
2.5 million deaths since the outbreak of the fighting in August 1998"
Of course you can ask which country has dropped the most real go-bang variety of bombs on the most other countries since WWII. Don't trust me, ask Google.
You can also ask which country has the greatest capacity for cyberwarfare and history of attacking other states with it. Ask Google about "crypto ag" and the NSA. "Echelon" is another nifty search term.
The problem is people are looking at Lines Of Code, not business priorities.
Lines Of Code are liabilities, not assets.
You can sell the right functionality, easily maintained, on time within a tight schedule, pixel-pushed and micromanaged to fit the needs and dark cravings of the PHB's. But LOC's themselves don't sell.
These schemes are addressing the wrong end. A good Open Source company could go forth and get business and sub-contract to programmers, but what will _never_ work is sitting there saying, we have programmers, give us your business.
Gold Owners _never_ buy LOC's. They don't understand code, don't trust 'em, couldn't used 'em if they tried. They buy warm bods to do stuff.
A business that went forth an presented it self to its clients as a run'o'the'mill software shop, with cheapish rates, but was powered by SourceX style flocks of part timers, that may work.
But you would have problems with delivery dates and schedules, since PHB's get quite antsy about that...
According to former and current DEA, military, and State Department officials, the cartel had assembled a database that contained both the office and residential telephone numbers of U.S. diplomats and agents based in Colombia, along with the entire call log for the phone company in Cali, which was leaked by employees of the utility. The mainframe was loaded with custom-written data-mining software. It cross-referenced the Cali phone exchange's traffic with the phone numbers of American personnel and Colombian intelligence and law enforcement officials. The computer was essentially conducting a perpetual internal mole-hunt of the cartel's organizational chart. "They could correlate phone numbers, personalities, locations -- any way you want to cut it," says the former director of a law enforcement agency. "Santacruz could see if any of his lieutenants were spilling the beans."
They were. A top Colombian narcotics security adviser says the system fingered at least a dozen informants -- and that they were swiftly assassinated by the cartel. A high-level DEA official would go only this far: "It is very reasonable to assume that people were killed as a result of this capability. Potential sources of information were compromised by the system."
Sounds like in the war between the cartel's and the gov.us, the cartel's have won. It seem's like they ownz you US folk already...
"Modern C++ Design - Generic Programming and Design Patterns Applied"
by Andrei Alexandrescu
The ACLU voted this book best C++ book of 2001.
Michael Feathers of ObjectMentor described this book as "Porno for
Programmers"
With accolades like that, it is obviously an, umm, ah, unusual, book.
Let me start with a "Once upon a time story...."
Once upon a time, I wrote a C++ library that should have become the
Standard Template Library. But Stepanov didn't play fair. Whilst I, in
my third worldish corner, dreamt up arcane workarounds for the
deficiencies of the C++ compilers of the day, Stepanov walked next
door to Bjarne Soustroup's office and told him to do it right.
I remember the frustration, I needed template template parameters. I
needed traits, above all I needed ways of manipulating types and
getting information about types. I did some amazingly convoluted and
tricksy things to make it work. My library did work, but bygorrah was it
arcane.
I gave up on C++ as an "almost" language that didn't quite make the
grade.
After Stepanov's bullying, C++ was better but it still lacked
things. RTTI is deficient. Types are not first-class objects.
I had given up hope of it ever becoming my dream language.
Now there is a new player on the C++ scene. Andrei Alexandrescu.
He is starting from the base that Stepanov, Moo and Soustroup
created. The nifty template template stuff is now in every
compiler. We all have partial specialization. We have traits.
Now he tackles the rest of the deficiencies with a bullheaded
aggressiveness that is amazing. He does horrible things. Truly evil
things. Things that make me blush to read. He then sweeps all the
nasty stuff under a nifty carpet called the "Loki Library".
Its neat and usable. All the deficiencies of C++ are gone.
The grand promise of very efficient reusable generic code lies open
before us with all the obstacles blasted away.
Java will never compete with C++ for efficiency and
flexibility. Alexandrescu has created a new future.
Now we know this book is important, so what is in it....
Policy Based Class design.
Every design has trade-offs. What works in one situation is slow in
another. What is safe here is a disaster there. Some situations allow
some really good optimizations, but other places not.
We want reusable components, but if we make it generic, we end up with
a "Lowest Common Denominator" kitchen sink mess.
Policy Based class design is the answer.
Read the chapter on Smart pointers. It is the best example of generic,
flexible, efficient, reusuable design I have ever seen.
Part I Techniques and Typelists.
This is the "porn" in the "porno for programmers". What he does here
is pure horror. This the Steven King of the software development
literature. This is why people who have looked deep into the heart of
C++ templates shudder.
But don't worry. He neatly prepackages it all and serves it up in a
nice tidy API. Enjoy.
Chapter 4. Small Object Allocation.
Don't discount this one. STL
actually quietly does a lot of this. It is amazingly effective.
Part II Components.
All the high-tech of Part one applied to roll out
generic, efficient and flexible implementations of 7 standard
patterns.
I always wanted a book entitled -
"Identification of Common garden Birds and small Mammals - A field guide for Cat lovers."
I'm sure with a digital camera and a good ratter, you could go quite far....
Someone could create a web site where you can submit photos of your own moggies trophies, and assist other proud owners in their identification and interpretation of entrails.
I remember walking, zombie-like at 2am to the bathroom and been struck by the question. "Why is there a large Rat, buried up to the neck in the concrete floor?"
Then my cat came up proudly going WowWowWowrrrr!
Closer inspection reveal that the rat wasn't buried, it was just that the rest of it was missing, presumably regurgitated under my bed.
Believe it or not, your processor isn't at 100% utilization as you type messages to Slashdot.
Ah, but my 'net connection is....(and that has nothing to do with my typing speed....) I trawl the 'net for info.
I think the authors mistake is calling it an OS. reading the article closer it isn't an OS. It is more a load balancing general purpose RPC stub with several huge problems....
Security. The ISOS allows any program to be run remotely on the clients PC. I don't think I need enumerate the long list of vulnerabilities associated with that.
It is only useful when the inputs and outputs are very very small compared to the compute time. Very few commercial apps satisfy that.
Why is this separation necessary?
You haven't been watching all this fuss about "inappropriate use" have you? Have you forgotten Borland's shenanigans where the bosses raided the employees email? No thanks, both the bosses and I want seperation of work and private.
Your computer is simply a resource that can be used to accomplish something.
Your head is simply a resource that can accomplish something. Can I borrow it for awhile...? Its obviously not at 100% utilization:-)
Then disconnect your machine from the Net, and you will be happy. However don't presume to speak for the vast majority of computer users who seem extremely happy to be part of a large, distributed network of machines and systems.
And are extremely unhappy when some sod from the 'net "ownz yoo". Doesn't that tell you something?
As I said, there is data that is mine to control and data since it is mine to do as I please, I wish to share.
I didn't say we want standalone net disconnected machines, we want machines that are masters operating amongst peers, not clients dependent on servers.
We don't want "The Network Is The Computer". Remember mainframes? Remember how we joyfully fled from them?
What we want is to really own our computer power.
We want a very clear sense of "This is my computer" and "This is my data". I can do what I like with it.
Think folks, what is all the fuss about security and file sharing? Ownership. This is my data to own (keep private) and my data to share (if I choose).
Complexity and installation difficulties steal our sense of ownership. When the computer is a burden, we don't want to own it. Complexity robs
us of choice.
The correct fix is not an ISOS, or retreat to mainframe days. The correct fix is to simplify and make things easy.
I don't want my work computer to be my home computer. My employer and I definitely want a strong sense of separation on that front thank you.
Forget these silly pipe dreams, and concentrate on easing the pains of ownership so that we have strength to share.
All this is a silly confusion over....
What I do and can do... (I want unlimited freedom and choice)
What intellectual product I create... (It costs nothing to make another copy, so why limit distribution?)
What hard product I create... (It costs much effort to make a copy, and requires hard inputs.)
What I own... (What I control)
What is private... (Thoughts and activities that concern me only thank you)
Remove the confusion between the above items and the desire for silly things like "The Network Is The Computer", DMCA etc goes away.
Slashdot Mirror
Thus the open source world can be way ahead of the proprietary guys if when a "live load" bug bites, the user can look at the source and poke in the memory with a debugger.
So the user doesn't have a clue. Well, not surprising is it? We hide everything. You will be shocked at how fast a smart but clueless user can pick up a clue or three if he has the source and a friendly debugger in front of him.
And yes he can download the source (and twenty libraries, recompile them (after working out how to switch debugging on), but he won't. No time, insufficient clue. The debug symbols and sources must always be present. Disk drives are way big enough now.
What I propose is more than just a powerful way of finding and fixing bugs. Its a way of generating clue.
If a package crashes or does the wrong thing? What do we do now?
Like a Windows Luser we shrug and go off and do something else.
In the bad old download, compile and install your own s/ware days you had the debug symbols and the source, if it crashed, you fixed it and sent the patch back to the developer.
Am I saying we should go back to the pre-dpkg pre rpm days?
No, but given the size of the disks we have the distros should include the source and the debug symbols and if a package crashes it should by default load up the debugger and point you at the problem.
I willing to bet hard cash that the pace of open source development would pick up by an order of magnitude if the distros did that.
Why?
I take it, like a Bone world rat creature, he is going to outwit us to death.
NEVER.
Damn yankee scammers. ;-))
While you there have a good browse around. Lots of good stuff for the "Midnight Engineer" as he calls us.
"AARGH!! There is a Quake coming in the next three weeks"
Bugger.
Now what do I do?
Hmm. Hoo, Hum. Dunno. May as well go fishing, was going to do that anyway...
You can also ask which country has the greatest capacity for cyberwarfare and history of attacking other states with it. Ask Google about "crypto ag" and the NSA. "Echelon" is another nifty search term.
Given how long OOP's and Wireless has been around, I dare say a serialized Object can travel quite far at the speed of light....
I was just thinking what would happen if I bought a set of those and attached them firmly to a cow-orkers Office Chair.
My only problem is to decide whether to set them all pointing in the same direction for maximal linear thrust, or like a catherine wheel?
Any suggestions?
Lines Of Code are liabilities, not assets.
You can sell the right functionality, easily maintained, on time within a tight schedule, pixel-pushed and micromanaged to fit the needs and dark cravings of the PHB's. But LOC's themselves don't sell.
These schemes are addressing the wrong end. A good Open Source company could go forth and get business and sub-contract to programmers, but what will _never_ work is sitting there saying, we have programmers, give us your business.
Gold Owners _never_ buy LOC's. They don't understand code, don't trust 'em, couldn't used 'em if they tried. They buy warm bods to do stuff.
A business that went forth an presented it self to its clients as a run'o'the'mill software shop, with cheapish rates, but was powered by SourceX style flocks of part timers, that may work.
But you would have problems with delivery dates and schedules, since PHB's get quite antsy about that...
Sounds like in the war between the cartel's and the gov.us, the cartel's have won. It seem's like they ownz you US folk already...
Why?
They have a driver for a 1100 british pound DAC, so how many guys are going to lay out that much cash for a radio?
And to listen / send what?
The ACLU voted this book best C++ book of 2001.
Michael Feathers of ObjectMentor described this book as "Porno for Programmers"
With accolades like that, it is obviously an, umm, ah, unusual, book.
Let me start with a "Once upon a time story...."
Once upon a time, I wrote a C++ library that should have become the Standard Template Library. But Stepanov didn't play fair. Whilst I, in my third worldish corner, dreamt up arcane workarounds for the deficiencies of the C++ compilers of the day, Stepanov walked next door to Bjarne Soustroup's office and told him to do it right.
I remember the frustration, I needed template template parameters. I needed traits, above all I needed ways of manipulating types and getting information about types. I did some amazingly convoluted and tricksy things to make it work. My library did work, but bygorrah was it arcane.
I gave up on C++ as an "almost" language that didn't quite make the grade.
After Stepanov's bullying, C++ was better but it still lacked things. RTTI is deficient. Types are not first-class objects.
I had given up hope of it ever becoming my dream language.
Now there is a new player on the C++ scene. Andrei Alexandrescu.
He is starting from the base that Stepanov, Moo and Soustroup created. The nifty template template stuff is now in every compiler. We all have partial specialization. We have traits.
Now he tackles the rest of the deficiencies with a bullheaded aggressiveness that is amazing. He does horrible things. Truly evil things. Things that make me blush to read. He then sweeps all the nasty stuff under a nifty carpet called the "Loki Library".
Its neat and usable. All the deficiencies of C++ are gone.
The grand promise of very efficient reusable generic code lies open before us with all the obstacles blasted away.
Java will never compete with C++ for efficiency and flexibility. Alexandrescu has created a new future.
Now we know this book is important, so what is in it....
Policy Based Class design.
Every design has trade-offs. What works in one situation is slow in another. What is safe here is a disaster there. Some situations allow some really good optimizations, but other places not.
We want reusable components, but if we make it generic, we end up with a "Lowest Common Denominator" kitchen sink mess.
Policy Based class design is the answer.
Read the chapter on Smart pointers. It is the best example of generic, flexible, efficient, reusuable design I have ever seen.
Part I Techniques and Typelists.
This is the "porn" in the "porno for programmers". What he does here is pure horror. This the Steven King of the software development literature. This is why people who have looked deep into the heart of C++ templates shudder.
But don't worry. He neatly prepackages it all and serves it up in a nice tidy API. Enjoy.
Chapter 4. Small Object Allocation.
Don't discount this one. STL actually quietly does a lot of this. It is amazingly effective.
Part II Components.
All the high-tech of Part one applied to roll out generic, efficient and flexible implementations of 7 standard patterns.
Say, all this super-quiet case tech could be used by dark-siders to cut down on the noisy breathing problem...
I'm sure with a digital camera and a good ratter, you could go quite far....
Someone could create a web site where you can submit photos of your own moggies trophies, and assist other proud owners in their identification and interpretation of entrails.
I remember walking, zombie-like at 2am to the bathroom and been struck by the question. "Why is there a large Rat, buried up to the neck in the concrete floor?"
Then my cat came up proudly going WowWowWowrrrr!
Closer inspection reveal that the rat wasn't buried, it was just that the rest of it was missing, presumably regurgitated under my bed.
Robert Jordan Wheel of Time series comes painfully to mind....
Except code is much worse. I will probably only read the Wheel of Time once, at worst twice. I reread the code I'm maintaining many times.
I know I'm on the right track when by deleting code I'm adding functionality.
It is my favourite thing to hammer the business folk with, LOC's are liabilities, not assets. They freak out totally when it finally penetrates...
Sort of like watching a caterpillar go into a "which foot first" fit.
It selects for "blood pumping action" by sending adrenaline junkies to dance with publicity junkies.
Both sides of the equation firmly believe in the maxim "_Anything_ for a good story."
Give this moron some oxy.
Ah, but my 'net connection is....(and that has nothing to do with my typing speed....) I trawl the 'net for info.
I think the authors mistake is calling it an OS. reading the article closer it isn't an OS. It is more a load balancing general purpose RPC stub with several huge problems....
- Security. The ISOS allows any program to be run remotely on the clients PC. I don't think I need enumerate the long list of vulnerabilities associated with that.
- It is only useful when the inputs and outputs are very very small compared to the compute time. Very few commercial apps satisfy that.
Why is this separation necessary?You haven't been watching all this fuss about "inappropriate use" have you? Have you forgotten Borland's shenanigans where the bosses raided the employees email? No thanks, both the bosses and I want seperation of work and private.
Your computer is simply a resource that can be used to accomplish something.
Your head is simply a resource that can accomplish something. Can I borrow it for awhile...? Its obviously not at 100% utilization :-)
As I said, there is data that is mine to control and data since it is mine to do as I please, I wish to share.
I didn't say we want standalone net disconnected machines, we want machines that are masters operating amongst peers, not clients dependent on servers.
We don't want "The Network Is The Computer". Remember mainframes? Remember how we joyfully fled from them?
What we want is to really own our computer power.
We want a very clear sense of "This is my computer" and "This is my data". I can do what I like with it.
Think folks, what is all the fuss about security and file sharing? Ownership. This is my data to own (keep private) and my data to share (if I choose).
Complexity and installation difficulties steal our sense of ownership. When the computer is a burden, we don't want to own it. Complexity robs us of choice.
The correct fix is not an ISOS, or retreat to mainframe days. The correct fix is to simplify and make things easy.
I don't want my work computer to be my home computer. My employer and I definitely want a strong sense of separation on that front thank you.
Forget these silly pipe dreams, and concentrate on easing the pains of ownership so that we have strength to share.
All this is a silly confusion over....
Remove the confusion between the above items and the desire for silly things like "The Network Is The Computer", DMCA etc goes away.
I used to like Borland.