I came to the conclusion yesterday that commercial software is for people who hate their job so much you have to pay them to do it. Free software is created by people who like to do it. If people in college decide they want to study what they like, and not what will make them employable, which community will see a more severe impact?
Belief in the The Free Market is the modern secular equivalent of belief in The True Cross, or of the Marxist's belief in Working Class Consciousness. This article is just another sermon.
I think could take these econopuritans with more equanimity if they had better hymns.
So, it's all free() and malloc(), but garbage collection lets you defer the decision to do free().
I once worked on a compiler which did garbage collection this way. When you compile a procedure, the signature is persistent, but the body isn't. That is to say, the data structure used to represent the signature will remain forever, but the data structures used to represent the bodies are not useful after you have finished compiling the body. So, you allocate space for the persistent stuff in permanent storage, and space for the body in some other, ephemeral blocks of storage. When you are finished with the procedure, you can just collect all the ephemeral stuff and throw it out, without thinking about it. This is much faster than using reference counts, since you don't pay a penalty for each pointer assignment. Since the language was Ada, what was persistent in one context was ephemeral in another. C or C++ wouldn't have the same problem. I thought it was a neat idea, and wish that it was *my* idea.
I understand that, but there are many more constraints in engineering projects than commercial constraints. The goal should be to create the most functional artifacts within the constraints imposed by resource and technology limitations. The model should be that if you do a good job, which includes understanding the end user requirements, and letting people know what you can do, you will be rewarded. Instead, the primary goal is to maximize profits. The model is do the least and charge the most. The problem is that its hard to decide what "least" and "most" are, so people take the optimistic route, underestimate the first, and overestimate the second. The result is that we get artifacts which do too little, and cost too much. I say artifacts rather than products, because this same phenomenon can describe things which are not necessarily sold.
Let me give you one simple example. One possible principle of engineering is "One cannot solve a problem without clearly stating it." This seems like it would be uncontroversial. However, I have worked for five companies since 1982, and in only one of them was I asked to write a design proposal for any project I have worked on. I write them as a matter of course, but even then, nobody knows what to do with them. Nobody reviews them, because there is no review process. As a result, my projects often don't do exactly what the managers wanted. Sometimes I provide a general, flexible solution where a quick hack was really wanted. Sometimes I do the opposite. Sometimes I don't understand the problem and solve some other problem. If I write a project proposal in which I state the problem and my proposed solution, and even then I frequently get it wrong, how can other people get it right when they don't write proposals, don't do designs, don't do anything like requirements analysis, and don't have any feedback to correct their initial misunderstandings.
The one company which had project proposals in their process only had a process because they wanted to say they were ISO9001 compliant, and they needed process steps for that. The project proposals were completely for show.
So, why is commerce fraud? Fraud is "A deception deliberately practiced in order to secure unfair or unlawful gain" according to my American Heritage Dictionary. Since no company says "We want your money, and we'll do anything short of working hard to get it", and their intent is to do anything short of working hard to get it, they are
Of the five companies I have worked for, only one has been in the PC business. Two were in the Electronics Testing business, one was in the compiler and language tools business, one was in the bank software business, and one is in the PC graphics business. The business model for the first four was not set by any vision of emulating Microsoft that I could tell. They were to my mind motivated entirely by unalloyed greed. I think you could even make a case for a model motivated by greed but tempered by wisdom, but that seems to be too much to ask.
This is a great article. I think I've decided, after something like 15 years of professional software engineering, that what RMS is reputed to believe is actually right. Commerce is fraud. I don't know if RMS believes it in fact, probably not, but it seems that when commerce drives engineering, quality is the victim. The value of Free Software is that people are creating things because they care about how well they work, and now how much lucre they can squeeze out of customers, so quality is the only criterion.
A year ago Sun increased their source code fee to 1.0 lots of dollars. People who paid this then are going to be very mad that they can get it for free now.
Slashdot Mirror
I came to the conclusion yesterday that commercial
software is for people who hate their job so
much you have to pay them to do it. Free
software is created by people who like to do
it. If people in college decide they want to
study what they like, and not what will make
them employable, which community will see a more
severe impact?
Belief in the The Free Market is the modern secular equivalent of belief in The True Cross, or of the Marxist's belief in Working Class Consciousness. This article is just another sermon.
I think could take these econopuritans with more equanimity if they had better hymns.
It's rude to point and laugh. I can't imagine any other reason to want to look at Micros~1 code.
You know, it doesn't evem pay to get out of bed on this day.
So, it's all free() and malloc(), but garbage collection lets you defer the decision to do free().
I once worked on a compiler which did garbage collection this way. When you compile a procedure, the signature is persistent, but the body isn't. That is to say, the data structure used to represent the signature will remain forever, but the data structures used to represent the bodies are not useful after you have finished compiling the body. So, you allocate space for the persistent stuff in permanent storage, and space for the body in some other, ephemeral blocks of storage. When you are finished with the procedure, you can just collect all the ephemeral stuff and throw it out, without thinking about it. This is much faster than using reference counts, since you don't pay a penalty for each pointer assignment. Since the language was Ada, what was persistent in one context was ephemeral in another. C or C++ wouldn't have the same problem. I thought it was a neat idea, and wish that it was *my* idea.
I understand that, but there are many more constraints in engineering projects than commercial constraints. The goal should be to create the most functional artifacts within the constraints imposed by resource and technology limitations. The model should be that if you do a good job, which includes understanding the end user requirements, and letting people know what you can do, you will be rewarded. Instead, the primary goal is to maximize profits. The model is do the least and charge the most. The problem is that its hard to decide what "least" and "most" are, so people take the optimistic route, underestimate the first, and overestimate the second. The result is that we get artifacts which do too little, and cost too much. I say artifacts rather than products, because this same phenomenon can describe things which are not necessarily sold.
Let me give you one simple example. One possible principle of engineering is "One cannot solve a problem without clearly stating it." This seems like it would be uncontroversial. However, I have worked for five companies since 1982, and in only one of them was I asked to write a design proposal for any project I have worked on. I write them as a matter of course, but even then, nobody knows what to do with them. Nobody reviews them, because there is no review process. As a result, my projects often don't do exactly what the managers wanted. Sometimes I provide a general, flexible solution where a quick hack was really wanted. Sometimes I do the opposite. Sometimes I don't understand the problem and solve some other problem. If I write a project proposal in which I state the problem and my proposed solution, and even then I frequently get it wrong, how can other people get it right when they don't write proposals, don't do designs, don't do anything like requirements analysis, and don't have any feedback to correct their initial misunderstandings.
The one company which had project proposals in their process only had a process because they wanted to say they were ISO9001 compliant, and they needed process steps for that. The project proposals were completely for show.
So, why is commerce fraud? Fraud is "A deception deliberately practiced in order to secure unfair or unlawful gain" according to my American Heritage Dictionary. Since no company says "We want your money, and we'll do anything short of working hard to get it", and their intent is to do anything short of working hard to get it, they are
Of the five companies I have worked for, only one has been in the PC business. Two were in the Electronics Testing business, one was in the compiler and language tools business, one was in the bank software business, and one is in the PC graphics business. The business model for the first four was not set by any vision of emulating Microsoft that I could tell. They were to my mind motivated entirely by unalloyed greed.
I think you could even make a case for a model motivated by greed but tempered by wisdom, but that seems to be too much to ask.
This is a great article. I think I've decided, after something like 15 years of professional software engineering, that what RMS is reputed to believe is actually right. Commerce is fraud. I don't know if RMS believes it in fact, probably not, but it seems that when commerce drives engineering, quality is the victim. The value of Free Software is that people are creating things because they care about how well they work, and now how much lucre they can squeeze out of customers, so quality is the only criterion.
A year ago Sun increased their source code fee to 1.0 lots of dollars. People who paid this then are going to be very mad that they can get it for free now.