Wrong, wrong, wrong. We're not talking about bearing children, or anything else physical. We're talking about aptitudes, which are determined more by nurture than nature.
When I first went away to university, hundreds of miles from my hometown, I heard a new thing. I heard that men were better at math than women. I looked around and sure enough there were more male math majors than female. And the men seemed to be getting higher grades in first year calculus. This was bizarre, because I had been told the opposite, and seen the opposite, since birth!
You see, mathematics aptitude is a product of upbringing. Where I grew up it was taken for granted that females were better at math than males. All my math teachers said so! And for me and my classmates, it was true. The girls consistantly got good grades in math. Only boys flunked algebra. All the adult bookkeepers and accountants were women. Looking back on my childhood, it's almost like my hometown was a cultural backwater who hadn't gotten the memo that said boys are better at math.
I see that you have two testicles, therefor you must stand over here with this group. And you, with the two ovaries, but stand over there with that group. I will now assign you a desktop based on the group your are standing with. This should be an easy task for me, since all members of a group have identical aptitudes.
There are statistically significant differences between men and women in terms of behaviors, aptitudes and motivations. But they aren't necessarily genetic. They arise because they have been shoved into two separate groups since birth, with the assumption that all members of a group are similar.
Boys are told to play competitive games like soccer and basebasll and girls are told to play cooperative games like house and tea parties. Boys who attend tea parties are sissies and girls who play baseball are tomboys. Boys are told to be engineers and girls to be secretaries. Boys are told to be doctors and girls to be nurses. I've seen this even in the most "progressive" of households. It's a hard cycle to break, because girls want to emulate their mothers and boys their fathers.
Why aren't their more women programmers? Because they are all told as girls that computers are difficult and to try something else. Boys are told that computers are difficult so they need to keep at it until they get it right.
I am not a feminist by denying a genetic basis for gender differences. Not by any means! But it doesn't take a rocket scientist (typically male) to realize that environment plays a huge role in child development.
Unless I copy GPLd software and create a proprietary modification of it. Then ten thousand Slashsdotters will howl that I have stolen the code.
If you don't have a stake in the copyright being infringed then it's merely free speech. If it's a copyright you have a personal stake in then it's a violent mugging.
That doesn't work here, because we don't have a parliamentary system like most European nations. In your country, if a third party gets five percent of the vote, they get five percent of the seats in the legislature. Here an individual candidate must personally gain a plurality of the vote in his or her district order to win.
I am not arguing against voting for third party candidates, however. Please do if you agree with their platforms and policies. It won't get any third party candidate elected, but it is a step towards making today's minor parties into tomorrows major parties.
The purpose of the executive branch of US government is to enforce the law, and not to judge its constitutionality. If you're going to blame someone, blame congress for passing this law to begin with. This law will eventually be visited by the Supreme Court, at which time it will decide its constitutionality.
The real problem here is that it is far too easy to enact laws and far too hard to repeal or overturn them.
I'm forced to change my NT account password every two months as well. I keep the same password and append a digit. I write the digit on a corner of my monitor so I can remember.
The really sad thing is that I only use NT once every two months, so it seems that I am changing my password everytime I use it. Sigh.
I always go into their accounts and mess stuff up. Like adding random slides with their animated password to the powerpoint presentation they're going to give in half an hour.
To me, almost every non-trivial structure is appropriate for a database.
You know all those posts that say OOP programmers go overboard and use OO for everything, and how they're so stupid for thinking everything can be an object in an inheritence tree? Welcome to the looking glass.
Every programming paradigm can be abused, and the database paradigm is no exception. For many problems, it is the appropriate solution. But for many others it is gross overkill. Our company builds an embedded UNIX system. One database programmer implemented "uname" with MySQL. Huh?
Databases work well for data that is numerous and uniform. It does not work well for data that is few and varied. Customer accounts? A database of course! A web browser? Not really, because an XML DOM tree is much simpler to model HTML with.
It can be a lot of work to overhaul or remove your state management when you need to move state into a database to be shared by other apps and languages.
If your data is inappropriate for a database, then it doesn't matter if it's in a C struct or a C++ class, they'll both be a bitch to convert. And if it is appropriate for a database, it's not going to be harder to port the OOP C++ version over the Structured C version.
No no no! That would be too much like X11R6! Remember, this is Slashdot, where the remote capabilities of X are perceived as a hinderance to 3D game performance under Linux.
If you check, you'll find that Qt based Kylix (as well as other commercial Qt based libs and tools) doesn't require developers to pay Trolltech. Ever wonder why? Because Borland, which is significantly smaller than Sun, negotiated a simple deal with Trolltech. I don't know what the details are, but I suspect it's a small percentage of each Kylix purchase. Sun could easily do the same thing with the same triple licensing that Trolltech uses, so that non-commercial development on Solaris would be free.
starts becoming expensive to distribute as a corporate entity
Huh? It's free to distribute! Even the proprietary version!
But it's not free to develop for. You need a per-developer license to create non-free software with Qt. But once you've done that, you can distribute the finished product any way you want, including in ways in which you don't have to pay any with no royalties for anyone. You can even distribute the Qt runtimes royalty free!
No, it's not free-beer, not even for billion dollar companies like Sun and HP that might have twenty or so people actually developing with Qt. But you don't have to pay to distribute your Qt software.
HP wants to write commercial proprietary applications for GNOME. They cannot do that when the development has not stabilized. HP does not want to develop for and support a moving target, and their customers won't want to install a patch every week just because someone at GNOME changed the API. Geez, even Windows managed to keep a stable API through three different desktops, nine major release versions, and one complete decade!
Well, since you're a GNOME user, I can understand why you don't understand the term, since it's so rare to see it. [ducks]
"Stable", among other things, means that the development APIs are not changing. It does NOT mean that development has stopped, only that they have finalized the interfaces, allowing other people to develop for it.
A stabilized GNOME 2 means that you don't have to rewrite your application next week when things change. Ideally, you shouldn't have to rewrite it when GNOME 3 comes out either. Consider the great unwashed evil that is KDE: the API is stable. It doesn't matter if you love or hate KDE, if you look at the project with an honest perspective, you have to agree that they have a relatively stable API. They may add new interfaces, but they keep their old ones as stable as possible. I ported several KDE 2 applications to KDE 3 for the FreeBSD ports collection. Average porting time was half an hour, including compilation and testing. And this was between MAJOR release versions!
An unstable API is a public announcement that the developers do not feel that the project is ready for public use, regardless of other statements to the contrary. GNOME is not alone in this regard, but that doesn't make the practice right.
A mere five years ago a "multigigabyte" drive would have been immense. So you'll have to excuse these guys for not being 15 year old gaming geeks who need to upgrade their hardware every few days.
I would even argue that you don't really need inheritance.
Consider a simple GUI. In this GUI you have a button, a checkbox, a label and an edit box. All of these things are widgets. Does it not make sense to use a base widget class instead of rewriting all the common related functionality they all have? Now add a radio button. It's related to a checkbox. Perhaps it also makes sense to have the checkbox and radio button share a common ancestor as well.
I can hear people say "that's what functions are for!". True, that is what they are for. But if a function is only useful for a widget related structure, doesn't it make sense to encapsulate that function with that structure?
No, you don't need inheritance. But sometimes it's damned useful.
I've seen some very complex software written in functional languages that was very easy to follow even though the had no real OO concepts
So have I. But no one here is arguing that OOP is the only appropriate paradigm for all problem domains.
In the past I have made some comments on what I think plain old C could become if it incorporated some modulization features.
Be careful with this. If you don't watch yourself it could end up becoming Objective C:-)
Re:OOP is frequently the wrong answer
on
The Post-OOP Paradigm
·
· Score: 2, Interesting
Don't lay this sin solely at OOP's door. We learned it from Structured Programmers, and are smugly pleased that Generic Programmers are knee deep in this iniquity as well.
Any formalized process can be taken to an extreme. And you yourself are arguing for a formalized process as well: "And never write OO when procedural programming can do the same trick in less space." Sometimes it might actually make sense to write it in OO, because other things may be more important than less space! Programming isn't a field of morality where you must choose a particular code of ethics to abide by no matter what.
My point about 'power users' installing software is something that can be set and controlled so that the software can also be available to ALL users on the machine. Not just in the home directory for the user that installed it. I have seen too many senerios where there have been 10 different Mozilla installations for 10 different users. This is just ridiculous.
And I've seen the case where one user wanted to update StarOffice and ruined the install for everyone because they didn't do it right. Whether it's NT or UNIX, one person needs to be in charge of the system. It may be inconvenient, but as the adage says, convenience is the inverse of security.
Keeping UNIX variants on par with what NT was doing 10 years ago is still a process of catch up, and yet the UNIX 'ideology' is far older. And its age of older 'ideologies' is not a plus in the technology world, instead they have to be molded and kludged to make them mimic what is natural in newer OS architectures like NT.
Were you able to see past the single word "duh" and understand the point I was trying to make? UNIX is not obsolete, it's just different paradigm. You don't throw away a technology just because it's old, not even software technology.
The UNIX philosophy that there shall be one and only one ultimate administration account is no less valid than NT's philosophy that there can be multiple ultimate administration accounts. Both can be used to implement a practical easy-to-use system for home desktop systems. XP and OSX are prime examples.
The most important part of token/object security model is its practical effect. You can achieve the same practical effect with UNIX using owners and groups. But to make it easy, it needs to be done by the system integrator (the distro) and not by the system administrator.
This story is about Lycoris. There is no reason an "entry-level" UNIX of any sort cannot implement a model similar in effect to NT's if they wanted to. The problem with many newbie Linux distros is that they don't want to remove themselves too far from the traditional server model of Linux.
Slashdot Mirror
Wrong, wrong, wrong. We're not talking about bearing children, or anything else physical. We're talking about aptitudes, which are determined more by nurture than nature.
When I first went away to university, hundreds of miles from my hometown, I heard a new thing. I heard that men were better at math than women. I looked around and sure enough there were more male math majors than female. And the men seemed to be getting higher grades in first year calculus. This was bizarre, because I had been told the opposite, and seen the opposite, since birth!
You see, mathematics aptitude is a product of upbringing. Where I grew up it was taken for granted that females were better at math than males. All my math teachers said so! And for me and my classmates, it was true. The girls consistantly got good grades in math. Only boys flunked algebra. All the adult bookkeepers and accountants were women. Looking back on my childhood, it's almost like my hometown was a cultural backwater who hadn't gotten the memo that said boys are better at math.
Precisely!
I see that you have two testicles, therefor you must stand over here with this group. And you, with the two ovaries, but stand over there with that group. I will now assign you a desktop based on the group your are standing with. This should be an easy task for me, since all members of a group have identical aptitudes.
There are statistically significant differences between men and women in terms of behaviors, aptitudes and motivations. But they aren't necessarily genetic. They arise because they have been shoved into two separate groups since birth, with the assumption that all members of a group are similar.
Boys are told to play competitive games like soccer and basebasll and girls are told to play cooperative games like house and tea parties. Boys who attend tea parties are sissies and girls who play baseball are tomboys. Boys are told to be engineers and girls to be secretaries. Boys are told to be doctors and girls to be nurses. I've seen this even in the most "progressive" of households. It's a hard cycle to break, because girls want to emulate their mothers and boys their fathers.
Why aren't their more women programmers? Because they are all told as girls that computers are difficult and to try something else. Boys are told that computers are difficult so they need to keep at it until they get it right.
I am not a feminist by denying a genetic basis for gender differences. Not by any means! But it doesn't take a rocket scientist (typically male) to realize that environment plays a huge role in child development.
Unless I copy GPLd software and create a proprietary modification of it. Then ten thousand Slashsdotters will howl that I have stolen the code.
If you don't have a stake in the copyright being infringed then it's merely free speech. If it's a copyright you have a personal stake in then it's a violent mugging.
That doesn't work here, because we don't have a parliamentary system like most European nations. In your country, if a third party gets five percent of the vote, they get five percent of the seats in the legislature. Here an individual candidate must personally gain a plurality of the vote in his or her district order to win.
I am not arguing against voting for third party candidates, however. Please do if you agree with their platforms and policies. It won't get any third party candidate elected, but it is a step towards making today's minor parties into tomorrows major parties.
The purpose of the executive branch of US government is to enforce the law, and not to judge its constitutionality. If you're going to blame someone, blame congress for passing this law to begin with. This law will eventually be visited by the Supreme Court, at which time it will decide its constitutionality.
The real problem here is that it is far too easy to enact laws and far too hard to repeal or overturn them.
"One database programmer implemented "uname" with MySQL. Huh?"
It worked for him/her (I presume, not having seen it).
It's a single record of three static unchanging fields! This goes beyond overkill and into the realm of "shock and awe".
I'm forced to change my NT account password every two months as well. I keep the same password and append a digit. I write the digit on a corner of my monitor so I can remember.
The really sad thing is that I only use NT once every two months, so it seems that I am changing my password everytime I use it. Sigh.
I always go into their accounts and mess stuff up. Like adding random slides with their animated password to the powerpoint presentation they're going to give in half an hour.
To me, almost every non-trivial structure is appropriate for a database.
You know all those posts that say OOP programmers go overboard and use OO for everything, and how they're so stupid for thinking everything can be an object in an inheritence tree? Welcome to the looking glass.
Every programming paradigm can be abused, and the database paradigm is no exception. For many problems, it is the appropriate solution. But for many others it is gross overkill. Our company builds an embedded UNIX system. One database programmer implemented "uname" with MySQL. Huh?
Databases work well for data that is numerous and uniform. It does not work well for data that is few and varied. Customer accounts? A database of course! A web browser? Not really, because an XML DOM tree is much simpler to model HTML with.
You basically said what I was saying, in a slightly different way. The Gnome API is not stabilized, and commercial developers want stable APIs.
It can be a lot of work to overhaul or remove your state management when you need to move state into a database to be shared by other apps and languages.
If your data is inappropriate for a database, then it doesn't matter if it's in a C struct or a C++ class, they'll both be a bitch to convert. And if it is appropriate for a database, it's not going to be harder to port the OOP C++ version over the Structured C version.
Here's a OOP C++ member function:
void MyClass::doSomething()
{
m_var = somethingElse();
}
Here's the exact same function in non-OO C:
void doSomething(MyStruct *m)
{
m->var = somethingElse();
}
In terms of data, the only difference between them is encapsulation.
No no no! That would be too much like X11R6! Remember, this is Slashdot, where the remote capabilities of X are perceived as a hinderance to 3D game performance under Linux.
How about "I can run Framemaker on my FreeBSD box?" When you think about it, that's pretty damned cool.
Gnome appears to me to be at least as stable as CDE!
Stable does not merely mean that the software doesn't crash much. It also means that the software has a consistant and reliable API.
Not only do you want your desktop not to crash, you also want the software you write for it today to be valid tomorrow.
If you check, you'll find that Qt based Kylix (as well as other commercial Qt based libs and tools) doesn't require developers to pay Trolltech. Ever wonder why? Because Borland, which is significantly smaller than Sun, negotiated a simple deal with Trolltech. I don't know what the details are, but I suspect it's a small percentage of each Kylix purchase. Sun could easily do the same thing with the same triple licensing that Trolltech uses, so that non-commercial development on Solaris would be free.
starts becoming expensive to distribute as a corporate entity
Huh? It's free to distribute! Even the proprietary version!
But it's not free to develop for. You need a per-developer license to create non-free software with Qt. But once you've done that, you can distribute the finished product any way you want, including in ways in which you don't have to pay any with no royalties for anyone. You can even distribute the Qt runtimes royalty free!
No, it's not free-beer, not even for billion dollar companies like Sun and HP that might have twenty or so people actually developing with Qt. But you don't have to pay to distribute your Qt software.
HP wants to write commercial proprietary applications for GNOME. They cannot do that when the development has not stabilized. HP does not want to develop for and support a moving target, and their customers won't want to install a patch every week just because someone at GNOME changed the API. Geez, even Windows managed to keep a stable API through three different desktops, nine major release versions, and one complete decade!
So what about the paid GNOME developers at Redhat and Sun?
What does "stabilize" mean, anyway?
Well, since you're a GNOME user, I can understand why you don't understand the term, since it's so rare to see it. [ducks]
"Stable", among other things, means that the development APIs are not changing. It does NOT mean that development has stopped, only that they have finalized the interfaces, allowing other people to develop for it.
A stabilized GNOME 2 means that you don't have to rewrite your application next week when things change. Ideally, you shouldn't have to rewrite it when GNOME 3 comes out either. Consider the great unwashed evil that is KDE: the API is stable. It doesn't matter if you love or hate KDE, if you look at the project with an honest perspective, you have to agree that they have a relatively stable API. They may add new interfaces, but they keep their old ones as stable as possible. I ported several KDE 2 applications to KDE 3 for the FreeBSD ports collection. Average porting time was half an hour, including compilation and testing. And this was between MAJOR release versions!
An unstable API is a public announcement that the developers do not feel that the project is ready for public use, regardless of other statements to the contrary. GNOME is not alone in this regard, but that doesn't make the practice right.
Ooh ooh! Elitist alert! He thinks his system is better than mine.
Bollocks! All I need to do to get a complete GNOME installation under FreeBSD is:
make install
p.s. Yeah, I omitted the cvsup step, but you omitted your rsync step as well.
p.p.s. I can also install prebuilt binary packages so I can use GNOME now, and save the compile for a 2am cron job.
A mere five years ago a "multigigabyte" drive would have been immense. So you'll have to excuse these guys for not being 15 year old gaming geeks who need to upgrade their hardware every few days.
I would even argue that you don't really need inheritance.
:-)
Consider a simple GUI. In this GUI you have a button, a checkbox, a label and an edit box. All of these things are widgets. Does it not make sense to use a base widget class instead of rewriting all the common related functionality they all have? Now add a radio button. It's related to a checkbox. Perhaps it also makes sense to have the checkbox and radio button share a common ancestor as well.
I can hear people say "that's what functions are for!". True, that is what they are for. But if a function is only useful for a widget related structure, doesn't it make sense to encapsulate that function with that structure?
No, you don't need inheritance. But sometimes it's damned useful.
I've seen some very complex software written in functional languages that was very easy to follow even though the had no real OO concepts
So have I. But no one here is arguing that OOP is the only appropriate paradigm for all problem domains.
In the past I have made some comments on what I think plain old C could become if it incorporated some modulization features.
Be careful with this. If you don't watch yourself it could end up becoming Objective C
Don't lay this sin solely at OOP's door. We learned it from Structured Programmers, and are smugly pleased that Generic Programmers are knee deep in this iniquity as well.
Any formalized process can be taken to an extreme. And you yourself are arguing for a formalized process as well: "And never write OO when procedural programming can do the same trick in less space." Sometimes it might actually make sense to write it in OO, because other things may be more important than less space! Programming isn't a field of morality where you must choose a particular code of ethics to abide by no matter what.
the US that sold Saddam most of his chemical stockpile
The claim by the grandparent post was that the US that sold Saddam most of his chemical stockpile. Your references don't support that claim.
My point about 'power users' installing software is something that can be set and controlled so that the software can also be available to ALL users on the machine. Not just in the home directory for the user that installed it. I have seen too many senerios where there have been 10 different Mozilla installations for 10 different users. This is just ridiculous.
And I've seen the case where one user wanted to update StarOffice and ruined the install for everyone because they didn't do it right. Whether it's NT or UNIX, one person needs to be in charge of the system. It may be inconvenient, but as the adage says, convenience is the inverse of security.
Keeping UNIX variants on par with what NT was doing 10 years ago is still a process of catch up, and yet the UNIX 'ideology' is far older. And its age of older 'ideologies' is not a plus in the technology world, instead they have to be molded and kludged to make them mimic what is natural in newer OS architectures like NT.
Were you able to see past the single word "duh" and understand the point I was trying to make? UNIX is not obsolete, it's just different paradigm. You don't throw away a technology just because it's old, not even software technology.
The UNIX philosophy that there shall be one and only one ultimate administration account is no less valid than NT's philosophy that there can be multiple ultimate administration accounts. Both can be used to implement a practical easy-to-use system for home desktop systems. XP and OSX are prime examples.
The most important part of token/object security model is its practical effect. You can achieve the same practical effect with UNIX using owners and groups. But to make it easy, it needs to be done by the system integrator (the distro) and not by the system administrator.
This story is about Lycoris. There is no reason an "entry-level" UNIX of any sort cannot implement a model similar in effect to NT's if they wanted to. The problem with many newbie Linux distros is that they don't want to remove themselves too far from the traditional server model of Linux.