My body doesn't react well to giving blood. When I donate they have to rest a lot longer than normal before I can leave the bed, and then I need help getting to a chair. I'm okay after a while, but I can tell the rest of the day that things are not right. I've stopped giving blood (sad because I'm a low risk donor, but my body won't take it). I faint easily in normal situations anyway.
Just personal experience, but I suspect the original poster has similar experience.
I know people who have had the blood test done. They report that a lot of blood is taken. Not as much as a full donation, but not much less either. Still I'm with you, if they accuse me of being drunk I'll demand a blood test. (Since I never drink this is unlikely, but it could happen)
This machine has not been verified. Maybe it works, maybe it doesn't. It is up to the prosecution to prove it works. I'm innocent by that machine until you prove that it works. If there is even a.1% change that it is wrong, I'm innocent.
I want the drunks locked up. I do not want anyone who is not drunk to get locked up by mistake. We have not yet established that this machine can tell the difference to a reasonable degree of accuracy.
As I recall the banks always closed at 3pm, except on Friday they were open until 7, but anything done after 3pm Friday was just put in a box and not processed until the next Monday.
I'm told that it was because they didn't have computers back then, so everything was processed by hand, and they used the last 2 hours to balance the books. I don't know that I believe that though - I'm young enough that computers have always been around in banks. (They didn't reach general business until latter, but computers in banks were old news by then)
About a month ago there was an article on slashdot about spyware that bypassed SSL. (They of course claim they are not spyware) Just install a certificate of yourself into the machine, then set up all connections to proxy through your machine. Then just generate whatever keys you need to sign any page they connect to.
Are you trying to tell me that if I go to Trolltech and say "We made a mistake, and used the freeversion. How do I make this right so we can release our software?" They won't come up with something? I'm sure it will be more than if we had just bought the license like we were supposed to. I'm sure an offer of twice what I would have had to pay by doing it right to start with would be accepted.
Of course they could refuse to come to any agreement, but will they? Money is on the table, if they remain reasonable they can take it. If not a good product gets written off, and the company refuses to use qt for any other products.
If I act alone you are correct. However if all Linux users would act like me, that would make a difference. 5% extra return rate on products that are not linux compatible will get the retailer to start an investigation to find out what is up. Their goal is lower prices and all those returns are cutting into their profit meaning they need to charge more. They won't like that, so they will demand the supplier do something about it.
It is one thing to know how long something will take, but delivering features should be a business decision.
When you say you can do something, and they agree it would be nice, the proper response is "Settle the price with my boss and I'll get right on it." Given your experience with then original meeting you should then secretly write down on paper your time estimate so the boss has a clue how to bill.
Actually this isn't correct yet. Before the meeting you tell your boss that you can do something and you think they will want it. Then let the boss bring it up. Remember your job is in large part to make your boss look good, and your advance on his coat tails. So tell him to deal with it. If it is trivial you do it before hand, but #ifdef out (or whatever your language does), so that you can deliver when the boss says it is ready.
From their misguided point of view: Linux is only a tiny fraction of the market. We can afford to loose that fraction, as it isn't worth the costs to do anything to get it.
I'll leave pointing out the flaws to someone else.
Nearly all of that documentation needs to be written anyway. It is the only reasonable way for the programmers to know how to write the drivers.
Oh sure, you could have the programmers bug the hardware guys directory for every detail, but between the programmer forgetting (where is foo again? Bit 7 or byte offset 0xff3a...), and the interuptions of the train of thought for both guys, you need the paper anyway. If as an investor I found out you used that process I would start a shareholder lawsuit - when someone quits you loose all their knowledge, and that is criminal.
Now there is some more effort in making it publicly releaseable, but it should be much above what you have to do. If it is your documentation isn't usefull internally either.
Your analogy is wrong. You have consider taking I5, and CA-22. How you never considered teleportation, helicopter, or airplanes. Your thinking is constrained because the US and your state put a lot of money into building roads.
If your state had not put money into roads, and thus your road choices were dirt (not even gravel) tracks, you would have developed other modes of transportation. Because you are not the only one in the position of needing to get home from work, there would be great demands for whatever you use instead.
Lets call that replacement helicopters, to pick one. Because so many people are buying helicopters, there would be hundreds of models to choose from. The Japanese would have been beating the US's lunch in the 70s, but today the Koreans would be trying hard, with the Chinese looking for their way into the market.
Would this be better or worse? We will never know because we have a road system that is already a sunk cost.
I have in fact read the GPL. Did it back in 1998 in fact. Since the GPL hasn't changed since then I don't have to read it again. The BSD license hasn't changed in a while either. (just the drop of the advertising clause)
I read the EULA when I installed XP SP2 as well, and thought "Thank God I'm only agreeing to this on behalf of my company, I don't think I'd agree for my personal machines." I don't have Java installed on my home FreeBSD machine because I didn't like the EULA (I object to the provision that it can't be automatically downloaded and installed by FreeBSD's port system). Not to mention all the other license agreements I have to deal with.
We can turn Linux support into a downhill battle if we try. Everytime you want to buy hardware, find out what doesn't work in linux.
Go to the local retailer
buy the version that doesn't work
Open the box
install it
Find it doesn't work
place back in box with all parts
bring back to store
Tell clerk that you are returning it because it isn't linux compatible.
Repeat until you have tried all non-working parts
Buy something that does work (from same retailer if they have it)
Get all your friends to do the same.
The last step is important. There are enough linux users that we can make this work for hardware - IF we all work together to make it unprofitable to carry anything without linux drivers.
I live near Best Buy's headquarters. No Fry's in this state, or even the next one (AFAIK). Nobody in my area cares about Linux, yet I know there are many linux users around here.
I couldn't care less how much linux is on the shelf at best buy. I'm a BSD guy by choice, so I wouldn't have a use for it anyway. Put all the Windows software on the shelf you want, I don't care.
I want hardware that will work. When I want a wireless adapter for my laptop I want it today, with no hassles otherwise I'd buy it mail order. So I often find myself in Best Buy looking at some box, and wondering if it will work on my system.
My solution: research. First I find out what will work with BSD, and what will not. Then I go in, and buy something that will not. Open the box, installed it and play a little, and sure enough, it won't work with BSD - return it. Buy the part that does work. I'm doing my best to make it expensive to stock hardware that isn't BSD compatible.
Flame retardant material used on the space ship to protect the astronauts is used in fire fighter equipment.
Not to mention flame retardant material developed for use in movies.
Microwave (you know the stuff people use to cook with) was invented for astronauts.
No, Microwave was invented when engineers noticed that candybars in the pockets started to melt when they stood in front of WWII radar. (which wasn't a smart thing to do, but in the war you cared more about winning than your own life). After the war those engineers worked on making a product out of it. The first Microwaves were used on luxury ships because they were too large for any home (note that today many home microwaves have a large oven chamber and more power than those first ones)
Now I will grant that some things have come sooner because of space research. However what is the cost in things that we could have now if engineers hadn't been focused on space? This question cannot of course be answered, which is why I reject all arguments that space was really good for us. Maybe it was, maybe it wasn't, since we don't have a proper scientific controlled experiment we cannot know.
The human factor of umpires that are failable make the game. Despite a few who are dishonest, most officials try their best to be fair, even when it is against the team they want to win.
I don't want to see robots play sports. In theory (the Jetson's universe) watching robots play is just a case of waiting to see who's bearings wear out sooner.
Now in amateur sports a robot ump would be nice. When it is just me and someone at my level on the racquetball court it would be nice to have something that knew all the rules to tell them to us, not to mention call violations where neither of us know that rule. However I cannot afford to pay for a device to call racquetball games, nor could I afford membershim in a gym that would have it.
Well your are off base, but not for obvious reasons. The problem is the monopoly has no incentive to cut costs, so they end up with a lot of wasted overhead that doesn't go to infrastructure improvements.
If your monopoly company would be just as competitive as they would be without the monopoly, then you would be right: the monopoly is better. However as a monopoly they have no incentive to actually watch the bottom line (after all they can raise prices to cover any loss), nor is there incentive to upgrade the infrastructure, because you don't have any ability to do something about it if they don't.
In the real world non-monopolies would have to compete, which means they would have to upgrade their network to handle new technologies that consumers want. If they don't upgrade all they can compete on is price and that is a race to the bottom - a situation no investor wants from his company as one wrong move will put your out of business. By upgrading the network you have something the other guy doesn't, so you can encourage people to switch for a higher cost.
Start with your local PUC (public utilities commission). Then contact your congressmen. Seriously, it can work if you make enough noise.
The phone company is allowed to have their wires run across your property only because they are serving the public good. Make it clear that if they won't give you good service you will reject their right to run wires across your property. You can't do this alone, but the PUC is assigned the job of distributing that power by congress. So start with the PUC, and if they refuse to help (which they will) talk to your congressmen.
The best way to contact your congressmen is with a face to face meeting. Do your research (know all the facts, practice your presentation with others. Then ask for a 15 minute meeting and deliver it. Make sure your presentation is less than 10 minutes, leaving the rest for questions (which you better be prepared to answer - get a devils advocate to write them). Do your best to be out before your time is up unless they insist.
Congressmen want corporate money to buy votes. So make sure your vote is for sale ONLY to someone who will take care of the issues you want them to take care of. And make sure you talk to your friends and neighbors about it. Come election time knock on all the doors in your neighborhood and tell everyone who to vote for - this is worth far more than any amount of money the phone company can counter with, but only if you do it.
If you are like most people though you will decide that the years it takes to change things is not worth it. Enjoy your beer on the deck or whatever, but don't be surprised when nothing changes.
The US rail system is well managed, with one exception: Amtrak. The US railroads have realized that freight does not care too much about how fast it is going, sitting still waiting for another train to pass, and not taking the shortest route point to point.
So the US rails have decided to focus on freight where they hold nearly 2/3rds of all traffic (compare to less than 1/3rd for Europe's rails). That is good management: do what you can do well, and let someone else deal with what you cannot do well. I would argue that Europe's rails are mismanaged, spending all their energy on moving people when it is much easier to move freight.
The point of a journaling file system is that you don't need to run fsck
No, the point is that fsck is very quick because you can read the journal so see all the places where something could possibly be wrong, instead of having to read the entire disk. On power failure your fsck times change from several hours (think a several terabyte RAID drive), to just a few seconds.
Mounting a several-hundred GB Reiser partition takes a few seconds, even if it was not cleanly unmounted. How much faster do you want that to be?
With softupdates mounting a several hundred GB partition is even less time. Of course you have to do a full fsck, but you can do that in the background whenever you get around to it.
Though I agree that it this point it is a stupid measure - if the difference might matter you should have a cluster with mirrors so that you can stand for an entire system to be down for hours.
Softupdates is very invasive to the filesystem. For some loads they are much faster (no need to write the journal to say what you are going to do, write the disk to do it, and then erase that part journal to say it is done). For other loads it is slower (I forget what offhand, but I don't want to give the impression that softupdates are always better - they are not)
For most people: the differences are trivial, and will not matter in your real world useage. Both are much better than the old style. Use whatever is native to your OS, it doesn't matter. There are a few professionals who deal with systems where the differences matter - leave figuring this whole mess to them.
Speaking as an Emacs guy, I'd be tempted to mod it flamebait. As would the vi users I work with.
Choice of editors is personal. So is choice of IDEs. People keep telling me that kdevelop is great. They might be right, but I like my Emacs.
If the above is too subtile: you are mistaking personal preference (likely influenced by the first useful thing you learned, like my preference) for better.
Now since you like Visual Studio, you are stuck with Windows. (Unless you decide something like Kdevelop is better one day). Perfectly valid response, but in light of all the vi-Emacs wars funny is the correct way to look at it.
If someone's going to do a new application, it's much more likely to be a Windows application.
Exactly. So why not create your software for linux and/or the Mac where you won't have as much competition. Number of potential users is not important, what is important is number of sales. If you don't dominate the Windows market you will get just a tiny share of the Windows sales. Same for the mac, but odds are much better than you can dominate and get nearly all the Mac sales.
A Chevy S10, 2wd, with the 4 cylinder engine can get as much as 33 mpg on the highway. That is as good as many tiny cars. (Sure the Geo Metro got 45, but that is a rare exception) reference (Yes I know the reference lists 28mpg - I used to have a 1988 model that got 33, emission controls are tighter now accounting to the difference)
That is also the only answer I like. I've seen many answers about round being space saving, or easiest to position manhole covers (any rotation will work). Problem is manholes are not bought by bureaucrats who do not have to actually manipulate the covers themselves. Thus practical considerations to the workmen are not accounted for.
Legend has this that this riddle was first used on the Microsoft Campus, where all the manholes are in fact rectangular (including square) not round.
Slashdot Mirror
My body doesn't react well to giving blood. When I donate they have to rest a lot longer than normal before I can leave the bed, and then I need help getting to a chair. I'm okay after a while, but I can tell the rest of the day that things are not right. I've stopped giving blood (sad because I'm a low risk donor, but my body won't take it). I faint easily in normal situations anyway.
Just personal experience, but I suspect the original poster has similar experience.
I know people who have had the blood test done. They report that a lot of blood is taken. Not as much as a full donation, but not much less either. Still I'm with you, if they accuse me of being drunk I'll demand a blood test. (Since I never drink this is unlikely, but it could happen)
This machine has not been verified. Maybe it works, maybe it doesn't. It is up to the prosecution to prove it works. I'm innocent by that machine until you prove that it works. If there is even a .1% change that it is wrong, I'm innocent.
I want the drunks locked up. I do not want anyone who is not drunk to get locked up by mistake. We have not yet established that this machine can tell the difference to a reasonable degree of accuracy.
As I recall the banks always closed at 3pm, except on Friday they were open until 7, but anything done after 3pm Friday was just put in a box and not processed until the next Monday.
I'm told that it was because they didn't have computers back then, so everything was processed by hand, and they used the last 2 hours to balance the books. I don't know that I believe that though - I'm young enough that computers have always been around in banks. (They didn't reach general business until latter, but computers in banks were old news by then)
About a month ago there was an article on slashdot about spyware that bypassed SSL. (They of course claim they are not spyware) Just install a certificate of yourself into the machine, then set up all connections to proxy through your machine. Then just generate whatever keys you need to sign any page they connect to.
Are you trying to tell me that if I go to Trolltech and say "We made a mistake, and used the freeversion. How do I make this right so we can release our software?" They won't come up with something? I'm sure it will be more than if we had just bought the license like we were supposed to. I'm sure an offer of twice what I would have had to pay by doing it right to start with would be accepted.
Of course they could refuse to come to any agreement, but will they? Money is on the table, if they remain reasonable they can take it. If not a good product gets written off, and the company refuses to use qt for any other products.
If I act alone you are correct. However if all Linux users would act like me, that would make a difference. 5% extra return rate on products that are not linux compatible will get the retailer to start an investigation to find out what is up. Their goal is lower prices and all those returns are cutting into their profit meaning they need to charge more. They won't like that, so they will demand the supplier do something about it.
It is one thing to know how long something will take, but delivering features should be a business decision.
When you say you can do something, and they agree it would be nice, the proper response is "Settle the price with my boss and I'll get right on it." Given your experience with then original meeting you should then secretly write down on paper your time estimate so the boss has a clue how to bill.
Actually this isn't correct yet. Before the meeting you tell your boss that you can do something and you think they will want it. Then let the boss bring it up. Remember your job is in large part to make your boss look good, and your advance on his coat tails. So tell him to deal with it. If it is trivial you do it before hand, but #ifdef out (or whatever your language does), so that you can deliver when the boss says it is ready.
Microsoft stopped doing that years ago. Such bundling killed OS/2, but the government stepped in and killed such arrangements years ago.
From their misguided point of view: Linux is only a tiny fraction of the market. We can afford to loose that fraction, as it isn't worth the costs to do anything to get it.
I'll leave pointing out the flaws to someone else.
Nearly all of that documentation needs to be written anyway. It is the only reasonable way for the programmers to know how to write the drivers.
Oh sure, you could have the programmers bug the hardware guys directory for every detail, but between the programmer forgetting (where is foo again? Bit 7 or byte offset 0xff3a...), and the interuptions of the train of thought for both guys, you need the paper anyway. If as an investor I found out you used that process I would start a shareholder lawsuit - when someone quits you loose all their knowledge, and that is criminal.
Now there is some more effort in making it publicly releaseable, but it should be much above what you have to do. If it is your documentation isn't usefull internally either.
Your analogy is wrong. You have consider taking I5, and CA-22. How you never considered teleportation, helicopter, or airplanes. Your thinking is constrained because the US and your state put a lot of money into building roads.
If your state had not put money into roads, and thus your road choices were dirt (not even gravel) tracks, you would have developed other modes of transportation. Because you are not the only one in the position of needing to get home from work, there would be great demands for whatever you use instead.
Lets call that replacement helicopters, to pick one. Because so many people are buying helicopters, there would be hundreds of models to choose from. The Japanese would have been beating the US's lunch in the 70s, but today the Koreans would be trying hard, with the Chinese looking for their way into the market.
Would this be better or worse? We will never know because we have a road system that is already a sunk cost.
I have in fact read the GPL. Did it back in 1998 in fact. Since the GPL hasn't changed since then I don't have to read it again. The BSD license hasn't changed in a while either. (just the drop of the advertising clause)
I read the EULA when I installed XP SP2 as well, and thought "Thank God I'm only agreeing to this on behalf of my company, I don't think I'd agree for my personal machines." I don't have Java installed on my home FreeBSD machine because I didn't like the EULA (I object to the provision that it can't be automatically downloaded and installed by FreeBSD's port system). Not to mention all the other license agreements I have to deal with.
We can turn Linux support into a downhill battle if we try. Everytime you want to buy hardware, find out what doesn't work in linux.
Go to the local retailer
buy the version that doesn't work
Open the box
install it
Find it doesn't work
place back in box with all parts
bring back to store
Tell clerk that you are returning it because it isn't linux compatible.
Repeat until you have tried all non-working parts
Buy something that does work (from same retailer if they have it)
Get all your friends to do the same.
The last step is important. There are enough linux users that we can make this work for hardware - IF we all work together to make it unprofitable to carry anything without linux drivers.
I live near Best Buy's headquarters. No Fry's in this state, or even the next one (AFAIK). Nobody in my area cares about Linux, yet I know there are many linux users around here.
I couldn't care less how much linux is on the shelf at best buy. I'm a BSD guy by choice, so I wouldn't have a use for it anyway. Put all the Windows software on the shelf you want, I don't care.
I want hardware that will work. When I want a wireless adapter for my laptop I want it today, with no hassles otherwise I'd buy it mail order. So I often find myself in Best Buy looking at some box, and wondering if it will work on my system.
My solution: research. First I find out what will work with BSD, and what will not. Then I go in, and buy something that will not. Open the box, installed it and play a little, and sure enough, it won't work with BSD - return it. Buy the part that does work. I'm doing my best to make it expensive to stock hardware that isn't BSD compatible.
Flame retardant material used on the space ship to protect the astronauts is used in fire fighter equipment.
Not to mention flame retardant material developed for use in movies.
Microwave (you know the stuff people use to cook with) was invented for astronauts.
No, Microwave was invented when engineers noticed that candybars in the pockets started to melt when they stood in front of WWII radar. (which wasn't a smart thing to do, but in the war you cared more about winning than your own life). After the war those engineers worked on making a product out of it. The first Microwaves were used on luxury ships because they were too large for any home (note that today many home microwaves have a large oven chamber and more power than those first ones)
Now I will grant that some things have come sooner because of space research. However what is the cost in things that we could have now if engineers hadn't been focused on space? This question cannot of course be answered, which is why I reject all arguments that space was really good for us. Maybe it was, maybe it wasn't, since we don't have a proper scientific controlled experiment we cannot know.
The human factor of umpires that are failable make the game. Despite a few who are dishonest, most officials try their best to be fair, even when it is against the team they want to win.
I don't want to see robots play sports. In theory (the Jetson's universe) watching robots play is just a case of waiting to see who's bearings wear out sooner.
Now in amateur sports a robot ump would be nice. When it is just me and someone at my level on the racquetball court it would be nice to have something that knew all the rules to tell them to us, not to mention call violations where neither of us know that rule. However I cannot afford to pay for a device to call racquetball games, nor could I afford membershim in a gym that would have it.
Well your are off base, but not for obvious reasons. The problem is the monopoly has no incentive to cut costs, so they end up with a lot of wasted overhead that doesn't go to infrastructure improvements.
If your monopoly company would be just as competitive as they would be without the monopoly, then you would be right: the monopoly is better. However as a monopoly they have no incentive to actually watch the bottom line (after all they can raise prices to cover any loss), nor is there incentive to upgrade the infrastructure, because you don't have any ability to do something about it if they don't.
In the real world non-monopolies would have to compete, which means they would have to upgrade their network to handle new technologies that consumers want. If they don't upgrade all they can compete on is price and that is a race to the bottom - a situation no investor wants from his company as one wrong move will put your out of business. By upgrading the network you have something the other guy doesn't, so you can encourage people to switch for a higher cost.
Start with your local PUC (public utilities commission). Then contact your congressmen. Seriously, it can work if you make enough noise.
The phone company is allowed to have their wires run across your property only because they are serving the public good. Make it clear that if they won't give you good service you will reject their right to run wires across your property. You can't do this alone, but the PUC is assigned the job of distributing that power by congress. So start with the PUC, and if they refuse to help (which they will) talk to your congressmen.
The best way to contact your congressmen is with a face to face meeting. Do your research (know all the facts, practice your presentation with others. Then ask for a 15 minute meeting and deliver it. Make sure your presentation is less than 10 minutes, leaving the rest for questions (which you better be prepared to answer - get a devils advocate to write them). Do your best to be out before your time is up unless they insist.
Congressmen want corporate money to buy votes. So make sure your vote is for sale ONLY to someone who will take care of the issues you want them to take care of. And make sure you talk to your friends and neighbors about it. Come election time knock on all the doors in your neighborhood and tell everyone who to vote for - this is worth far more than any amount of money the phone company can counter with, but only if you do it.
If you are like most people though you will decide that the years it takes to change things is not worth it. Enjoy your beer on the deck or whatever, but don't be surprised when nothing changes.
The US rail system is well managed, with one exception: Amtrak. The US railroads have realized that freight does not care too much about how fast it is going, sitting still waiting for another train to pass, and not taking the shortest route point to point.
So the US rails have decided to focus on freight where they hold nearly 2/3rds of all traffic (compare to less than 1/3rd for Europe's rails). That is good management: do what you can do well, and let someone else deal with what you cannot do well. I would argue that Europe's rails are mismanaged, spending all their energy on moving people when it is much easier to move freight.
The point of a journaling file system is that you don't need to run fsck
No, the point is that fsck is very quick because you can read the journal so see all the places where something could possibly be wrong, instead of having to read the entire disk. On power failure your fsck times change from several hours (think a several terabyte RAID drive), to just a few seconds.
Mounting a several-hundred GB Reiser partition takes a few seconds, even if it was not cleanly unmounted. How much faster do you want that to be?
With softupdates mounting a several hundred GB partition is even less time. Of course you have to do a full fsck, but you can do that in the background whenever you get around to it.
Though I agree that it this point it is a stupid measure - if the difference might matter you should have a cluster with mirrors so that you can stand for an entire system to be down for hours.
Softupdates is very invasive to the filesystem. For some loads they are much faster (no need to write the journal to say what you are going to do, write the disk to do it, and then erase that part journal to say it is done). For other loads it is slower (I forget what offhand, but I don't want to give the impression that softupdates are always better - they are not)
For most people: the differences are trivial, and will not matter in your real world useage. Both are much better than the old style. Use whatever is native to your OS, it doesn't matter. There are a few professionals who deal with systems where the differences matter - leave figuring this whole mess to them.
Speaking as an Emacs guy, I'd be tempted to mod it flamebait. As would the vi users I work with.
Choice of editors is personal. So is choice of IDEs. People keep telling me that kdevelop is great. They might be right, but I like my Emacs.
If the above is too subtile: you are mistaking personal preference (likely influenced by the first useful thing you learned, like my preference) for better.
Now since you like Visual Studio, you are stuck with Windows. (Unless you decide something like Kdevelop is better one day). Perfectly valid response, but in light of all the vi-Emacs wars funny is the correct way to look at it.
If someone's going to do a new application, it's much more likely to be a Windows application.
Exactly. So why not create your software for linux and/or the Mac where you won't have as much competition. Number of potential users is not important, what is important is number of sales. If you don't dominate the Windows market you will get just a tiny share of the Windows sales. Same for the mac, but odds are much better than you can dominate and get nearly all the Mac sales.
A Chevy S10, 2wd, with the 4 cylinder engine can get as much as 33 mpg on the highway. That is as good as many tiny cars. (Sure the Geo Metro got 45, but that is a rare exception) reference (Yes I know the reference lists 28mpg - I used to have a 1988 model that got 33, emission controls are tighter now accounting to the difference)
That is also the only answer I like. I've seen many answers about round being space saving, or easiest to position manhole covers (any rotation will work). Problem is manholes are not bought by bureaucrats who do not have to actually manipulate the covers themselves. Thus practical considerations to the workmen are not accounted for.
Legend has this that this riddle was first used on the Microsoft Campus, where all the manholes are in fact rectangular (including square) not round.