I just thought of this while I was reading the article summary, so this isn't exactly well thought out, but...
I'm surprised some enterprising sort hasn't created a blacklist for use by mailing list operators that tracks the likelihood of a domain's customers illegitimately reporting valid mail as spam. Then, newsletter admins could use that score as a guideline to how many hoops a would-be subscriber has to jump through before getting added to the list.
Coming in from a private domain that's never mis-reported ham as spam? Your reply to the confirmation email is enough to subscribe you. Signing up from moron.com with a mis-reporting likelihood of 35%? You can't subscribe until your mailserver admins have also acknowledged a confirmation message explaining what you're asking for and that you've already explicitly asked to do it.
Hmmm, I've been looking for a new project to start...
I understand that, but I can quite easily imagine the inner loop of a compute intensive task that might see a huge benefit from filling 90% of available cache with a larger number of more efficient instructions instead of 70% with a small number of slower opcodes.
I'll admit that's probably not the most common case, and for general use then your scenario would probably give better system wide performance. Still, it's not really possible to say that smaller (or larger) binaries are "better" - there're too many other variables to account for.
As far as I'm concerned, unless you're using "-Os" because you're deliberately building small binaries at the expense of all else - say, for embedded development - the resulting binary size is completely irrelevant as a compiler benchmark. What if the smaller result uses a slower, naive algorithm (which in this case would mean choosing an obviously-correct set of opcodes to implement a line of C instead of a less-obvious but faster set)?
Second, the runtime benchmarks were close enough to be statistically meaningless in most cases. The author concludes with:
Is GCC 4.0 better than its predecessors?
In terms of raw numbers, the answer is a definite "no".
My take would have been "in terms of raw numbers, it's not really any better yet." It's close enough to equal (and slower in few enough cases that I'd be willing to accept them), though, that I'd be willing to switch to it if I could do so without having to modify a lot of incompatible code. It's clearly the way of the future, and as long as it's not worse than the current gold standard, why not?
But what you could do is slip into an alternate universe which is exactly like ours...
...at which point "an alternate universe" becomes the new "Canada" where the hypothetical girlfriends of young geeks live: "I went to Probability 5439954634 with my parents this summer, and I met this great girl, but her parents won't let her come see me (or write or call)."
You move away from human readability/editability and into the realm where only a machine can fathom the format.
You misspelled "also" as "only". Having a featureful editor than groks every single config file on my system, supports context-sensitive help for each possible option, and shows me what values are legal for all of them sounds like a Good Thing. Wouldn't you like an "etc-mode" for Emacs, or GEtcEditor, or KonfigEditor, or similar?
This should also be eagerly awaited by network administrators who occasionally have to roll out the same config change to hundreds of servers. A command-line XML config file editor should be trivially easy to write, and I image "GNU etcmanager" would be available in a few months' time.
The college grads aren't going to impress much with their stunning knowledge of literature or micro-biology.
One of my interviewers (in America) asked me a question in German. I replied in Spanish. They were amused enough that I got the job. No class you'll take in a trade school will help you with those kinds of curveballs.
It's also not unheard of for a boss to drag his admin in front of some customers, and I assure you that they probably won't want to talk shop over dinner. Being able to discuss philosophy, art, and current events goes a long way.
No, I'll stand by my point that most of the interesting employers want a well-rounded applicant. I don't want to sound harsh, but trade college graduates are a dime a dozen in the post dot-bomb world, and given two applicants with comparable technical skills, most employers will pick the one that can do other stuff if the need arises.
The University will waste your time and money with classes that have no bearing on your job. The University usually focuses on theory and generalizations of concepts with some specifics. Universities are more interested in creating a well rounded educated employee. But, employers don't care if you took Western Civ.
My boss pays me to be well-spoken and literate so that when I come up with a new way to make his system work better, I can explain it to him and our customers. I don't know where all of these one-track employers are, outside of Dilbert, but the ones I've worked with hire people and not androids. Then again, I've been fortunate enough to afford to be a bit picky about who I work with. Is it really that bad out there?
I use Python for almost all new development. Why? Because it offers at least a bit of Lisp's functional goodness, but is similar enough to the other common programming languages that other people actually use it. If I were writing purely for my own entertainment without having to interact with others, I'd probably take a closer look at Lisp. However, that's a luxury I can't afford right now and I've therefore had to pick something a little more palatable to the people around me.
That doesn't mean that I can't use some of its concepts, though. Our company's application server (running on Zope) makes heavy use of passing functions here and there ("report generator, make this data into a PDF, and use this dictionary of lambda functions to format the data in it and use this other function to generate the filename based on the results"). With Python, we have the option of using baby steps that my eventual replacement can understand, rather than having to find someone willing to jump directly into a Lisp codebase with no prior experience.
Is that the ideal? No, but it's certainly not awful, either.
First, me too on the tab completion. I'm partial to Zsh's implementation, but I was the maintainer for FreeBSD's bash-completion port for a while so I'm sort of fond of it, too. Basically, it's crack for CLI users - once they've had it, there's no going back.
Second, setting up a bunch of "-i" aliases is an extremely bad idea for newbies. Considering that they're starting with a blank slate, this is the time when they can most afford to learn that the machine will always do what they ask and not what they mean. Do you really want them to get used to relying on rm -i not nuking their home directory and then having it disappear one day after they make a mistake in their.bashrc and the alias doesn't get set? Train them to think before they hit enter and make backups often.
Finally, my first original comment: place a call to fortune(6) in their login script, learn to make a fortune file, and start treating it like a wiki or blog. Have a tip for your users? Put it in the fortune file. A user makes a particularly brilliant mistake that you want to warn the others about? Put it in the fortune file. The randomness of it is a lot more interesting than the ol' MOTD that everyone trains themselves to ignore.
I'm with you. "Because you can" is a poor excuse for attempting something like that. There are plenty of F/OSS and proprietary development environments that are infinitely more suitable for application development than a scripted spreadsheet ever could be.
Put another way, I don't do accounting in Python - why would I want to write applications in Excel? Spreadsheets are the right tool for quite a few jobs, but this isn't one of them.
I've got like 100 gigs of bandwidth to use before the end of the month.
Did you really... No, he didn't - did he? OK, then. Can you the distant thunder of a herd of clicking Slashbots? Got a fire extinguisher in the server room?
What's better from a marketing perspective -- an ad that 50% of your potential audience really likes, and 50% of of them really dislike, or an ad that 100% of your audience likes?
I'll take "50% really likes" over "100% merely likes" any day of the week.
The fact is that many of us who've been in the business for a few years hate Network Solutions (I've explained it in more detail during other discussions). Seeing them fall from the top seat is excellent news and someone we wish would have happened a long time ago.
MetalliQaZ was right: competition is good. I mean, if Network Solutions is this universally despised today, imagine how much we all liked them years ago when they were the only game in town.
Let's see: $61 million - (100 employes * $60000 [1]) - DNS and web infrastructure of $1,000,000 [2] - Super Bowl ad of $2,000,000 = $52 million in profit.
I'd have to be of by orders of magnitude in a few areas for that not to be a sure-fire profit center.
[1] I doubt it takes 100 professional-level employees to run something like that, but you never know.
[2] Remember, you're only hosting the "domain parking" pages, and DNS for people with sites so small that the marginal costs per site are almost zero. The big customers (if any) pay extra for full-fledged services.
The thing is, though, that I liked the ad. As long as people who agree with me outnumber the people who share your opinion, then they're on the right track.
I'm glad that you're working on it, just as the rest of us are (and have been), but I think that "making maintenance easier" is a far cry from obsoleting an entire class of workers.
I don't want to make an enemy here, but you've hit on one of my personal hot buttons.
I generally like your posts, but this one was kind of dumb. Look, we've been hearing this promise for about 30 years now, and I don't think it's any more true today than it was then. The fact is that companies staff all of their mission-critical business functions and probably always will.
Examples? My company is not a shipper, but we have a full-time employee that handles shipping arrangements, puts incoming parcels where the belong, and has outgoing boxes ready when FedEx gets here. We're also not a staffing company, but we have an HR person. Neither are we a construction company, but we have a maintenance guy who also remodels our building as needed. Finally, we're not an IT consultant, but we have IT people on staff.
IT people will go away whenever companies no longer use IT. Until then, every place that depends on their services for daily operation will have employees that run them, just as they also have shipping, HR, and maintenance workers. I like your company (and would like them even more if you sent some free stuff my way, hint-hint), but you've done an excellent job of advancing the state of the art of the computers on the average employee's desk. That's just the tip of the iceburg for a lot of us, and no amount of CUPS-style printer autoconfiguration will change it.
Private or custom software is software developed for one user (typically an organization or company). That user keeps it and uses it, and does not release it to the public either as source code or as binaries.
[...]
In general we do not believe it is wrong to develop a program and not release it. There are occasions when a program is so useful that withholding it from release is treating humanity badly. However, most programs are not that marvelous, and withholding them is not particularly harmful. Thus, there is no conflict between the development of private or custom software and the principles of the free software movement.
It seems that proprietary software meant to control a piece of hardware that a company is selling as a stand-alone "black box" unit would be OK by RMS. In other words, in the cases I mentioned, the purchaser is really buying an MRI unit or an ABS system - the fact that it comes with software is mostly beside the point.
Slashdot Mirror
I'm surprised some enterprising sort hasn't created a blacklist for use by mailing list operators that tracks the likelihood of a domain's customers illegitimately reporting valid mail as spam. Then, newsletter admins could use that score as a guideline to how many hoops a would-be subscriber has to jump through before getting added to the list.
Coming in from a private domain that's never mis-reported ham as spam? Your reply to the confirmation email is enough to subscribe you. Signing up from moron.com with a mis-reporting likelihood of 35%? You can't subscribe until your mailserver admins have also acknowledged a confirmation message explaining what you're asking for and that you've already explicitly asked to do it.
Hmmm, I've been looking for a new project to start...
Signed, a parent of three.
I'll admit that's probably not the most common case, and for general use then your scenario would probably give better system wide performance. Still, it's not really possible to say that smaller (or larger) binaries are "better" - there're too many other variables to account for.
Second, the runtime benchmarks were close enough to be statistically meaningless in most cases. The author concludes with:
My take would have been "in terms of raw numbers, it's not really any better yet." It's close enough to equal (and slower in few enough cases that I'd be willing to accept them), though, that I'd be willing to switch to it if I could do so without having to modify a lot of incompatible code. It's clearly the way of the future, and as long as it's not worse than the current gold standard, why not?
...at which point "an alternate universe" becomes the new "Canada" where the hypothetical girlfriends of young geeks live: "I went to Probability 5439954634 with my parents this summer, and I met this great girl, but her parents won't let her come see me (or write or call)."
You misspelled "also" as "only". Having a featureful editor than groks every single config file on my system, supports context-sensitive help for each possible option, and shows me what values are legal for all of them sounds like a Good Thing. Wouldn't you like an "etc-mode" for Emacs, or GEtcEditor, or KonfigEditor, or similar?
This should also be eagerly awaited by network administrators who occasionally have to roll out the same config change to hundreds of servers. A command-line XML config file editor should be trivially easy to write, and I image "GNU etcmanager" would be available in a few months' time.
One of my interviewers (in America) asked me a question in German. I replied in Spanish. They were amused enough that I got the job. No class you'll take in a trade school will help you with those kinds of curveballs.
It's also not unheard of for a boss to drag his admin in front of some customers, and I assure you that they probably won't want to talk shop over dinner. Being able to discuss philosophy, art, and current events goes a long way.
No, I'll stand by my point that most of the interesting employers want a well-rounded applicant. I don't want to sound harsh, but trade college graduates are a dime a dozen in the post dot-bomb world, and given two applicants with comparable technical skills, most employers will pick the one that can do other stuff if the need arises.
My boss pays me to be well-spoken and literate so that when I come up with a new way to make his system work better, I can explain it to him and our customers. I don't know where all of these one-track employers are, outside of Dilbert, but the ones I've worked with hire people and not androids. Then again, I've been fortunate enough to afford to be a bit picky about who I work with. Is it really that bad out there?
That doesn't mean that I can't use some of its concepts, though. Our company's application server (running on Zope) makes heavy use of passing functions here and there ("report generator, make this data into a PDF, and use this dictionary of lambda functions to format the data in it and use this other function to generate the filename based on the results"). With Python, we have the option of using baby steps that my eventual replacement can understand, rather than having to find someone willing to jump directly into a Lisp codebase with no prior experience.
Is that the ideal? No, but it's certainly not awful, either.
You misspelled "Best Buy".
Second, setting up a bunch of "-i" aliases is an extremely bad idea for newbies. Considering that they're starting with a blank slate, this is the time when they can most afford to learn that the machine will always do what they ask and not what they mean. Do you really want them to get used to relying on rm -i not nuking their home directory and then having it disappear one day after they make a mistake in their .bashrc and the alias doesn't get set? Train them to think before they hit enter and make backups often.
Finally, my first original comment: place a call to fortune(6) in their login script, learn to make a fortune file, and start treating it like a wiki or blog. Have a tip for your users? Put it in the fortune file. A user makes a particularly brilliant mistake that you want to warn the others about? Put it in the fortune file. The randomness of it is a lot more interesting than the ol' MOTD that everyone trains themselves to ignore.
Put another way, I don't do accounting in Python - why would I want to write applications in Excel? Spreadsheets are the right tool for quite a few jobs, but this isn't one of them.
Oh, I know. You just happened to be on the receiving end of me being argumentative today. :-)
Did you really... No, he didn't - did he? OK, then. Can you the distant thunder of a herd of clicking Slashbots? Got a fire extinguisher in the server room?
I'll take "50% really likes" over "100% merely likes" any day of the week.
Ironically, I (the one who defended the ad) am one of those hated red-staters. You need to update your prejudices.
MetalliQaZ was right: competition is good. I mean, if Network Solutions is this universally despised today, imagine how much we all liked them years ago when they were the only game in town.
Let's see: $61 million - (100 employes * $60000 [1]) - DNS and web infrastructure of $1,000,000 [2] - Super Bowl ad of $2,000,000 = $52 million in profit.
I'd have to be of by orders of magnitude in a few areas for that not to be a sure-fire profit center.
[1] I doubt it takes 100 professional-level employees to run something like that, but you never know.
[2] Remember, you're only hosting the "domain parking" pages, and DNS for people with sites so small that the marginal costs per site are almost zero. The big customers (if any) pay extra for full-fledged services.
The thing is, though, that I liked the ad. As long as people who agree with me outnumber the people who share your opinion, then they're on the right track.
I'm glad that you're working on it, just as the rest of us are (and have been), but I think that "making maintenance easier" is a far cry from obsoleting an entire class of workers.
I generally like your posts, but this one was kind of dumb. Look, we've been hearing this promise for about 30 years now, and I don't think it's any more true today than it was then. The fact is that companies staff all of their mission-critical business functions and probably always will.
Examples? My company is not a shipper, but we have a full-time employee that handles shipping arrangements, puts incoming parcels where the belong, and has outgoing boxes ready when FedEx gets here. We're also not a staffing company, but we have an HR person. Neither are we a construction company, but we have a maintenance guy who also remodels our building as needed. Finally, we're not an IT consultant, but we have IT people on staff.
IT people will go away whenever companies no longer use IT. Until then, every place that depends on their services for daily operation will have employees that run them, just as they also have shipping, HR, and maintenance workers. I like your company (and would like them even more if you sent some free stuff my way, hint-hint), but you've done an excellent job of advancing the state of the art of the computers on the average employee's desk. That's just the tip of the iceburg for a lot of us, and no amount of CUPS-style printer autoconfiguration will change it.
In certain technical forums, those are referred to as "one" and "zero".
$40,000 ain't so hot if you're $500,000 in debt. Not that Debian is, mind you, but numbers without context are just pixels on a monitor.
It seems that proprietary software meant to control a piece of hardware that a company is selling as a stand-alone "black box" unit would be OK by RMS. In other words, in the cases I mentioned, the purchaser is really buying an MRI unit or an ABS system - the fact that it comes with software is mostly beside the point.
...and welcome to 1998. It wasn't true then, either.