You've been offered a job and you're worried about the pay? It's better to be worried about finding a job...
I have to disagree here. You shouldn't be worried about finding a job; you should be worried about finding a field in which you have sufficient ability.
If you have the necessary talent, finding a job in computing isn't a problem. I've turned down many job offers in the past six months, and declined even more invitations to interview (I want to finish my thesis). The jobs are out there.
If you don't have the necessary talent... go find a different field. Who knows, you might be a really good waiter. But computing has far too many poor coders already, and we certainly don't need any more.
The students have an excuse for lousy code. The professors don't.
You might like to take a look at the full titles of said professors: "Professor of Computer Science". Not "Professor of Computer Programming", and certainly not "Computer Programmer". The greatest architects of the world are often entirely ignorant of the latest techniques in plumbing, welding, and electric wiring.
So imagine that you're the manager of the UNIX dev group. What you see are a bunch of old graybeards with 20+ years of software experience (and salaries to match!), who are relying on a twentysomething not six months out of college to tell them how to make their code compile.
What would you as the manager think? Would you think "damn, that kid must be really hot!", or would you think "damn, experience in programming is really overrated!"?
No, I'd think "twentysomething programmers are cheap. Let's hire two more, for the sole purpose of cleaning up after the greybeards and making sure that their code compiles". Let the experts do what they're good at; hire a janitor to do the janitorial work.
i.e. the case where webwasher incorrectly denies access to your site to a large audience (i.e. its entire product base) and where you lose revenue.
Unfortunately, it doesn't look like this will happen. The situation is identical to when non-spammers have their email blocked by anti-spam blacklists; the people who provide those lists just say "we're not blocking anything, we're just providing a list". I'm sure webwasher will say the same thing: "Don't blame us, we're just providing a list."
(Oddly enough, this excuse is accepted in this context, while it was never accepted as a defense of those who compiled lists of Jews during the second world war; evidently compiling an accurate list for an illegal purpose is worse than compiling an inaccurate list for a good purpose.)
And what about the famous "magic number", 7 +/- 2? These people seem to be offering 4 +1/-2.
It was found that the famous "5-9 digits" resulted from a bogus test. Rather than testing short-term memory, it was testing the "auditory loop" -- people weren't remembering the digits, they were mentally replaying the sound of someone speaking the digits.
When people are given the digits via non-auditory means, 3-5 digits seems to be the norm.
So I was thinking about how us non-ubergeeks could contribute in a helpful way, and I think a separate section on bsdforums (say "PR Candidates") could be created, where we, users would test out some things to make 100% sure it is really a bug, it is reproducable, etc. before we submit it. Originator would be bsdforums, the threads could be used as reference, and thread participants would volunteer to test out the patches sent back.
That would be great; but much less than that would still be helpful. Having someone go through and identify PRs which 1. Report a reproducible bug, and 2. Contain a patch which fixes that bug would be useful just by itself.
To be perfectly honest, you've been lucky. Ports PRs tend to get resolved fairly quickly; src PRs often get lost in the shuffle. The big problem is that most PRs are poorly written, either lacking necessary information or lacking coherant English; as a result, most src committers won't take the time necessary to comb through the database in order to find the relatively few good PRs.
Prior to getting a commit bit of my own, I often had bug fixes sit in the PR database for months... the trick, as I learned, is to send in the PR with a patch, wait a couple weeks, and then start sending emails to committers.
Dealing with PRs is certainly a major issue which we'd like to improve upon, but in the end it's all a question of time and money; reading through PRs is rather dull work, and if we're not going to pay people (and there isn't any money available for this) then there simply isn't enough committer-time to do as well as we should.
And, as phk says, you'll get much more, than just some binary updates.
A good thing, too, considering how much I've gathered in donations from FreeBSD Update: $10 plus a new computer which is dedicated to the task of building those updates. (That's with over 3000 users.)
It may be where his budget balances, but if he expects to live off the kindness of strangers, he needs to adjust his budget substantially.
He's not *expecting* anything. This is an experiment: See if the FreeBSD community is willing to pay for someone to work full-time on FreeBSD. If not, well, he finds more contract work, earns the same amount (or more), and works on FreeBSD in his spare time.
RedHat charged people $60/year for access to binary updates (the company which has taken over supplying updates to old RedHat releases also charges the same rate). MandrakeClub costs at least $60/year, with a "Recommended level" of $120/year.
As phk wrote, "Imagine if some of our users sent $1/month for each FreeBSD machine they were running." There are a lot of people and companies running FreeBSD, and it wouldn't take much from each of them to pay for several people to work full-time on FreeBSD.
Why not correct for camera jitter by digitally re-aligning the frames afterwards? Ok, you'd lose the edges of your image, and parallax might be a problem, but I doubt these would be major issues in most movie sequences.
During Clinton's term in office, the unemployment rate dropped from 7.5% to 4.0%. During the first three years of Bush's term, it rose from 4.0% to 6.0%.
"Some Americans have yet to receive their forty acres and mules."
This is not recorded anywhere in any historical document. This is a legend that has been passed on over the years.
Not quite. During the civil war, General William Tecumseh Sherman issued an order to provide some blacks with 40 acres, and for the army to loan them mules. However, he had no authority to do this, so his order (and promises) were worthless.
http://www.snopes.com/business/taxes/blacktax.as p
When compared to Mozilla 1.6, Lynx is 99% faster at startup, 99% faster at window open time, has 50% faster pageloading times, and is 90% smaller in binary size.
In all seriousness, it's easy to improve figures like this just by removing features.
Why not overclock network cards as well as CPU and graphics cards?
Two reasons: First, you'd have to overclock every network card by exactly the same amount, or they'd not be able to communicate. Second, the limiting factor is the ability to get a signal through the cables; unless you're going to have LN2 jackets around all your cables, there's not much you can do about this.
In many cases, I don't think the incompatibilities were deliberate. The great thing about standards is that there are so many to choose from... Microsoft had to choose something, and it's not their fault that most people picked a different standard.
To take the sockets API as an example, the UNIX world had BSD sockets, while the Windows world had Winsock. Microsoft didn't invent Winsock -- it was put together by people who wanted TCP/IP on Windows back when Microsoft didn't provide this. When Microsoft added TCP/IP into Windows, they had a choice: They could use the "standard" BSD sockets, and break all the code people were already running on Windows, or they could build their version of the API people were already using.
In the case of threads, it's easy to ask why Microsoft doesn't use POSIX threads... except that the POSIX threads standard didn't exist until three years after Windows NT was first released. Oops.
As for graphics: If compatibility would mean supporting X windows, breaking compatibility is a Good Thing.
I'm not going to argue about that, but just point out that it is a very strong statement, and that most people would disagree with you.
Lots of people use Microsoft Windows.
This becomes even more obvious when you consider that replacing elementary particles is a no-op.
Nope. This is a logical fallacy called the Paradox of the heap.
I've always known it called the "'piece of string' fallacy", but terminology aside, this just illustrates my point: It is impossible to naively define a {heap, length of a piece of string, library, identity}, ergo those must be considered to be nothing beyond social constructs.
The win32 threading API is very nice. Seriously -- I'm a FreeBSD developer, and there isn't much about Windows which I like, but win32 threads are really well thought out and intuitive.
Slashdot Mirror
You've been offered a job and you're worried about the pay? It's better to be worried about finding a job...
I have to disagree here. You shouldn't be worried about finding a job; you should be worried about finding a field in which you have sufficient ability.
If you have the necessary talent, finding a job in computing isn't a problem. I've turned down many job offers in the past six months, and declined even more invitations to interview (I want to finish my thesis). The jobs are out there.
If you don't have the necessary talent... go find a different field. Who knows, you might be a really good waiter. But computing has far too many poor coders already, and we certainly don't need any more.
The students have an excuse for lousy code. The professors don't.
You might like to take a look at the full titles of said professors: "Professor of Computer Science". Not "Professor of Computer Programming", and certainly not "Computer Programmer". The greatest architects of the world are often entirely ignorant of the latest techniques in plumbing, welding, and electric wiring.
So imagine that you're the manager of the UNIX dev group. What you see are a bunch of old graybeards with 20+ years of software experience (and salaries to match!), who are relying on a twentysomething not six months out of college to tell them how to make their code compile.
What would you as the manager think? Would you think "damn, that kid must be really hot!", or would you think "damn, experience in programming is really overrated!"?
No, I'd think "twentysomething programmers are cheap. Let's hire two more, for the sole purpose of cleaning up after the greybeards and making sure that their code compiles". Let the experts do what they're good at; hire a janitor to do the janitorial work.
i.e. the case where webwasher incorrectly denies access to your site to a large audience (i.e. its entire product base) and where you lose revenue.
Unfortunately, it doesn't look like this will happen. The situation is identical to when non-spammers have their email blocked by anti-spam blacklists; the people who provide those lists just say "we're not blocking anything, we're just providing a list". I'm sure webwasher will say the same thing: "Don't blame us, we're just providing a list."
(Oddly enough, this excuse is accepted in this context, while it was never accepted as a defense of those who compiled lists of Jews during the second world war; evidently compiling an accurate list for an illegal purpose is worse than compiling an inaccurate list for a good purpose.)
And what about the famous "magic number", 7 +/- 2? These people seem to be offering 4 +1/-2.
It was found that the famous "5-9 digits" resulted from a bogus test. Rather than testing short-term memory, it was testing the "auditory loop" -- people weren't remembering the digits, they were mentally replaying the sound of someone speaking the digits.
When people are given the digits via non-auditory means, 3-5 digits seems to be the norm.
So I was thinking about how us non-ubergeeks could contribute in a helpful way, and I think a separate section on bsdforums (say "PR Candidates") could be created, where we, users would test out some things to make 100% sure it is really a bug, it is reproducable, etc. before we submit it. Originator would be bsdforums, the threads could be used as reference, and thread participants would volunteer to test out the patches sent back.
That would be great; but much less than that would still be helpful. Having someone go through and identify PRs which
1. Report a reproducible bug, and
2. Contain a patch which fixes that bug
would be useful just by itself.
The team is very responsive...
Yes, well... (*ahem*).
To be perfectly honest, you've been lucky. Ports PRs tend to get resolved fairly quickly; src PRs often get lost in the shuffle. The big problem is that most PRs are poorly written, either lacking necessary information or lacking coherant English; as a result, most src committers won't take the time necessary to comb through the database in order to find the relatively few good PRs.
Prior to getting a commit bit of my own, I often had bug fixes sit in the PR database for months... the trick, as I learned, is to send in the PR with a patch, wait a couple weeks, and then start sending emails to committers.
Dealing with PRs is certainly a major issue which we'd like to improve upon, but in the end it's all a question of time and money; reading through PRs is rather dull work, and if we're not going to pay people (and there isn't any money available for this) then there simply isn't enough committer-time to do as well as we should.
And, as phk says, you'll get much more, than just some binary updates.
A good thing, too, considering how much I've gathered in donations from FreeBSD Update: $10 plus a new computer which is dedicated to the task of building those updates. (That's with over 3000 users.)
USD$5500/month? That's more than my net take home
He still has to pay taxes, you know...
It may be where his budget balances, but if he expects to live off the kindness of strangers, he needs to adjust his budget substantially.
He's not *expecting* anything. This is an experiment: See if the FreeBSD community is willing to pay for someone to work full-time on FreeBSD. If not, well, he finds more contract work, earns the same amount (or more), and works on FreeBSD in his spare time.
RedHat charged people $60/year for access to binary updates (the company which has taken over supplying updates to old RedHat releases also charges the same rate). MandrakeClub costs at least $60/year, with a "Recommended level" of $120/year.
As phk wrote, "Imagine if some of our users sent $1/month for each FreeBSD machine they were running." There are a lot of people and companies running FreeBSD, and it wouldn't take much from each of them to pay for several people to work full-time on FreeBSD.
Why not correct for camera jitter by digitally re-aligning the frames afterwards? Ok, you'd lose the edges of your image, and parallax might be a problem, but I doubt these would be major issues in most movie sequences.
For the people who like useless links: You are here.
If you can't perform a CRC in your head, you shouldn't be surfing the 'Net. :)
TCP/IP doesn't use a CRC.
I was looking at yearly statistics -- monthly statistics are too volatile to be useful.
5.6% unemployment: Low for Clinton, High for Bush.
The Bureau of Labour Statistics doesn't agree with you.
During Clinton's term in office, the unemployment rate dropped from 7.5% to 4.0%. During the first three years of Bush's term, it rose from 4.0% to 6.0%.
Not quite. During the civil war, General William Tecumseh Sherman issued an order to provide some blacks with 40 acres, and for the army to loan them mules. However, he had no authority to do this, so his order (and promises) were worthless.
http://www.snopes.com/business/taxes/blacktax.a
Don't you mean liberate mars?
Not all the world is the United States. In Oxford, for example, I know of several buildings with 999 year leases.
(Of course, I'm writing this from inside a room which predates the USA by a couple centuries, so...)
When compared to Mozilla 1.6, Lynx is 99% faster at startup, 99% faster at window open time, has 50% faster pageloading times, and is 90% smaller in binary size.
In all seriousness, it's easy to improve figures like this just by removing features.
I mean, there's only so far Hyperthreading will take you...
And, by extension, there's only so far that multiple processors can take you.
Why not overclock network cards as well as CPU and graphics cards?
Two reasons: First, you'd have to overclock every network card by exactly the same amount, or they'd not be able to communicate. Second, the limiting factor is the ability to get a signal through the cables; unless you're going to have LN2 jackets around all your cables, there's not much you can do about this.
In many cases, I don't think the incompatibilities were deliberate. The great thing about standards is that there are so many to choose from... Microsoft had to choose something, and it's not their fault that most people picked a different standard.
To take the sockets API as an example, the UNIX world had BSD sockets, while the Windows world had Winsock. Microsoft didn't invent Winsock -- it was put together by people who wanted TCP/IP on Windows back when Microsoft didn't provide this. When Microsoft added TCP/IP into Windows, they had a choice: They could use the "standard" BSD sockets, and break all the code people were already running on Windows, or they could build their version of the API people were already using.
In the case of threads, it's easy to ask why Microsoft doesn't use POSIX threads... except that the POSIX threads standard didn't exist until three years after Windows NT was first released. Oops.
As for graphics: If compatibility would mean supporting X windows, breaking compatibility is a Good Thing.
Lots of people use Microsoft Windows.
Nope. This is a logical fallacy called the Paradox of the heap.
I've always known it called the "'piece of string' fallacy", but terminology aside, this just illustrates my point: It is impossible to naively define a {heap, length of a piece of string, library, identity}, ergo those must be considered to be nothing beyond social constructs.
When every component of the [baby] has been replaced at least once, is it still the same [baby]?
Yes. The law says that it is the same ship (err, baby); laws are a codification of social norms; and identity is nothing other than a social norm.
This becomes even more obvious when you consider that replacing elementary particles is a no-op.
Try writing a multi-threaded app on Windows.
The win32 threading API is very nice. Seriously -- I'm a FreeBSD developer, and there isn't much about Windows which I like, but win32 threads are really well thought out and intuitive.
CVSup; make buildworld && make installworld
For people who've never done this before (such as myself), this is an intimidating operation; care to walk me through it?
If you're intimidated by buildworld, there's an easier option:
# freebsd-update fetch
# freebsd-update install