My girlfriend is currently in medical school, and mentioned having been told that surgeons who play video games, especially just prior to surgery, actually make fewer mistakes during operations.
Wired had a story which provides some information about it.
I'm convinced that Microsoft's ultimate goal is to release just often enough that customers with Enterprise agreements are reminded that they aren't just paying for support, they're getting the software with it.
The price point is high enough that support alone isn't necessarily worth the annual cost. Only Microsoft can support the software (this isn't strictly true, but only they have the ability to publish patches, so it's effectively true). Companies are willing to pay an annual fee for that support. They're willing to pay much more for it if they get software licenses "for free" with the contract, but they need occasional new releases to feel like they're getting the full benefit of the agreement (as opposed to just paying for support, and getting software à la carte.
The home user isn't the market, since most don't purchase MS software directly and don't have much choice as to whether they receive it. It's businesses that matter, and upgrades are a way to ensure continued support and maintenance revenue.
...but theft prevention is a separate issue from what is addressed by the article. Yes, theft prevention exists, but the article's focus is on the use of RFID within the warehouse.
I heard about this while talking with a guy who deals with warehousing systems for a local supermarket chain.
It is a misconception that this is for use within the retail stores. In reality this is for use within the warehouses that supply the retail stores. I blame the reporter for making the assumption, and to a lesser extent the summary for running with the bait.
RFID is still too expensive to be placed within each individual package of Ramen noodles. It won't replace bar codes on the packages bought by consumers, but it is already replacing bar codes within the distribution centers.
In other words, each crate of Doritos will have an RFID chip that identifies the product. This is useful within the warehouse, as the warehouse deals with crates of product, not with individual packages of Charmin. You'll still see bar codes on products you buy.
This would only be true if it's actually unlawful to prescribe medication across state lines. I don't believe that's true, but it may be. The article mentions a charge of "distributing controlled substances and introducing misbranded drugs into interstate commerce". It's unclear as to whether "distributing" involves prescription, especially as it fails to list charges filed against each defendant individually. It seems reasonable to interpret this as referring to the actual distribution. The prescription probably falls under the listed charge of "conspiracy to dispense controlled substances".
I'm not clear as to whether it is unlawful to write prescriptions outside of one's home state. I can't find anything conclusive either way.
At any rate, the sentence as written simply didn't parse properly, and the post to which I replied pretended to provide clarification but failed. Your suggested interpretation is valid, but isn't quite supported by the article in question.
That's not two things. It's not criminal for a doctor not to evaluate someone. The average doctor will not have time to evaluate even a significant portion of the population before retiring.
It's one action: providing prescriptions. The phrase you've labelled as #2 simply explains the circumstances which made the action unlawful. You [hopefully] wouldn't say there are two actions involved in "I ate a sandwich at Arby's", despite similar structure.
Apparently there was some use to going over sentence structure in high school after all....
Their Enterprise contracts are probably a bit much for your friend's needs, but they may offer single-incident support for optimization and tuning assistance.
If he doesn't mind delving into DBA-land, he may want to buy a book. If he values the time it would take him to get up to speed and would rather spend it on other pursuits, it may well be worth the money to get some help.
Either way, he'll have to spend something (time or money) -- it's a question of how much his time is worth to him.
The problem is that language is a tool for expression. To me, a lack of concern regarding the clarity of your communication shows two things:
You don't consider your opinion to be worth the effort required to express it clearly.
You don't care to show respect for those reading your opinions. By placing more burden upon those interpreting your statements you indicate that you aren't as concerned with their ability to easily understand your statements as you are with simply feeling like you're contributing in some way.
If you care about your opinion and whether readers can understand it well, it's reasonable to expect that you will make some effort to express it clearly. For instance, confusing "it's" with "its" actually does make it harder for skilled readers to focus on your content because of the backwards lookup required when they suddenly realize that "it is" doesn't make sense in context. By forcing them to back up and reinterpret you redirect the attention from the content to the presentation.
Listening to somebody speak is a different case -- when you hear "its" or "it's" you automatically recognize that you'll have to do additional contextual resolution in order to interpret meaning. However, when I see the word "it's" I automatically resolve the meaning as the same concept expressed by "it is". When you have to stop and interpret each word phonetically an extra layer is introduced -- word->meaning becomes word->audio->meaning -- and those who have enough reading skill to avoid this extra step when reading well-written text will take performance hits when reading poorly-written text.
This isn't to say that you should be anal about every expression you write. However, if you respect your opinion and your audience it is worth the time to consider their end of the transaction -- your statements are only worth the reader's ability to properly interpret what you've expressed.
Katamari Damacy kicks some ass. It's a seemingly innovative game with the option to play short rounds (a few minutes) up to much longer ones (30 minutes, although after meeting certain criteria some areas have a timeless option).
There are no complex controls to remember, and replay value has been there so far. It's not addictive, but remains genuinely fun. There's also the additional factor that females seem to be just as likely to love it, so if you're looking for a relatively cheap game with replay value that's not just another rehash of some 15-year-old game it's definitely worth a look.
The only problem is that it can be hard to find a copy, but since it's been out for a little while that's getting better. Some EB stores stock a copy, as does Best Buy.
If it's half as large as you would indicate, your company already pays millions upon millions for software annually. You'd be amazed by what companies pay for Oracle, PeopleSoft, Windows, Office, Exchange, etc.
Hell, the per-user list price for an Exchange Client Access License (CAL) is $67 a seat. Assuming you have 10,000 employees, each of whom has a PC. That's $670,000 annually for email, not including the base price of the Exchange server. Note that each Exchange CAL gives you an Outlook CAL "for free", which pushes client usage towards Outlook on Windows.
Sure, you can buy through resellers and get volume discounts, but we're talking hundreds of thousands of dollars per year for freaking email.
I don't think your group is very familiar with enterprise software licensing schemes, or you'd have mentioned the number of users you expect to access the software in companies that might consider buying it. If this software package is as good as you claim, it's certainly possible to imagine a company tossing out a few hundred thousand for a license. You can make up numbers based on your experience as a coder, but until you do a little research into enterprise software licensing and costs you're doing your company a disservice by throwing out numbers so casually.
Hell, your company invested $2 million into development. Obviously they made an economic decision, and decided to go ahead with in-house development. If your company paid $2 million for the app, it has to be worth something to other businesses.
Projects may go over initial estimates -- what was originally budgeted for development of the application? That's the price your company was willing to pay for it. I'm sure they budgeted more than $10k. Find out this initial estimate, and use that as an estimate of what you'd expect to sell it for, and base your estimate on something real, not some number you pulled out of your ass.
It's been pointed out by others that many people simply have trouble saving money.
I used to work in a bank, and you'd see the same people come in every two weeks and talking about how they were sooo glad to be getting paid, then in the next breath mention that "I hope I have time to go next door and play my numbers." These are people who generally spend every damned cent they have regardless of the amount on the check.
I'm pretty bad at saving money myself. It would make more sense to invest in something long-term, sure. However, many people find this form of saving so convenient because there's no opt-out mechanism, and it's easier to pay extra and get back than have to risk underpaying and have to pull it out of their asses. I used to hold paper paychecks until I was in need of more money -- if it's not in my account, I can't spend it quite so easily.
As far as investments go, you'll not get enough interest to make interest your sole reason to do something like this. The most direct analog to paying in every pay period and having to wait before being able to get money back out would be a step-up CD. However, this is more work to set up than simply writing something on your tax forms when getting hired, and with CD rates so pathetically low it's simply not worth the hassle for most people, especially those with lower incomes.
This isn't a government program to help us manage money -- it's a way of getting an approximately correct estimate of what will be owed later on.
If you want to bitch about stupid investments, keeping your money in a savings account usually won't come close to inflation, and using it as a substitute for a legitimate long-term investment is still a waste. By that logic, why not just spend it all immediately, when your buying power is greatest? However, some things are important to people -- with taxes, the idea of having to do very little in order to get an "unexpected" windfall annually is a nice option, especially for those in the habit of spending every penny they earn immediately no matter how much it is.
Yes, it's a personal problem, but perhaps someone could be teaching this stuff and providing examples. Ideally the parents would do it. Realistically there are many children whose parents make awful financial decisions.
The problem also is that you can't necessarily manage multiple groups easily and make sure that machines are part of the appropriate groups. You could probably pull something together with tons of symlinks and a private repository per-machine, but rollback is still difficult, and you're still dealing with the problem of moving targets (for instance, a while ago a Debian stable upgrade broke SMTP auth because a configuration file had changed).
FreeBSD had a similar problem, where an update brought down an SSH configuration file change. Fortunately our automatic update script didn't restart services automatically, and I noticed the change before it caused trouble. Had services started automatically I'd have had to call customers all over the place and step them through making changes... not my idea of fun, and we only had a few sites to deal with at the time.
You're also losing the benefit of having a common target -- known versions of libraries, and known APIs and configuration file formats. Vendors like that, and I don't blame them.
One other RHN feature I'd forgotten to mention is that you can drop a text file on every machine in the group pretty easily -- good for things such as a new/etc/hosts or nsswitch.conf or similar. You can even schedule events (such as reboots, etc.) without having to log in and update crontab on every... damned... machine.
I'm looking at this right now at work. I've been a long-term Debian and occasional FreeBSD user, so I'm new to a lot of the considerations involved. If anyone can point out any misunderstandings or inaccuracies it would help readers and perhaps allow me to make more informed decisions as well.
We have a large number of desktops (numbered in four digits) across a large number of sites (numbered in three digits). When you get to this point, the idea of a free OS isn't quite the issue. Sure, it's nice that you can pop in the RHEL (Red Hat Enterprise Linux) or Fedora CDs and get a functional desktop. Automated hardware detection is good when anything other than a preconfigured, flawlessly automated installation simply isn't an option. This is something that Fedora has covered, as do many other free distributions.
However, where Fedora falls short is remote management -- obviously you can update a single machine easily, but imagine having several thousand identical systems to update in a secure and timely manner.
Add the fact that when there's nobody at an installation site capable of fixing problems should they arise you need a way to test new packages and updates prior to deployment. What is needed is a way to set up groups of machines in such a way that you can push software out to them easily in a controlled manner. Set up a testing lab, for instance, and test new software in the labs before risking a breakage in remote settings. Then, when you've tested, push updates out easily, and offer rollback support if something unexpected happens.
This is the state of what Red Hat Network allows you to do (at least with their highest level of support), all via a web-based tool. You can even delegate administrative access to subsets of machines to given administrators. They promise a stable distribution, even to the extent that APIs and ABIs of provided software will not change. This is something not promised under Fedora.
RHEL clones exist, but they do not allow access to the administrative features that the Red Hat Network provides. Nor are they supported by vendors -- if your SAN is supported under RHEL and doesn't work under RHEL, you can complain to the vendor until they fix it. If your SAN is supported under RHEL and doesn't work under CentOS, there's little you can do.
Windows allows system management like this through a product called Systems Management Server. Unsupported Linux distributions don't offer the features, and they don't offer this level of guaranteed API/ABI stability and vendor support.
Yes, you can do without support. However, when you're dealing with application vendors who expect to have a known set of installed software, it's much easier when everyone is on the same page. (This is a pain when it comes to JREs required by Windows apps -- everyone wants a different version, and if you don't have enough pull with the vendor to get them to support what you've got rolled out, you're stuck rolling out yet another JRE).
Cheap software means nothing if you can't tailor your environment to your needs, including deployment, testing, and administration on a wide scale. That's why companies pay for Red Hat support and subscriptions -- because of the management tools and guaranteed support you don't get with Fedora or CentOS.
I'm ignoring Novell here entirely, by the way, but they do offer similar features. They just can't seem to make up their minds whether to push open products or ones based on NetWare. Their interoperability is wicked, but they also like to push their consulting services so it's hard to find direct information much of the time.
Finally, Red Hat's desktop stance is quite unclear. Novell is pushing the hell out of Windows migrations, but they are short on details such as pricing and return on investment (except pushing products such as ZENworks, which is similar in function to the RHN stuff I've described). You can't get any information on Windows-to-Red Hat migrations either -- all of the Red Hat case studies are UNIX-to-
I tried to get back into it a few years ago and was disappointed to find out that none of the local Radio Shacks had any of the Engineer's Mini Notebooks. The folks at the counter hadn't heard of them, either.
Just two weeks ago a friend serendipitously returned something that I'd lent and forgotten about -- Mims' "Getting Started in Electronics", which the cover tells me was published exclusively for Radio Shack. It was a great introduction to electronics and covered enough theory to make it worth reading even if you didn't actually make anything functional.
It's definitely looking for a copy, or writing to Radio Shack to see if they know of any sources. It probably set me back about $10 at the time, and is a great introductory handbook.
I caught this post via MM2, and wanted to respond.
IE7 may depend on features provided by a particular OS, but that doesn't mean that it isn't a "real, independent web browser, safely seperated from the OS".
The Windows build of Firefox depends on Windows 98 or above. That doesn't mean that it's not separate from the OS. It means that it uses features not available in earlier versions of Windows. So it may be with IE7. It's entirely possible that XPSP2 provides functionality needed by IE7 (and thus is the minimum OS version required) as '98 provides functionality needed by Firefox.
"A requires B" does not necessarily mean "B also requires A". "IE7 needs XPSP2+" does not necessarily mean "XPSP2 needs IE7". It may indeed require IE7, but you can't assume that it follows logically from the initial statement without additional information.
To my knowledge, no one actually sells a (support) license or a retail box for any of these products. (Maybe OO.org... but definitely not FF or T-bird.) (emphasis mine)
You should really ask someone to update mozilla.org then -- they seem to think that "[t]elephone support at 1-888-586-4539 is available for Firefox 0.9 and above, Thunderbird 0.8 and above, and final Mozilla release versions 1.5, 1.6, and above. $39.95 per incident." When you mention this to the webmaster, mention that you saw the text above at http://www.mozilla.org/support/.
So, who is going to support these additional software packages? Who is going to be responsible for the software if/when it breaks, or if/when Firefox leads to a virus infecting the computer?
Who is responsible if/when Internet Explorer breaks, or if/when it leads to a virus infecting the computer? Last I checked, Microsoft won't "be responsible" either, in terms of cleaning up messes they allow to be created through bugs in their software.
Geeks often forget that a major part of the computer industry is support. [...] At least, when you have a licensed copy of MS office, and it goes whacko, you can call up MS and demand that they fix it. Can the same be said of the OSS that you listed?
Support is often a consideration with enterprise-level deployments. For such cases, there are third parties that are more than happy to offer support for OSS software. Sun themselves offers support contracts for OpenOffice.
For home users of Office, Microsoft offers two free phone support incidents, but only if your copy came from them. If it came preinstalled, you'll have to call your PC vendor (who will likely do very little to help) or pay Microsoft $35 per incident ($245 for "advanced issues"). I'd imagine that most people didn't buy a retail copy of Office, which means that there's no free support for the average user.
and there is no 1-800 number to call for help, either.
Microsoft offers an 800 number only for the fee-based support. If you bought a retail copy of Office (which is the example you gave), and are taking advantage of one of your two free incidents, you'll be calling (425) 635-7056.
Maybe before citing the myriad of support options available from OEMs at reasonable costs to home users, you might consider the real-world availability to those users. Bundled software is up to the PC manufacturer to support, and generally their support ends by suggesting use of the system restore CD. This means that bundled software is virtually unsupported, and the average user isn't likely to pay tons of money for a retail box -- they'll borrow from a friend long before, and anyways are so used to being pushed aside by ineffectual support departments that they expect very little in terms of vendor support anyways.
Different tools for different folks -- I'm rarely at the console of the machine I'm maintaining at the time, and I generally prefer command-line usage anyways.
Thanks for the reply though -- it's nice to know that there are usable graphical tools for those who prefer a GUI.
man editkeep and forget about it. Also see the -e or --exclude option -- apt-get --purge remove `deborphan -e libc6-i686,j2re-x` is still easier than the extra indirection of saving and editing a list.
You really should be using dpkg --purge instead of apt-get to remove programs that you're actually trying to remove.
The difference is that by default apt-get will remove files included in the package in a manner equivalent to dpkg --remove. However, as with dpkg --remove, it will not remove configuration files, and can thus leave some cruft behind on your system.
Debian kicks ass in part because you can keep a system clean for years without unneeded effort. Using "remove" where you really want to "purge" is one way to give up that advantage to a small degree.
See man apt-get, specifically the --purge and the APT::Get::Purge configuration option in apt.conf.
Or, just use dpkg --purge for removal, and stick to apt-get for installation and upgrades.
Slashdot Mirror
My girlfriend is currently in medical school, and mentioned having been told that surgeons who play video games, especially just prior to surgery, actually make fewer mistakes during operations.
Wired had a story which provides some information about it.
I'm convinced that Microsoft's ultimate goal is to release just often enough that customers with Enterprise agreements are reminded that they aren't just paying for support, they're getting the software with it.
The price point is high enough that support alone isn't necessarily worth the annual cost. Only Microsoft can support the software (this isn't strictly true, but only they have the ability to publish patches, so it's effectively true). Companies are willing to pay an annual fee for that support. They're willing to pay much more for it if they get software licenses "for free" with the contract, but they need occasional new releases to feel like they're getting the full benefit of the agreement (as opposed to just paying for support, and getting software à la carte.
The home user isn't the market, since most don't purchase MS software directly and don't have much choice as to whether they receive it. It's businesses that matter, and upgrades are a way to ensure continued support and maintenance revenue.
Just some thoughts.
...but theft prevention is a separate issue from what is addressed by the article. Yes, theft prevention exists, but the article's focus is on the use of RFID within the warehouse.
I heard about this while talking with a guy who deals with warehousing systems for a local supermarket chain.
It is a misconception that this is for use within the retail stores. In reality this is for use within the warehouses that supply the retail stores. I blame the reporter for making the assumption, and to a lesser extent the summary for running with the bait.
RFID is still too expensive to be placed within each individual package of Ramen noodles. It won't replace bar codes on the packages bought by consumers, but it is already replacing bar codes within the distribution centers.
In other words, each crate of Doritos will have an RFID chip that identifies the product. This is useful within the warehouse, as the warehouse deals with crates of product, not with individual packages of Charmin. You'll still see bar codes on products you buy.
This would only be true if it's actually unlawful to prescribe medication across state lines. I don't believe that's true, but it may be. The article mentions a charge of "distributing controlled substances and introducing misbranded drugs into interstate commerce". It's unclear as to whether "distributing" involves prescription, especially as it fails to list charges filed against each defendant individually. It seems reasonable to interpret this as referring to the actual distribution. The prescription probably falls under the listed charge of "conspiracy to dispense controlled substances".
I'm not clear as to whether it is unlawful to write prescriptions outside of one's home state. I can't find anything conclusive either way.
At any rate, the sentence as written simply didn't parse properly, and the post to which I replied pretended to provide clarification but failed. Your suggested interpretation is valid, but isn't quite supported by the article in question.
That's not two things. It's not criminal for a doctor not to evaluate someone. The average doctor will not have time to evaluate even a significant portion of the population before retiring.
It's one action: providing prescriptions. The phrase you've labelled as #2 simply explains the circumstances which made the action unlawful. You [hopefully] wouldn't say there are two actions involved in "I ate a sandwich at Arby's", despite similar structure.
Apparently there was some use to going over sentence structure in high school after all....
You might suggest that your friend consider asking MySQL for a quote:
http://www.mysql.com/company/contact/
Their Enterprise contracts are probably a bit much for your friend's needs, but they may offer single-incident support for optimization and tuning assistance.
If he doesn't mind delving into DBA-land, he may want to buy a book. If he values the time it would take him to get up to speed and would rather spend it on other pursuits, it may well be worth the money to get some help.
Either way, he'll have to spend something (time or money) -- it's a question of how much his time is worth to him.
Even worse, this isn't Google adding another feature, it's Taco going ape over a feature that's been in Google for months, if not longer.
What's next, a "Java implements multithreading" story?
If you care about your opinion and whether readers can understand it well, it's reasonable to expect that you will make some effort to express it clearly. For instance, confusing "it's" with "its" actually does make it harder for skilled readers to focus on your content because of the backwards lookup required when they suddenly realize that "it is" doesn't make sense in context. By forcing them to back up and reinterpret you redirect the attention from the content to the presentation.
Listening to somebody speak is a different case -- when you hear "its" or "it's" you automatically recognize that you'll have to do additional contextual resolution in order to interpret meaning. However, when I see the word "it's" I automatically resolve the meaning as the same concept expressed by "it is". When you have to stop and interpret each word phonetically an extra layer is introduced -- word->meaning becomes word->audio->meaning -- and those who have enough reading skill to avoid this extra step when reading well-written text will take performance hits when reading poorly-written text.
This isn't to say that you should be anal about every expression you write. However, if you respect your opinion and your audience it is worth the time to consider their end of the transaction -- your statements are only worth the reader's ability to properly interpret what you've expressed.
Katamari Damacy kicks some ass. It's a seemingly innovative game with the option to play short rounds (a few minutes) up to much longer ones (30 minutes, although after meeting certain criteria some areas have a timeless option).
There are no complex controls to remember, and replay value has been there so far. It's not addictive, but remains genuinely fun. There's also the additional factor that females seem to be just as likely to love it, so if you're looking for a relatively cheap game with replay value that's not just another rehash of some 15-year-old game it's definitely worth a look.
The only problem is that it can be hard to find a copy, but since it's been out for a little while that's getting better. Some EB stores stock a copy, as does Best Buy.
The first thing that springs to mind is that they're the only ones small enough to fit!
Well, I was amused.
I was out of town for a week -- belated thanks for the clarification.
If it's half as large as you would indicate, your company already pays millions upon millions for software annually. You'd be amazed by what companies pay for Oracle, PeopleSoft, Windows, Office, Exchange, etc.
Hell, the per-user list price for an Exchange Client Access License (CAL) is $67 a seat. Assuming you have 10,000 employees, each of whom has a PC. That's $670,000 annually for email, not including the base price of the Exchange server. Note that each Exchange CAL gives you an Outlook CAL "for free", which pushes client usage towards Outlook on Windows.
Sure, you can buy through resellers and get volume discounts, but we're talking hundreds of thousands of dollars per year for freaking email.
I don't think your group is very familiar with enterprise software licensing schemes, or you'd have mentioned the number of users you expect to access the software in companies that might consider buying it. If this software package is as good as you claim, it's certainly possible to imagine a company tossing out a few hundred thousand for a license. You can make up numbers based on your experience as a coder, but until you do a little research into enterprise software licensing and costs you're doing your company a disservice by throwing out numbers so casually.
Hell, your company invested $2 million into development. Obviously they made an economic decision, and decided to go ahead with in-house development. If your company paid $2 million for the app, it has to be worth something to other businesses.
Projects may go over initial estimates -- what was originally budgeted for development of the application? That's the price your company was willing to pay for it. I'm sure they budgeted more than $10k. Find out this initial estimate, and use that as an estimate of what you'd expect to sell it for, and base your estimate on something real, not some number you pulled out of your ass.
It's been pointed out by others that many people simply have trouble saving money.
I used to work in a bank, and you'd see the same people come in every two weeks and talking about how they were sooo glad to be getting paid, then in the next breath mention that "I hope I have time to go next door and play my numbers." These are people who generally spend every damned cent they have regardless of the amount on the check.
I'm pretty bad at saving money myself. It would make more sense to invest in something long-term, sure. However, many people find this form of saving so convenient because there's no opt-out mechanism, and it's easier to pay extra and get back than have to risk underpaying and have to pull it out of their asses. I used to hold paper paychecks until I was in need of more money -- if it's not in my account, I can't spend it quite so easily.
As far as investments go, you'll not get enough interest to make interest your sole reason to do something like this. The most direct analog to paying in every pay period and having to wait before being able to get money back out would be a step-up CD. However, this is more work to set up than simply writing something on your tax forms when getting hired, and with CD rates so pathetically low it's simply not worth the hassle for most people, especially those with lower incomes.
This isn't a government program to help us manage money -- it's a way of getting an approximately correct estimate of what will be owed later on.
If you want to bitch about stupid investments, keeping your money in a savings account usually won't come close to inflation, and using it as a substitute for a legitimate long-term investment is still a waste. By that logic, why not just spend it all immediately, when your buying power is greatest? However, some things are important to people -- with taxes, the idea of having to do very little in order to get an "unexpected" windfall annually is a nice option, especially for those in the habit of spending every penny they earn immediately no matter how much it is.
Yes, it's a personal problem, but perhaps someone could be teaching this stuff and providing examples. Ideally the parents would do it. Realistically there are many children whose parents make awful financial decisions.
The problem also is that you can't necessarily manage multiple groups easily and make sure that machines are part of the appropriate groups. You could probably pull something together with tons of symlinks and a private repository per-machine, but rollback is still difficult, and you're still dealing with the problem of moving targets (for instance, a while ago a Debian stable upgrade broke SMTP auth because a configuration file had changed).
/etc/hosts or nsswitch.conf or similar. You can even schedule events (such as reboots, etc.) without having to log in and update crontab on every... damned... machine.
FreeBSD had a similar problem, where an update brought down an SSH configuration file change. Fortunately our automatic update script didn't restart services automatically, and I noticed the change before it caused trouble. Had services started automatically I'd have had to call customers all over the place and step them through making changes... not my idea of fun, and we only had a few sites to deal with at the time.
You're also losing the benefit of having a common target -- known versions of libraries, and known APIs and configuration file formats. Vendors like that, and I don't blame them.
One other RHN feature I'd forgotten to mention is that you can drop a text file on every machine in the group pretty easily -- good for things such as a new
I'm looking at this right now at work. I've been a long-term Debian and occasional FreeBSD user, so I'm new to a lot of the considerations involved. If anyone can point out any misunderstandings or inaccuracies it would help readers and perhaps allow me to make more informed decisions as well.
We have a large number of desktops (numbered in four digits) across a large number of sites (numbered in three digits). When you get to this point, the idea of a free OS isn't quite the issue. Sure, it's nice that you can pop in the RHEL (Red Hat Enterprise Linux) or Fedora CDs and get a functional desktop. Automated hardware detection is good when anything other than a preconfigured, flawlessly automated installation simply isn't an option. This is something that Fedora has covered, as do many other free distributions.
However, where Fedora falls short is remote management -- obviously you can update a single machine easily, but imagine having several thousand identical systems to update in a secure and timely manner.
Add the fact that when there's nobody at an installation site capable of fixing problems should they arise you need a way to test new packages and updates prior to deployment. What is needed is a way to set up groups of machines in such a way that you can push software out to them easily in a controlled manner. Set up a testing lab, for instance, and test new software in the labs before risking a breakage in remote settings. Then, when you've tested, push updates out easily, and offer rollback support if something unexpected happens.
This is the state of what Red Hat Network allows you to do (at least with their highest level of support), all via a web-based tool. You can even delegate administrative access to subsets of machines to given administrators. They promise a stable distribution, even to the extent that APIs and ABIs of provided software will not change. This is something not promised under Fedora.
RHEL clones exist, but they do not allow access to the administrative features that the Red Hat Network provides. Nor are they supported by vendors -- if your SAN is supported under RHEL and doesn't work under RHEL, you can complain to the vendor until they fix it. If your SAN is supported under RHEL and doesn't work under CentOS, there's little you can do.
Windows allows system management like this through a product called Systems Management Server. Unsupported Linux distributions don't offer the features, and they don't offer this level of guaranteed API/ABI stability and vendor support.
Yes, you can do without support. However, when you're dealing with application vendors who expect to have a known set of installed software, it's much easier when everyone is on the same page. (This is a pain when it comes to JREs required by Windows apps -- everyone wants a different version, and if you don't have enough pull with the vendor to get them to support what you've got rolled out, you're stuck rolling out yet another JRE).
Cheap software means nothing if you can't tailor your environment to your needs, including deployment, testing, and administration on a wide scale. That's why companies pay for Red Hat support and subscriptions -- because of the management tools and guaranteed support you don't get with Fedora or CentOS.
I'm ignoring Novell here entirely, by the way, but they do offer similar features. They just can't seem to make up their minds whether to push open products or ones based on NetWare. Their interoperability is wicked, but they also like to push their consulting services so it's hard to find direct information much of the time.
Finally, Red Hat's desktop stance is quite unclear. Novell is pushing the hell out of Windows migrations, but they are short on details such as pricing and return on investment (except pushing products such as ZENworks, which is similar in function to the RHN stuff I've described). You can't get any information on Windows-to-Red Hat migrations either -- all of the Red Hat case studies are UNIX-to-
I tried to get back into it a few years ago and was disappointed to find out that none of the local Radio Shacks had any of the Engineer's Mini Notebooks. The folks at the counter hadn't heard of them, either.
Just two weeks ago a friend serendipitously returned something that I'd lent and forgotten about -- Mims' "Getting Started in Electronics", which the cover tells me was published exclusively for Radio Shack. It was a great introduction to electronics and covered enough theory to make it worth reading even if you didn't actually make anything functional.
It's definitely looking for a copy, or writing to Radio Shack to see if they know of any sources. It probably set me back about $10 at the time, and is a great introductory handbook.
Oops -- I linked to a mirror, instead of to the original. Perhaps not the worst mistake to make, but unintentional nonetheless.
This is why contributors to GNU software are expected to assign copyright to the FSF.
This issue has been addressed, and the FSF has shown one way to handle it properly. There's nothing to see here.
One of the best resources available...
I caught this post via MM2, and wanted to respond.
IE7 may depend on features provided by a particular OS, but that doesn't mean that it isn't a "real, independent web browser, safely seperated from the OS".
The Windows build of Firefox depends on Windows 98 or above. That doesn't mean that it's not separate from the OS. It means that it uses features not available in earlier versions of Windows. So it may be with IE7. It's entirely possible that XPSP2 provides functionality needed by IE7 (and thus is the minimum OS version required) as '98 provides functionality needed by Firefox.
"A requires B" does not necessarily mean "B also requires A". "IE7 needs XPSP2+" does not necessarily mean "XPSP2 needs IE7". It may indeed require IE7, but you can't assume that it follows logically from the initial statement without additional information.
You should really ask someone to update mozilla.org then -- they seem to think that "[t]elephone support at 1-888-586-4539 is available for Firefox 0.9 and above, Thunderbird 0.8 and above, and final Mozilla release versions 1.5, 1.6, and above. $39.95 per incident." When you mention this to the webmaster, mention that you saw the text above at http://www.mozilla.org/support/.
Who is responsible if/when Internet Explorer breaks, or if/when it leads to a virus infecting the computer? Last I checked, Microsoft won't "be responsible" either, in terms of cleaning up messes they allow to be created through bugs in their software.
Support is often a consideration with enterprise-level deployments. For such cases, there are third parties that are more than happy to offer support for OSS software. Sun themselves offers support contracts for OpenOffice.
For home users of Office, Microsoft offers two free phone support incidents, but only if your copy came from them. If it came preinstalled, you'll have to call your PC vendor (who will likely do very little to help) or pay Microsoft $35 per incident ($245 for "advanced issues"). I'd imagine that most people didn't buy a retail copy of Office, which means that there's no free support for the average user.
Microsoft offers an 800 number only for the fee-based support. If you bought a retail copy of Office (which is the example you gave), and are taking advantage of one of your two free incidents, you'll be calling (425) 635-7056.
Maybe before citing the myriad of support options available from OEMs at reasonable costs to home users, you might consider the real-world availability to those users. Bundled software is up to the PC manufacturer to support, and generally their support ends by suggesting use of the system restore CD. This means that bundled software is virtually unsupported, and the average user isn't likely to pay tons of money for a retail box -- they'll borrow from a friend long before, and anyways are so used to being pushed aside by ineffectual support departments that they expect very little in terms of vendor support anyways.
Different tools for different folks -- I'm rarely at the console of the machine I'm maintaining at the time, and I generally prefer command-line usage anyways.
Thanks for the reply though -- it's nice to know that there are usable graphical tools for those who prefer a GUI.
man editkeep and forget about it. Also see the -e or --exclude option -- apt-get --purge remove `deborphan -e libc6-i686,j2re-x` is still easier than the extra indirection of saving and editing a list.
You really should be using dpkg --purge instead of apt-get to remove programs that you're actually trying to remove.
The difference is that by default apt-get will remove files included in the package in a manner equivalent to dpkg --remove. However, as with dpkg --remove, it will not remove configuration files, and can thus leave some cruft behind on your system.
Debian kicks ass in part because you can keep a system clean for years without unneeded effort. Using "remove" where you really want to "purge" is one way to give up that advantage to a small degree.
See man apt-get, specifically the --purge and the APT::Get::Purge configuration option in apt.conf.
Or, just use dpkg --purge for removal, and stick to apt-get for installation and upgrades.
</pedantic>