eh? Phantom references just pass the collection along a level - if you store a db connection in an object, but place that object in another as a phantom reference simply ensures the db object gets collected after the enclosing object is collected.
Weak and phantom references are always collected if their referent is not marked. When a phantom reference is processed, its referent is marked so it will persist until the following garbage collection cycle or until the phantom reference is processed if it is associated with a reference queue. When it is determined that a reference is eligible for collection, it is either queued to its associated reference queue or removed from the reference list.
It persists until the following GC cycle.. sounds like one problem that finalisers are accused of.
Still, you need to execute code to clear the DB connection - GC of a phantom reference still does not process code when collected, you still need a finaliser to do that. If you never call the db close method (or whatever is required) the connection will remain in-use and effectively leaked.
which is why this guy is doing the right thing (hopefully)
ie, I'll code his app up for him, as long as he.. plans, describes, organises, documents, tests, QAs, packages, advertises, and generally does all the 'extra' stuff that I don't really want to do. I'm a programmer, I have to do all that stuff at work, I want to code for fun after hours.
I reckon that's a good split of effort, and if he's prepared to do all the additional stuff, then it will actually get done, unlike a lot of projects that have good code but a scrap of a website with some out-of-date documentation.
so, as a non-programmer he'll just have to show something better communicated than most projects. Show a good plan, mock-up, descriptoon etc and you'll get people who'll want to make it real.
Its like business, most projects are started by "businessmen" who usually have few skills other than planning and organising.
There's different kinds of anonynimity, there's the one where no-one truly knows who you are (eg you send an anonymous letter to a journalist), then there's the one where someone knows who you are but refuses to divulge that information (eg a journalist who has interviewed you and posts your story as an anonymous source).
Both provide you with the same anonymity, but the latter obviously carries more authority.
If the courts have decided that you have the right to anonymity online, then its surely ok for (say) your ISP to know who you are - they cannot reveal that information unless they get a court order allowing them to violate your legal right.
This latter form of anonyminity wouldn't apply to spammers, scammers, bullies and other malicious scum (ie the courts would grant a warrant everytime) and so might help to stop them and would make the internet an altogther better place to be.
Perhaps your skills get to the required 'junior' level after a few months of tinkering and learning, then you get better to become a 'full' sysadmin after a year or so - depending on how much you use the system day-to-day.
Two things are certain though - you'll never know everything as it continually changes so you may be an expert, but you'll still need to know more; and someone who knows nothing should not be in charge of a system. There's inexperienced and needing to learn, and then there's dangerous in charge of a live system.
oh, how I wish the OP would post a followup saying how he knew he was a good sysasmin because he'd passed all the MCSE exams.... I'm sure all the "don't give the guy a hard time" posts would become "prepare to be 0wn3d, n00b" instead:-)
Incidentally, the only real answer is the one usually given: use google. That's how I learn stuff nowadays.
I'd recommend the notebook approach, but I prefer to use a wiki. There's less chance of it being destroyed... because the first thing you learned was how to make backups wasn't it.
A Wiki is better because:
you can cut and paste commands into it without errors - including urls you can always read what you type into it you will never spill coffee over it age will never destroy it you will never lose it in the office moves you can share it with your colleagues it will always be there when you're doing things at your computer (assuming you work with LAMP) you can upload zips of config files, packages, etc
Whilst you could store passwords on it, I'd recommend against doing that:) a notebook (or keepass) is much better for them.
I recently took a new job as a network and server administration for a small IT company. I am absolutely shocked at how much is taking place within this company that I have little to no experience with
Its one thing to ask advice - you get it all the time on slashdot where people as "what's the best/how do I/what do you suggest", but this guy tells us straightaway, he has no experienbce int he job he's been hired to do. I don't know what jobs you go for, but I tend to be honest about my skills and not apply for jobs I have no chance of doing. The negative responses are surely picking up on this.
Havn't you been annoyed by colleagues who are incompetant in their roles?
At least this guy is admitting his weaknesses so he will get a lot of positive comments as well.0 Besides, some of the negative comments are just plain funny, stop being so negative about them:)
The only downside is that no two packages put everything in the same place on the local file system.
Amen to that. This is the biggest issue I have with Linux - the author's documentation says "edit the file in/usr/local/app/etc" but the distro decided to put it in/etc instead.
Now which one is right is another matter - ie put your apps in the right place, like all config files in/etc so no-one needs to hunt for them, or put them where the author decided they should go.
If you use yum/rpm then you can see where rpm put the files using "rpm -ql "
Interviewer: so, hw much do you know about server admin?
Poster: well, I freely admit I don't know everything, obviously nobody can know everything, but I know how to find the information for those things I'm not too familiar with.
Interviewer: good answer, you're hired.
Poster: excellent. You do, err, allow work-related internet access don't you....
'm willing to bet that if they make their own version it would be designed to be difficult to move to other systems
They have one already (sortof) - HP-UX. Perhaps they're thinking of making their own Linux version to make it easier to move to other systems, like all the x86_64 boxes they currently ship with Windows on. If they stop producing HP-UX and port a lot of the code they have in it to Linux, they get the best of all worlds - fancy stuff for their fancy servers, and fancy stuff for their mass-market servers and workstations.
They can also slap the Linux brand on it, so everyone becomes more comfortable running it.
I'd say he'd prefer a type of 'profile' system, where you change a tab to work under a different 'user' and you get the benefits of cookies and history and bookmarks, but they don't get mixed up with your other browsing.
It would not just be for porn, you could have a profile for porn, general surfing crap, work related surfing, etc. Possibly one for each interest you have. It could help manage ever growing bookmark lists. You could set privacy options for a profile to delete data at the end of the session, or keep it instead, and perhaps password-protct changing to a specific profile.
This kind of thing would be an improvement over the simple 'kill all data' sandbox system proposed.
so we'll probably end up showing you products that you probably aren't interested in.
You do that already. All your products are ones I'm not interested in, if I want something I'll go hunt it out and buy it.
Incidentally, you should be promoting the 'porn mode'. If suddenly all my 'off topic' browsing disappears from your advertising database, you have better quality statistics and significantly less noise. If I browse for work purposes, you don't want to be considering those sites to target advertising at me - I won't be interested, so splitting a browser into tab-based 'profiles' would be beneficial to you.
Sometimes it's hard to remember the contribution that people made so long ago. But we need to remember our history and respect those people who brought us here.
true. But it is our responsibility not to let things they did years ago to cloud our judgement of what they do today. You'd think Robert Mugabe was a hero otherwise, and ignore the politically-inspired violence of late.
Don't think I'm comparing Miguel with Mugabe, but the world would be better if he created similarly great advances in F/OSS software than just parroting Microsoft's latest moving target.
now that's a good idea - after all, the user interface is the computer to most people (ask my boss why I don't do prototypes anymore;) ). This would probably have the biggest impact to Linux takeup of all the code changes suggested so far.
so what configuration should you put in smb.conf? Cool, now email your reply to Canonical and see if someone will add it to the 'add smb networking' code.
I think Canonical has 'blueprints' for such additions (ie 'lightweight feature requests)
so it only takes an hour to copy a large directory over a WAN instead of an hour and ten minutes? BFD.
the most common operation is same-drive file copy, and Vista sucked so bad at it, it was the first fix they put into SP1. Its still not that good, Vista still pops up a dialog telling my how long it'd take to copy the files that takes longer than actually copying them.
the current 'real competition' for MS is open-source developers.
MS has always known that if they treat their devs right, more people will write software for Windows and so people will buy Windows because more software is written for it, businesses will use Windows because there are so many experienced Windows developers (roughly).
Now a recent Evans Data survey says the number of developers targetting Windows dropped from 74% in 2006 to 65% in 2007. This happened roughly the same time that MS started giving 'donations' to open source companies and becoming more open source 'friendly'. They know they can't stop devs from liking OSS, but they really, really want those devs to write their OSS code on the Windows platform. Hence all the tuning guides for PHP on Windows 2008 and marketing how PHP runs great on Windows. Apache is next in the sights and they're so desperate for new things that they're pushing C# as the one true language and platform. There are no Windows developers anymore, just.NET ones. See how they're throwing away all the old in favour of tempting in and locking up developers?
MS know that this year is difficult for them, web-technologies are popular, Google is hovering around attracting mindshare (at least) from MSs markets, and now developers are jumping ship for more interesting platforms. We could look back in a 5 years time and think of MS in the same way we think of IBM.
I'm not so sure. NT4 driver model was replaced in Win2000 and I didn't hear any complaints, grumblings or even a mention of it on/.
By all accounts the new model is simpler, and offers ore stability but I cannot see why so many 3rd party manufacturers didn't want (or couldn't) modify their existing drivers in time for Vista.
I suppose the problem then is that MS wants to have its cake and eat it too. It sells Windows to OEMs and then they sell it to you (or is it they licence it to OEMs, or are they licencing it to you indirectly via the OEM... its confusing) but they still want the MS branding on it. Dell cannot take Microsoft Windows and rebrand it as Dell.
So its MS Windows you get with your OEM. As a result MS could insist that the OEM supply an install disc, or could supply them itself, but the chance to save a few cents means more than the user benefits.
MS got itself stuck in a hard place, before the antitrust they did insist which icons were installed and prominent - IE and WMP. If they'd asked nicely instead of forcing them on through their market dominance, they might be able to affect that at least the worst crap doesn't get installed by default.
You can catch exceptions in c++ threads you know, and if you catch SEH exceptions you're catching the same exceptions as.NET apps do. So, its the platform that matters to get such stability, not the programming environment.
IPC is probably the biggest issue with threads v processes. I recall a quick rule of thumb that says if you get 1 million calls per second with in-process calls, cross-process runs at only 1000 calls per second, and RPC/network calls at 100. The cost of moving outside the app is significant. That was a few years back so things may not be quite so bad now with hypertransport buses, and I'm sure it didn't take into account shared memory - but then, shared memory is still memory in both processes and so not as reliable as a named pipe or socket.
You'll also need a cross-process synchronisation object, and they only exist in the kernel so you have the cost of locking that is very high compared to an in-process only lock.
Slashdot Mirror
hehehe, I've an idea.. get access to her Windows system, unlock the taskbar and drag it to the top of the screen.
eh? Phantom references just pass the collection along a level - if you store a db connection in an object, but place that object in another as a phantom reference simply ensures the db object gets collected after the enclosing object is collected.
Weak and phantom references are always collected if their referent is not marked. When a phantom reference is processed, its referent is marked so it will persist until the following garbage collection cycle or until the phantom reference is processed if it is associated with a reference queue. When it is determined that a reference is eligible for collection, it is either queued to its associated reference queue or removed from the reference list.
It persists until the following GC cycle.. sounds like one problem that finalisers are accused of.
Still, you need to execute code to clear the DB connection - GC of a phantom reference still does not process code when collected, you still need a finaliser to do that. If you never call the db close method (or whatever is required) the connection will remain in-use and effectively leaked.
which is why this guy is doing the right thing (hopefully)
ie, I'll code his app up for him, as long as he .. plans, describes, organises, documents, tests, QAs, packages, advertises, and generally does all the 'extra' stuff that I don't really want to do. I'm a programmer, I have to do all that stuff at work, I want to code for fun after hours.
I reckon that's a good split of effort, and if he's prepared to do all the additional stuff, then it will actually get done, unlike a lot of projects that have good code but a scrap of a website with some out-of-date documentation.
so, as a non-programmer he'll just have to show something better communicated than most projects. Show a good plan, mock-up, descriptoon etc and you'll get people who'll want to make it real.
Its like business, most projects are started by "businessmen" who usually have few skills other than planning and organising.
then its ok if they can find out who you are.
There's different kinds of anonynimity, there's the one where no-one truly knows who you are (eg you send an anonymous letter to a journalist), then there's the one where someone knows who you are but refuses to divulge that information (eg a journalist who has interviewed you and posts your story as an anonymous source).
Both provide you with the same anonymity, but the latter obviously carries more authority.
If the courts have decided that you have the right to anonymity online, then its surely ok for (say) your ISP to know who you are - they cannot reveal that information unless they get a court order allowing them to violate your legal right.
This latter form of anonyminity wouldn't apply to spammers, scammers, bullies and other malicious scum (ie the courts would grant a warrant everytime) and so might help to stop them and would make the internet an altogther better place to be.
Perhaps your skills get to the required 'junior' level after a few months of tinkering and learning, then you get better to become a 'full' sysadmin after a year or so - depending on how much you use the system day-to-day.
Two things are certain though - you'll never know everything as it continually changes so you may be an expert, but you'll still need to know more; and someone who knows nothing should not be in charge of a system. There's inexperienced and needing to learn, and then there's dangerous in charge of a live system.
Now where's a car analogy when you need one :)
I would like to see you back up your implication that those two skill sets are mutually exclusive.
you're reading slashdot and expect to see a better example of the exclusivity of those 2 skill sets?!
oh, how I wish the OP would post a followup saying how he knew he was a good sysasmin because he'd passed all the MCSE exams.... I'm sure all the "don't give the guy a hard time" posts would become "prepare to be 0wn3d, n00b" instead :-)
Incidentally, the only real answer is the one usually given: use google. That's how I learn stuff nowadays.
I'd recommend the notebook approach, but I prefer to use a wiki. There's less chance of it being destroyed ... because the first thing you learned was how to make backups wasn't it.
A Wiki is better because:
you can cut and paste commands into it without errors - including urls
you can always read what you type into it
you will never spill coffee over it
age will never destroy it
you will never lose it in the office moves
you can share it with your colleagues
it will always be there when you're doing things at your computer (assuming you work with LAMP)
you can upload zips of config files, packages, etc
Whilst you could store passwords on it, I'd recommend against doing that :) a notebook (or keepass) is much better for them.
I recently took a new job as a network and server administration for a small IT company. I am absolutely shocked at how much is taking place within this company that I have little to no experience with
Its one thing to ask advice - you get it all the time on slashdot where people as "what's the best/how do I/what do you suggest", but this guy tells us straightaway, he has no experienbce int he job he's been hired to do. I don't know what jobs you go for, but I tend to be honest about my skills and not apply for jobs I have no chance of doing. The negative responses are surely picking up on this.
Havn't you been annoyed by colleagues who are incompetant in their roles?
At least this guy is admitting his weaknesses so he will get a lot of positive comments as well.0 Besides, some of the negative comments are just plain funny, stop being so negative about them :)
The only downside is that no two packages put everything in the same place on the local file system.
Amen to that. This is the biggest issue I have with Linux - the author's documentation says "edit the file in /usr/local/app/etc" but the distro decided to put it in /etc instead.
Now which one is right is another matter - ie put your apps in the right place, like all config files in /etc so no-one needs to hunt for them, or put them where the author decided they should go.
If you use yum/rpm then you can see where rpm put the files using "rpm -ql "
Interviewer: so, hw much do you know about server admin?
Poster: well, I freely admit I don't know everything, obviously nobody can know everything, but I know how to find the information for those things I'm not too familiar with.
Interviewer: good answer, you're hired.
Poster: excellent. You do, err, allow work-related internet access don't you....
'm willing to bet that if they make their own version it would be designed to be difficult to move to other systems
They have one already (sortof) - HP-UX. Perhaps they're thinking of making their own Linux version to make it easier to move to other systems, like all the x86_64 boxes they currently ship with Windows on. If they stop producing HP-UX and port a lot of the code they have in it to Linux, they get the best of all worlds - fancy stuff for their fancy servers, and fancy stuff for their mass-market servers and workstations.
They can also slap the Linux brand on it, so everyone becomes more comfortable running it.
I'd say he'd prefer a type of 'profile' system, where you change a tab to work under a different 'user' and you get the benefits of cookies and history and bookmarks, but they don't get mixed up with your other browsing.
It would not just be for porn, you could have a profile for porn, general surfing crap, work related surfing, etc. Possibly one for each interest you have. It could help manage ever growing bookmark lists. You could set privacy options for a profile to delete data at the end of the session, or keep it instead, and perhaps password-protct changing to a specific profile.
This kind of thing would be an improvement over the simple 'kill all data' sandbox system proposed.
so we'll probably end up showing you products that you probably aren't interested in.
You do that already. All your products are ones I'm not interested in, if I want something I'll go hunt it out and buy it.
Incidentally, you should be promoting the 'porn mode'. If suddenly all my 'off topic' browsing disappears from your advertising database, you have better quality statistics and significantly less noise. If I browse for work purposes, you don't want to be considering those sites to target advertising at me - I won't be interested, so splitting a browser into tab-based 'profiles' would be beneficial to you.
don't worry, Google still knows you did it even if you've tried hard to forget.
Sometimes it's hard to remember the contribution that people made so long ago. But we need to remember our history and respect those people who brought us here.
true. But it is our responsibility not to let things they did years ago to cloud our judgement of what they do today. You'd think Robert Mugabe was a hero otherwise, and ignore the politically-inspired violence of late.
Don't think I'm comparing Miguel with Mugabe, but the world would be better if he created similarly great advances in F/OSS software than just parroting Microsoft's latest moving target.
why coudn't they dump ubuntu and continue with kubuntu instead then? Wouldn't it save everyone a heap of work?
now that's a good idea - after all, the user interface is the computer to most people (ask my boss why I don't do prototypes anymore ;) ). This would probably have the biggest impact to Linux takeup of all the code changes suggested so far.
so what configuration should you put in smb.conf? Cool, now email your reply to Canonical and see if someone will add it to the 'add smb networking' code.
I think Canonical has 'blueprints' for such additions (ie 'lightweight feature requests)
so it only takes an hour to copy a large directory over a WAN instead of an hour and ten minutes? BFD.
the most common operation is same-drive file copy, and Vista sucked so bad at it, it was the first fix they put into SP1. Its still not that good, Vista still pops up a dialog telling my how long it'd take to copy the files that takes longer than actually copying them.
the current 'real competition' for MS is open-source developers.
MS has always known that if they treat their devs right, more people will write software for Windows and so people will buy Windows because more software is written for it, businesses will use Windows because there are so many experienced Windows developers (roughly).
Now a recent Evans Data survey says the number of developers targetting Windows dropped from 74% in 2006 to 65% in 2007. This happened roughly the same time that MS started giving 'donations' to open source companies and becoming more open source 'friendly'. They know they can't stop devs from liking OSS, but they really, really want those devs to write their OSS code on the Windows platform. Hence all the tuning guides for PHP on Windows 2008 and marketing how PHP runs great on Windows. Apache is next in the sights and they're so desperate for new things that they're pushing C# as the one true language and platform. There are no Windows developers anymore, just .NET ones. See how they're throwing away all the old in favour of tempting in and locking up developers?
MS know that this year is difficult for them, web-technologies are popular, Google is hovering around attracting mindshare (at least) from MSs markets, and now developers are jumping ship for more interesting platforms. We could look back in a 5 years time and think of MS in the same way we think of IBM.
I'm not so sure. NT4 driver model was replaced in Win2000 and I didn't hear any complaints, grumblings or even a mention of it on /.
By all accounts the new model is simpler, and offers ore stability but I cannot see why so many 3rd party manufacturers didn't want (or couldn't) modify their existing drivers in time for Vista.
I suppose the problem then is that MS wants to have its cake and eat it too. It sells Windows to OEMs and then they sell it to you (or is it they licence it to OEMs, or are they licencing it to you indirectly via the OEM... its confusing) but they still want the MS branding on it. Dell cannot take Microsoft Windows and rebrand it as Dell.
So its MS Windows you get with your OEM. As a result MS could insist that the OEM supply an install disc, or could supply them itself, but the chance to save a few cents means more than the user benefits.
MS got itself stuck in a hard place, before the antitrust they did insist which icons were installed and prominent - IE and WMP. If they'd asked nicely instead of forcing them on through their market dominance, they might be able to affect that at least the worst crap doesn't get installed by default.
You can catch exceptions in c++ threads you know, and if you catch SEH exceptions you're catching the same exceptions as .NET apps do. So, its the platform that matters to get such stability, not the programming environment.
IPC is probably the biggest issue with threads v processes. I recall a quick rule of thumb that says if you get 1 million calls per second with in-process calls, cross-process runs at only 1000 calls per second, and RPC/network calls at 100. The cost of moving outside the app is significant. That was a few years back so things may not be quite so bad now with hypertransport buses, and I'm sure it didn't take into account shared memory - but then, shared memory is still memory in both processes and so not as reliable as a named pipe or socket.
You'll also need a cross-process synchronisation object, and they only exist in the kernel so you have the cost of locking that is very high compared to an in-process only lock.