2. File formats are a more dangerous issue than the physical media. Text-based formats (including XML) is something someone else could write a reader for later. That's a lot easier than trying to find both the software and the machine it ran on. Virtualization helps -- it may be worthwhile to keep a VMWare image with your stuff so that you have a copy of the software required. Emulators exist for most any obsolete system I can think of, and I'm pretty sure you're going to find x86 emulators for span of your grandkid's lifetimes. If paranoid, use an open-source emulation platform instead of vmware, and include the source of the emulator.
3. Archival formats are another problem. I love zfs dumps for their current usability, but I know I can't trust something that system-specific to last terribly long.
So, here's a procedure for you. Note that this is for important data. Don't interpret this as a method for keeping your Hentai collection. Of course, each has their own priorities in this area...
1. Choose something you know will be around "for a while." Easily available, good track record, easy to use. Don't worry about extreme-long term. Step 3 addresses that. More important are the short-term technological concerns: Is it simple & reliable enough that you _will_ back up reliably with it?
2. Keep a Master Image of what you want to back up. For example, a root directory with a per-year, per-project, or per-customer set of subdirectories. When you add stuff to your backup, you're appending to the master image. As storage (capacity & demand) is lightly exponential, you can usually afford to store all your old backups with the new (long term), as the old stuff is relatively small.
Note that I'm *not* saying copy this stuff every time you back up. The master image is append-only. You'll probably want more than 1 copy of the whole thing, but that can be two stacks of DVDs, two boxes of tapes, two RAID arrays, whatever.
3. Plan to migrate this growing master image to new formats. The entire master image moves to the new format. When you migrate, all your backup data is in the new format, and your old-format backups can be stored as a hedge against your new media having surprising failures later.
4. Treat this stuff *seriously*. Don't skimp or cheap-out. Spend a month or two every couple of years setting this stuff up correctly, and make it something you can rely on. A little organization and automation once makes worthwhile. Cron et al for Unix, and a good commercial backup app for everyone else.
5. TEST YOUR RESTORES. TEST YOUR RESTORES. TEST YOUR RESTORES. TEST YOUR RESTORES. TEST YOUR RESTORES. TEST YOUR RESTORES. TEST YOUR RESTORES. TEST YOUR RESTORES. TEST YOUR RESTORES. TEST YOUR RESTORES. TEST YOUR RESTORES. TEST YOUR RESTORES. TEST YOUR RESTORES.
There's a lot of solaris-specific software out there. Linux users tend to forget there was a Unix community long before they showed up.
As for IPS, you know you can just roll back through ZFS, right? As for why they're using IPS, why not ask Ian Murdock? He's the founder of Debian and works for Sun and worked (he's been promoted) on OpenSol.
Sun's workstations have stagnated in 2008, I don't know why. Their Amd64 line was the best deal from a real vendor when they came out.
As for the rest... yeah, support matters to some people. It's nice being able to talk the one who wrote the code that's giving you problems, and then getting a patch from them that'll be in the next release.
Also, have you considered the Enter key on your keyboard? It's to the right of the apostrophe on many layouts, and it makes your text easier to read.
Ask anyone who's been ordered to do so by the court. You can volunteer to do community service, but that's the combination of two distinct concepts.
As for doing it as part of school, I had to: it was part of my curriculum. Honestly I was better off for it. I helped the organizers of a marathon, set up some computers for a poorer school, and tutored a few in some basic programming skills.
I was a TA for the unix course here, and the final exam was a shell script that validated XML files using only your basic tools: shell scripts, pipes, sed, awk, and grep.
Piping through ( subshells ) at times, as a giant pipeline. Well, that was my solution to it.
Here's one I've started using: Make a.shortcuts directory. In it, put symlinks to the various places you go on your box.
Then, pop this in your.bashrc:
function go () {
if [ -x ~/.shortcuts/$1 ]
then
pushd ~/.shortcuts/$1 >/dev/null
DIR=`/bin/pwd -P`
popd >/dev/null
pushd $DIR
else
echo "not found"
fi }
This way, you can put this on the cmd line go pubs/foo/bar
For a symlink named pubs, with subdirectory foo, subdir of that bar.
Use multiple symlinks for the same place, for misspellings and abbreviations.
Honestly, really big. Lots of groups (I'll avoid names to protect the guilty) still use fortran on desktops and don't know any better. They order a stock machine from Dell or whoever and pull up a DOS window to run their fortran compiler.
I'd actually say that the majority of the people out there that could be using bigger computers for this stuff don't. Honestly, half of CS & IT people don't use computers effectively...
And 25k is a lot cheaper than the 3 extra people you hired to help you do your work.
No, it takes a fairly intelligent person's views and makes them public. The idea is this: everyone figures out what they want the government to do/be. Then they find the closest mapping, out of 2 options. If your views are similar to this exemplar (e.g. Cerf), then his reasoning should be useful to you. If not, then not.
That's really the question, does the person have any financial ties to the situation? Not just selling short on Apple, but owning microsoft or dell shares would also profit.
Of course, there are other malicious, but legal, reasons to do this sort of stuff. Maybe just a rabid anti-apple finance-savvy windows fanboi? (Apologies for the huge noun string).
The fact that you talk about your degree like it's some golden ticket, instead of a set of skills that you've learned, is indicative. The best shops will hire people based on skill, because their hiring people know actual skill when they see it. Shit shops look at degrees as checklists, as they don't know what they're doing and hope you do.
You've gotta seriously consider the fact that you do suck. It's ok, it happens to all of us sometime. Just go back now and actually learn the material. Do the studying you didn't do then. Make up for past intellectual sins & all that.
Experience is dead-simple now, just contribute to open source projects.
Go in the interview and say "I made GCC generate better code in this, this, and this case." Or "this one I/O board made by this mad-genius physicist in a Hong Kong sanitarium now runs under Linux because of me. It has a 5% better sampling rate than anyone else."
Look, the simple fact is that a lot of people aren't sure you're worth hiring.
Why? We all know people in our degree program, who graduated, who we think are useless jackasses. I knew quite a few CpEs who barely passed digital logic or programming and just wanted to build PCs for gaming all day. Some people finish CpE programs and design silicon for the big boys, others, who sat next to them in the same courses at the same time, tell people how to operate the power switch to make Windows work again. Your resume says you're in the latter group. Not only do they think you're worthless, they hate you for making their degrees look bad.
The fact that you got the degree but haven't found a use for it is a huge red flag. Especially in something useful like CpE. Good God man.
Despite that, it's not some scapegoat chicken & the egg problem. It's a technical degree, and you're applying to technical jobs. People can test what you actually know. The interview should include technical questions. Thats your chance to prove you're not a jackass. I've never been to an interview for a technical job without them asking me to talk code (I'm CS). Perhaps you've already forgotten too much? Or never really knew? Time to go through your textbooks and learn up.
Additionally, listing the courses and grades you got in them would help a lot. Unless you really are in the second category I listed, in which case you should look for the dumbest fucking hiring manager you can, and try and weasel your way into a mediocre admin job and let him think it's his lucky fucking day for getting a "massively overqualified" CpE.
Let's face it: Linux has stagnated. It used to be the hot new kid around, bringing together all the new-world desktop technology with the old-school unix reliability, modularity, and maintainability. It did it by being somewhere in between the two. And it took off! People loved it over windows.
But, look at it now. When's the last time it did something *well*? Its standard was always Windows, which is a very low bar to aim for. It's full of sprawl and half-implemented ideas that you have to constantly hack at to make the system work. It's been 10 years and Linux is still a maintenance pain-in-the-ass.
Most people don't care: they're happy to run a tomcat or php stack on top of it. For them, it's really just a SATA and ethernet device driver, which then lets you use your favorite app server.
Containers/Zones, ZFS, dtrace, and SMF blow linux out of the water in every category the affect. What's linux done in the last 4 years? More Windows Ketchup? Thanks but no thanks. I'll take good documentation, manageability, and stable behavior any day. Sorry, but Linux is still too much of a hobby instead of an OS for my book.
1. Document more than is necessary. B/c your estimates of what's necessary will always obsolesce over enough time.
2. Test cases. Tests that make sure the code runs, tests that you can run later to see what broke what.
3. Don't be a savage. Use real identifier names (i,j,k are only ok in very specific cases), well indented code, version control, etc. Code Complete was suggested above, it's a good place to start. If your shop doesn't do version control, that's no excuse for you not to.
Buy minutes in bulk @ virgin mobile. The only problem is, unless you pump $20/mon into the acct each month, the minutes expire in a month.
But $20/mon isn't too bad for me, and I can just let those $20 additions accumulate until I can either bulk-buy another 5c/min.
Also, not worrying about a broken/lost phone is pretty nice. A decent pitcher of beer costs more than the damage I could do to the phone from drinking the pitcher!
May I propose looking into the prepaid phones? They are the perfect hacker phone:
1. Cheap for low-minute users. Save your money for RAM. 2. Little to no personal information. No credit check. You can tell them whatever you want, and they don't care enough to validate. Helps relieve paranoia. 3. No contract, no obligation. Just let the minutes expire and dump the phone. Helps relieve paranoia. 4. Easily switchable. Psycho ex keep calling? Get a new account in a few minutes. With the GSM devices, just swap SIMs and burn the old one. Helps relieve paranoia.
I end up paying about $20/mon to keep my acct active with enough minutes. I pay about a nickel a minute to talk. If I break the phone, I hit wal-mart and get a replacement for $10, which they activate onto my existing account.
And the phone's a basic flip-device. No bells, no whistles. Did I mention the phone was $10 new? $10.
Slashdot Mirror
Some thoughts on this issue:
1. I wouldn't trust technologies that haven't been used for a decade or two yet, so people know what makes them last & fail. See here:
http://findarticles.com/p/articles/mi_m0BRZ/is_8_23/ai_109665179/pg_1?tag=artBody;col1
2. File formats are a more dangerous issue than the physical media. Text-based formats (including XML) is something someone else could write a reader for later. That's a lot easier than trying to find both the software and the machine it ran on. Virtualization helps -- it may be worthwhile to keep a VMWare image with your stuff so that you have a copy of the software required. Emulators exist for most any obsolete system I can think of, and I'm pretty sure you're going to find x86 emulators for span of your grandkid's lifetimes. If paranoid, use an open-source emulation platform instead of vmware, and include the source of the emulator.
3. Archival formats are another problem. I love zfs dumps for their current usability, but I know I can't trust something that system-specific to last terribly long.
So, here's a procedure for you. Note that this is for important data. Don't interpret this as a method for keeping your Hentai collection. Of course, each has their own priorities in this area...
1. Choose something you know will be around "for a while." Easily available, good track record, easy to use. Don't worry about extreme-long term. Step 3 addresses that. More important are the short-term technological concerns: Is it simple & reliable enough that you _will_ back up reliably with it?
2. Keep a Master Image of what you want to back up. For example, a root directory with a per-year, per-project, or per-customer set of subdirectories. When you add stuff to your backup, you're appending to the master image. As storage (capacity & demand) is lightly exponential, you can usually afford to store all your old backups with the new (long term), as the old stuff is relatively small.
Note that I'm *not* saying copy this stuff every time you back up. The master image is append-only. You'll probably want more than 1 copy of the whole thing, but that can be two stacks of DVDs, two boxes of tapes, two RAID arrays, whatever.
3. Plan to migrate this growing master image to new formats. The entire master image moves to the new format. When you migrate, all your backup data is in the new format, and your old-format backups can be stored as a hedge against your new media having surprising failures later.
4. Treat this stuff *seriously*. Don't skimp or cheap-out. Spend a month or two every couple of years setting this stuff up correctly, and make it something you can rely on. A little organization and automation once makes worthwhile. Cron et al for Unix, and a good commercial backup app for everyone else.
5. TEST YOUR RESTORES. TEST YOUR RESTORES. TEST YOUR RESTORES. TEST YOUR RESTORES. TEST YOUR RESTORES. TEST YOUR RESTORES. TEST YOUR RESTORES. TEST YOUR RESTORES. TEST YOUR RESTORES. TEST YOUR RESTORES. TEST YOUR RESTORES. TEST YOUR RESTORES. TEST YOUR RESTORES.
No, but they want you to pay the data fee for all 2 years of your contract.
Destroying the iPhone would break your EULA.
There's a lot of solaris-specific software out there. Linux users tend to forget there was a Unix community long before they showed up.
As for IPS, you know you can just roll back through ZFS, right? As for why they're using IPS, why not ask Ian Murdock? He's the founder of Debian and works for Sun and worked (he's been promoted) on OpenSol.
Sun's workstations have stagnated in 2008, I don't know why. Their Amd64 line was the best deal from a real vendor when they came out.
As for the rest... yeah, support matters to some people. It's nice being able to talk the one who wrote the code that's giving you problems, and then getting a patch from them that'll be in the next release.
Also, have you considered the Enter key on your keyboard? It's to the right of the apostrophe on many layouts, and it makes your text easier to read.
How many politicians actually used the "series of tubes" analogy?
One, and he's a felon now.
Ugh, I just undid my mod points for this..
Community Service != Volunteering
Ask anyone who's been ordered to do so by the court. You can volunteer to do community service, but that's the combination of two distinct concepts.
As for doing it as part of school, I had to: it was part of my curriculum. Honestly I was better off for it. I helped the organizers of a marathon, set up some computers for a poorer school, and tutored a few in some basic programming skills.
I was a TA for the unix course here, and the final exam was a shell script that validated XML files using only your basic tools: shell scripts, pipes, sed, awk, and grep.
Piping through ( subshells ) at times, as a giant pipeline. Well, that was my solution to it.
A little time on Solaris, and you start to love pgrep, which searches process names and returns their pids.
One little function I love (in .bashrc):
# Search history
function hgrep () {
history | grep "$@"
}
Which quickly lets you regex-search your history in a single command.
Here's one I've started using: .shortcuts directory. In it, put symlinks to the various places you go on your box.
Make a
Then, pop this in your .bashrc:
function go () {
if [ -x ~/.shortcuts/$1 ]
then
pushd ~/.shortcuts/$1 >/dev/null
DIR=`/bin/pwd -P`
popd >/dev/null
pushd $DIR
else
echo "not found"
fi
}
This way, you can put this on the cmd line
go pubs/foo/bar
For a symlink named pubs, with subdirectory foo, subdir of that bar.
Use multiple symlinks for the same place, for misspellings and abbreviations.
Um... you can turn off 3G in the iPhone preferences.
-1 Offtopic.
Dude, nobody wants to actually *read* the policies, they just want to complain about them.
And, you're posting hyperlinks of detailed articles on /.
That's a double fail.
Honestly, really big. Lots of groups (I'll avoid names to protect the guilty) still use fortran on desktops and don't know any better. They order a stock machine from Dell or whoever and pull up a DOS window to run their fortran compiler.
I'd actually say that the majority of the people out there that could be using bigger computers for this stuff don't. Honestly, half of CS & IT people don't use computers effectively...
And 25k is a lot cheaper than the 3 extra people you hired to help you do your work.
No, it takes a fairly intelligent person's views and makes them public. The idea is this: everyone figures out what they want the government to do/be. Then they find the closest mapping, out of 2 options. If your views are similar to this exemplar (e.g. Cerf), then his reasoning should be useful to you. If not, then not.
That's really the question, does the person have any financial ties to the situation? Not just selling short on Apple, but owning microsoft or dell shares would also profit.
Of course, there are other malicious, but legal, reasons to do this sort of stuff. Maybe just a rabid anti-apple finance-savvy windows fanboi? (Apologies for the huge noun string).
I donno about that. Pump & dump schemes are pretty illegal.
Yeah... but have you seen how shitty the other stuff is?
Yeah, self-reply.
The fact that you talk about your degree like it's some golden ticket, instead of a set of skills that you've learned, is indicative. The best shops will hire people based on skill, because their hiring people know actual skill when they see it. Shit shops look at degrees as checklists, as they don't know what they're doing and hope you do.
You've gotta seriously consider the fact that you do suck. It's ok, it happens to all of us sometime. Just go back now and actually learn the material. Do the studying you didn't do then. Make up for past intellectual sins & all that.
Experience is dead-simple now, just contribute to open source projects.
Go in the interview and say "I made GCC generate better code in this, this, and this case." Or "this one I/O board made by this mad-genius physicist in a Hong Kong sanitarium now runs under Linux because of me. It has a 5% better sampling rate than anyone else."
Look, the simple fact is that a lot of people aren't sure you're worth hiring.
Why? We all know people in our degree program, who graduated, who we think are useless jackasses. I knew quite a few CpEs who barely passed digital logic or programming and just wanted to build PCs for gaming all day. Some people finish CpE programs and design silicon for the big boys, others, who sat next to them in the same courses at the same time, tell people how to operate the power switch to make Windows work again. Your resume says you're in the latter group. Not only do they think you're worthless, they hate you for making their degrees look bad.
The fact that you got the degree but haven't found a use for it is a huge red flag. Especially in something useful like CpE. Good God man.
Despite that, it's not some scapegoat chicken & the egg problem. It's a technical degree, and you're applying to technical jobs. People can test what you actually know. The interview should include technical questions. Thats your chance to prove you're not a jackass. I've never been to an interview for a technical job without them asking me to talk code (I'm CS). Perhaps you've already forgotten too much? Or never really knew? Time to go through your textbooks and learn up.
Additionally, listing the courses and grades you got in them would help a lot. Unless you really are in the second category I listed, in which case you should look for the dumbest fucking hiring manager you can, and try and weasel your way into a mediocre admin job and let him think it's his lucky fucking day for getting a "massively overqualified" CpE.
Let's face it: Linux has stagnated. It used to be the hot new kid around, bringing together all the new-world desktop technology with the old-school unix reliability, modularity, and maintainability. It did it by being somewhere in between the two. And it took off! People loved it over windows.
But, look at it now. When's the last time it did something *well*? Its standard was always Windows, which is a very low bar to aim for. It's full of sprawl and half-implemented ideas that you have to constantly hack at to make the system work. It's been 10 years and Linux is still a maintenance pain-in-the-ass.
Most people don't care: they're happy to run a tomcat or php stack on top of it. For them, it's really just a SATA and ethernet device driver, which then lets you use your favorite app server.
Containers/Zones, ZFS, dtrace, and SMF blow linux out of the water in every category the affect. What's linux done in the last 4 years? More Windows Ketchup? Thanks but no thanks. I'll take good documentation, manageability, and stable behavior any day. Sorry, but Linux is still too much of a hobby instead of an OS for my book.
Simple answer, 3 parts:
1. Document more than is necessary. B/c your estimates of what's necessary will always obsolesce over enough time.
2. Test cases. Tests that make sure the code runs, tests that you can run later to see what broke what.
3. Don't be a savage. Use real identifier names (i,j,k are only ok in very specific cases), well indented code, version control, etc. Code Complete was suggested above, it's a good place to start. If your shop doesn't do version control, that's no excuse for you not to.
Quick patch.
1.5 Fuck it up.
Btw, reasons like that are why programmers count by 10 :-)
Virgin mobile, which uses Sprint's network.
Buy minutes in bulk @ virgin mobile. The only problem is, unless you pump $20/mon into the acct each month, the minutes expire in a month.
But $20/mon isn't too bad for me, and I can just let those $20 additions accumulate until I can either bulk-buy another 5c/min.
Also, not worrying about a broken/lost phone is pretty nice. A decent pitcher of beer costs more than the damage I could do to the phone from drinking the pitcher!
Is that 0.09581 almost 1% or 9.5% of our internet-hopping population?
I'd really believe the latter, due to various malware on most people's machines.
Actually, it's fun to jump into a coffee shop and port scan everyone, just to see who's head pops up.
Then you look them in the eye, and smile. If you can, take their picture and walk out. It'll drive their little paranoid hearts crazy for weeks :-D
May I propose looking into the prepaid phones? They are the perfect hacker phone:
1. Cheap for low-minute users. Save your money for RAM.
2. Little to no personal information. No credit check. You can tell them whatever you want, and they don't care enough to validate. Helps relieve paranoia.
3. No contract, no obligation. Just let the minutes expire and dump the phone. Helps relieve paranoia.
4. Easily switchable. Psycho ex keep calling? Get a new account in a few minutes. With the GSM devices, just swap SIMs and burn the old one. Helps relieve paranoia.
I end up paying about $20/mon to keep my acct active with enough minutes. I pay about a nickel a minute to talk. If I break the phone, I hit wal-mart and get a replacement for $10, which they activate onto my existing account.
And the phone's a basic flip-device. No bells, no whistles. Did I mention the phone was $10 new? $10.
$10.