If you have to pay a phalanx of QA engineers to find bugs post-facto ("just as important as our development department"), you're doing it wrong. The bugs shouldn't have been there to begin with.
You could add unit tests, integration tests and the next framework test. You can be agile. You can use TDD. You can use everything under the sun. But when you have a large complex system with 200 processes churning out data at 200 messages a second to each other over sockets, then you can almost bet your top dollar that a bug will resurface.
The problem is, just as one can write bad code, one can write even worse unit tests. Unit tests are subjective, just as the production code is.
Or he might be coming from a different background where they don't emphasize on performance. I've had experience where platform support teams are on the back burner (or maybe just mis-managed ?) most of the time, that all they have time to do is perform installations and then move on to the next task.
Or, he could even come from a non-Unix sys-admin background ? He didn't really mention where his experience is.
Or, he could just be a junior member on his team with no-one around to mentor him ?
All we're doing here is speculation.
I wonder what happens once the hype of Half Life 2 dies off, that is in a couple of years, when they are crazily interested in a game, would they actually demolish their half life 2 artwork and put the pc in a ?
I've always had a decent case, one that does not break easily.. but one that looks simple, and I am happy. Just wondering what these guys do when all the hype has died away.
Re:I have a bit of a bone to pick.
on
Learning PHP 5
·
· Score: 1
PHP is maturing. Its version number is the same as Perl
In that case Emacs must be in the great great great great great great great great great grandparents era.
Hey, I guess it's time for.geek to come out as well, or even.linux,.bsd,.whatever,.pr0n,.h4x0r....
On that matter, are there more Linux users than travel agents worldwide? Why have a different group for travel agents? Why stop there? What about a special one for the kebab shop round the corner? Just a wonder...
My point is that it does not matter if there is one top level command. Developers will do what's needed to get it working.
MSFT have an obligation to make sure that they do updates properly. Agreed, but so do Linux distros. That's what they are there for. To coordinate that none of their updates brake other applications. I highly doubt if Mandrake/RedHat,Fedora/Gentoo/ actually would release an update if it breaks their system.
No difference whatsoever. The top level is just the same. Hence that is why I thought your comment on the capabilities of Linux to deploy a huge update is somewhat far fetched.
Huh? Are you sure with this? Because the other day I was installing an update on, doh, can't remember what it was (it was a small update), and it did ask me to reboot. However, I did just ignore it, but it kept bugging me about rebooting every 5 mins or so.
Looking at the dependencies and the trickle effect, I think looking at the DLL hell that Microsoft comes with is just the same picture. See, I can't say for sure how the internal development is in Microsoft.
I do develop programs in Windows and do get a bump in the head when there is a library gets bonkered with weird dependencies.
Dependencies will be there for whatever platform that you are writing/developing in. It's how programs are made modular. Otherwise, everyone would love to have statically linked libraries and carry around massive binaries.
Having one controller does not make a difference. To each of the development teams, that takes care of each aspect of their own part, like the IE developement team, or the MFC development team, or whatever component of the OS, will have to make sure their dependencies with other programs are fixed.
Ripples will always be there. Solving them is the key. Having one controller is not the key here. Having a collaboration between your development teams is the key! Moreover, solving dependencies will be much much easier is everything is open sourced. I hope within MSFT all development teams can peek at each other's code. Just the other day I had a problem with a DLL that someone built, and looked at his code and fixed my code (my program was buggy). Being open-sourced does matter in debugging.
How can we convince people not to run admin mode? It's easy at work, in UNIX land (most people don't get to know root pw.) But most Windows users I know don't even know the difference.
Agreed. I think it is how the system gets installed and auto-logins as the user itself, not making many users aware of an Administrator account and what it means.
However, in *nix land, in most installations, you are told wtf is a root user, during installation it asks you to set it up, remember? And most of them give warnings like "Don't use this account for everyday use". I think people will listen to that. When needed, most programs, like Yast and Mandrake's Control Centre do ask for the root password clearly. This does not invoke complications.
Frankly, I dont think Linux could come close to releasing a patch of this magnitude with as little side effects.
a) Linux distros does not need a 400MB update at one go!
b) Even if it does need, say if you upgraded a distro that has not been upgraded in a couple of years (if it does reach that much), it won't break anything (severely). What you have to consider is that all the Linux distro updates happen per package wise (even the kernel) and hence it can be thought of as a "rolling update". I know Microsoft do this as well. But your comment on Linux not being able to patch in such a huge magnitude is a little far fetched.
True to a certain extent. Even as we speak now, not many people can afford the 80000 CPU hours. So, Joe Schmoe won't be able to crack the MD5, unless he gets 1000 (say) of his mates to help hand in hand.
The problem is that these algorithms are meant to be one way. No two files are meant to share the same hash. At the end of the day, there may be many more files that may prove to have the same effect. As they say: it's only a matter of time.
It may be that this is only a one timer, it may be that they have found a major flaw in the algorithms and more combination of files may have the same hash. But these events are eye openers as they have a "proof of concept" theme to it, and one can just extend it further. Who knows, maybe one day some guy actually manages to craft another file that can have the same hash.
I remember that one, where Inland Revenue had a mock up in their system, so much so that some people could look into other people's tax returns etc etc. These things hold very confidential information. Putting them on the internet comes with a very high risk, which I feel the Fed Res should not take on.
I have discussed this with many people and I keep wondering why the Financial Industry still relies (to a certain extent) on Telex, rather than using the Internet's wonderful distributed system to route all information. Thing is old technology comes with a price. Security comes with a price. Risk comes with a price as well.
At the end of the day, everyone has to look and see what suites them. Fed Res having all transactions on the Internet will get DDoS from god-knows-who-is-bored-today. Fed Res should really just stick with a private network. It's more secure. The risk however that they will have is a fallover plan, which the Internet readily provides. But then, when was the last time that anyone heard "The Fed Res private network is down! We can't put in our trades on ABC options today!".
Running a private network is costly, but I feel the benefits of moving to the Internet does not marginally come close to the benefits of keeping on a private network
Funny thing is, most people (and admins as well) when they do get r00ted or whatever, don't realise it. There are a lot of rootkits and tools out there to hide everything nicely. This is where the kiddies come into play. Download, click and go. Kinda like a Windows analogy:)
Slashdot Mirror
If you have to pay a phalanx of QA engineers to find bugs post-facto ("just as important as our development department"), you're doing it wrong. The bugs shouldn't have been there to begin with.
You could add unit tests, integration tests and the next framework test. You can be agile. You can use TDD. You can use everything under the sun. But when you have a large complex system with 200 processes churning out data at 200 messages a second to each other over sockets, then you can almost bet your top dollar that a bug will resurface.
The problem is, just as one can write bad code, one can write even worse unit tests. Unit tests are subjective, just as the production code is.
So many times I've had to go in and make a small fix for a usability issue, and ended up having to refactor a couple hundred lines of code.
Totally agree with this. Just as there are many good coders, there are just as many bad ones. No structure, no style, no zing!
Or he might be coming from a different background where they don't emphasize on performance. I've had experience where platform support teams are on the back burner (or maybe just mis-managed ?) most of the time, that all they have time to do is perform installations and then move on to the next task. Or, he could even come from a non-Unix sys-admin background ? He didn't really mention where his experience is. Or, he could just be a junior member on his team with no-one around to mentor him ? All we're doing here is speculation.
I'm not sure how well a RAIH5 solution would work though but I'm sure there are plenty of people working on that though.
All a possibility since houses are now going for $1.
I wonder what happens once the hype of Half Life 2 dies off, that is in a couple of years, when they are crazily interested in a game, would they actually demolish their half life 2 artwork and put the pc in a ?
I've always had a decent case, one that does not break easily.. but one that looks simple, and I am happy. Just wondering what these guys do when all the hype has died away.
PHP is maturing. Its version number is the same as Perl
In that case Emacs must be in the great great great great great great great great great grandparents era.
Hey, I guess it's time for .geek to come out as well, or even .linux, .bsd, .whatever, .pr0n, .h4x0r ....
On that matter, are there more Linux users than travel agents worldwide? Why have a different group for travel agents? Why stop there? What about a special one for the kebab shop round the corner? Just a wonder...
That's all well and good. But, why does it crash so often? Or have a BSOD often?
Just want your point of view... (not starting a flame war)
Can this fall under "organized crime"??
Well, it was organized after all (in some sense), and all of them sound like Tommy Vercetti from Vice City.
... I'm going to make you eat those words.
Are you going to print out his comment and ask him to eat paper?
Actually, a better thing to do is to to use one of these , display the comment and stuff it down.
ditto :D
These litigious Bastards are back again! Can't they smell that being litigious Bastards is pretty bad. Why don't these litigious Bastards ever just give up and realise that being litigious Bastards is not going to help bring up their litigious Bastards stock prices. Damn litigious Bastards
My point is that it does not matter if there is one top level command. Developers will do what's needed to get it working.
MSFT have an obligation to make sure that they do updates properly. Agreed, but so do Linux distros. That's what they are there for. To coordinate that none of their updates brake other applications. I highly doubt if Mandrake/RedHat,Fedora/Gentoo/ actually would release an update if it breaks their system.
No difference whatsoever. The top level is just the same. Hence that is why I thought your comment on the capabilities of Linux to deploy a huge update is somewhat far fetched.
Huh? Are you sure with this? Because the other day I was installing an update on, doh, can't remember what it was (it was a small update), and it did ask me to reboot. However, I did just ignore it, but it kept bugging me about rebooting every 5 mins or so.
But my question is... how many more forks will we have?
:D
As many spoons and knifes that we have
Looking at the dependencies and the trickle effect, I think looking at the DLL hell that Microsoft comes with is just the same picture. See, I can't say for sure how the internal development is in Microsoft.
I do develop programs in Windows and do get a bump in the head when there is a library gets bonkered with weird dependencies.
Dependencies will be there for whatever platform that you are writing/developing in. It's how programs are made modular. Otherwise, everyone would love to have statically linked libraries and carry around massive binaries.
Having one controller does not make a difference. To each of the development teams, that takes care of each aspect of their own part, like the IE developement team, or the MFC development team, or whatever component of the OS, will have to make sure their dependencies with other programs are fixed.
Ripples will always be there. Solving them is the key. Having one controller is not the key here. Having a collaboration between your development teams is the key! Moreover, solving dependencies will be much much easier is everything is open sourced. I hope within MSFT all development teams can peek at each other's code. Just the other day I had a problem with a DLL that someone built, and looked at his code and fixed my code (my program was buggy). Being open-sourced does matter in debugging.
You will do well in sales :D
But he can run it as a cron :) and not need the system reboot for each application that gets updated :) except the kernel, of course.
How can we convince people not to run admin mode? It's easy at work, in UNIX land (most people don't get to know root pw.) But most Windows users I know don't even know the difference.
Agreed. I think it is how the system gets installed and auto-logins as the user itself, not making many users aware of an Administrator account and what it means.
However, in *nix land, in most installations, you are told wtf is a root user, during installation it asks you to set it up, remember? And most of them give warnings like "Don't use this account for everyday use". I think people will listen to that. When needed, most programs, like Yast and Mandrake's Control Centre do ask for the root password clearly. This does not invoke complications.
Frankly, I dont think Linux could come close to releasing a patch of this magnitude with as little side effects.
a) Linux distros does not need a 400MB update at one go!
b) Even if it does need, say if you upgraded a distro that has not been upgraded in a couple of years (if it does reach that much), it won't break anything (severely). What you have to consider is that all the Linux distro updates happen per package wise (even the kernel) and hence it can be thought of as a "rolling update". I know Microsoft do this as well. But your comment on Linux not being able to patch in such a huge magnitude is a little far fetched.
Where/Who is his Minnie Me?
Yeah, that's like Emacs, when it was released a whole operating system came out from it!
True to a certain extent. Even as we speak now, not many people can afford the 80000 CPU hours. So, Joe Schmoe won't be able to crack the MD5, unless he gets 1000 (say) of his mates to help hand in hand.
The problem is that these algorithms are meant to be one way. No two files are meant to share the same hash. At the end of the day, there may be many more files that may prove to have the same effect. As they say: it's only a matter of time.
It may be that this is only a one timer, it may be that they have found a major flaw in the algorithms and more combination of files may have the same hash. But these events are eye openers as they have a "proof of concept" theme to it, and one can just extend it further. Who knows, maybe one day some guy actually manages to craft another file that can have the same hash.
I remember that one, where Inland Revenue had a mock up in their system, so much so that some people could look into other people's tax returns etc etc. These things hold very confidential information. Putting them on the internet comes with a very high risk, which I feel the Fed Res should not take on.
I have discussed this with many people and I keep wondering why the Financial Industry still relies (to a certain extent) on Telex, rather than using the Internet's wonderful distributed system to route all information. Thing is old technology comes with a price. Security comes with a price. Risk comes with a price as well.
At the end of the day, everyone has to look and see what suites them. Fed Res having all transactions on the Internet will get DDoS from god-knows-who-is-bored-today. Fed Res should really just stick with a private network. It's more secure. The risk however that they will have is a fallover plan, which the Internet readily provides. But then, when was the last time that anyone heard "The Fed Res private network is down! We can't put in our trades on ABC options today!".
Running a private network is costly, but I feel the benefits of moving to the Internet does not marginally come close to the benefits of keeping on a private network
Just my $0.02
Funny thing is, most people (and admins as well) when they do get r00ted or whatever, don't realise it. There are a lot of rootkits and tools out there to hide everything nicely. This is where the kiddies come into play. Download, click and go. Kinda like a Windows analogy :)