Wow. I haven't thought about WindowMaker for years. I always enjoyed that wm. If I could be arsed to tear myself away from the joy of openbox + tint2 + conky, I would;)
If I could easily find the bug reports in multiple distros I would. This happened on multiple Broadcom cards against various AP vendors. The NIC doesn't disassociate unless the scan takes longer than a given period. In any given environment, I can count 10-15 networks in range and scanning each of those takes time. While scanning I suffer reduced throughput and if the scan takes too long, the connection is terminated and has to be reestablished.
I don't claim to be a hardware engineer. I don't know what the IETF says about how background scanning should be handled but the inability to disable the background scanning, the fact that it doesn't scale to larger numbers of available networks AND the fact that it causes reduced throughput while scanning is bothersome to me. The same problem doesn't happen under Windows on the same machines.
I've been running linux for a LONG time (Yggdrassil anyone?). The fact that something works under Windows and doesn't under Linux only has one thing to blame - the software. That can be driver or user space but unless it's something like the winmodems of yore, it's quite obvious that background scanning does work with the hardware.
Mainly because my main environment is a wired desktop. For that, NetworkManager is fine. I can deal with a single component in an entire distro being problematic because I can easily replace that single component where it makes sense.
And in the interest of full disclosure, I don't actually run the stock ubuntu desktop. I switched to openbox+tint2 about 4 or 5 months back and prefer that combination.
NetworkManager is a piece of shit w.r.t wireless. I've read every fucking thread out on various mailing lists and the author simply says "It's the driver's fault" despite the same problem happening across the board to multiple users of different cards.
The biggest problem is the stupid fucking background scanning it does. What happens is that when NetworkManager gets a wild hair up its ass and decides it time to scan for more networks, your wireless NIC will disassociate from the current AP until the scanning is over. God forbid there happens to be one shitty AP somewhere at the edge of your range and it takes too long to respond. Your connection is toast and you have to re-associate but meanwhile you've just lost connectivity for 2 minutes. Hope you didn't need that download anytime soon or that you remembered to screen that SSH session to a production server. Any machine I use that has wireless, is running WICD now instead of NetworkManager ( http://wicd.sourceforge.net/ )
I love Ubuntu. Honestly the only problems I've ever had were with the switch to PulseAudio. I grew out of tinkering with my distros a LONG time ago. I need my machine to work so I can work. I did a fresh install of Karmic and moved my home partition stuff around this time. The ONLY problem I had was with PulseAudio and my Audigy card ( https://bugs.launchpad.net/bugs/467732 ).
You don't understand how the BB Storm is designed.
The Storm2 ships with three "memory" areas (yes I'm using memory incorrectly): 256MB OS pool 1GB Internal Storage 16GB External Storage (microsd)
The 256MB is where the OS runs, applications run and messages are stored. The original storm had 128MB. It would take 50MB just after powering on. In this case, the Storm2 is just delaying the problem instead of fixing it. We'll have to see how much better memory management is in OS5.
The 1GB internal storage is where pictures and songs and ringtones and such are stored by default. You can change this.
The 16GB microsd is the data storage area.
The biggest problem that RIM has is they have too many fucking devices. In the interest of developer sanity, they can't do cool shit like move the message store outside the 256MB area to the microsd because that might not work on another model that doesn't have a microsd slot.
Having said that, the storm2 from a data storage perspective is on par with the iphone.
That's the thing I never understood. There was NOTHING preventing RIM from putting more than the measly 128M memory pool in there. It was a cheap shitty move. Even bumping up to 1GB may not solve the problems...just push out the time needed to reboot. RIM was dragged kicking and screaming out of the pager market and the hardware occasionally shows that.
We'll see. I've not loaded OS5 on my current storm yet so I have no idea how much better or worse the memory management is.
That's the point though. What's the point of adding Wifi to the Storm 2? Nothing will really take advantage of it. Why does the BB *HAVE* to use a BIS in this day and age? I can SOMEWHAT understand BES but honestly with ActiveSync and EWS, there's no point to BES anymore.
The biggest reason I like my Storm is the feedback I get from the screen while typing. But that's slowly becoming less of a reason for me to keep it.
This is the biggest thing I've come to hate about my Storm (and the Storm 2). ANY push functionality has to go via RIM. The wifi is next to pointless on the Storm 2 the more that I look at it because I can't DO anything with it other than browse the web faster than over 3G. Oh Podtrapper will get my podcasts faster too. Whoop dee!
I was strongly considering upgrade my Storm to the Storm 2 but I think I may hold off until Verizon gets the Wifi android phone next year. The whole BIS/BES + your PIN/physical phone hardware is who you are (versus me being who I am) is dated.
For the sake of not losing information, I'd definitely not be adverse to creating a temporary copy of a document when a user opens it for editing, auto-saving to this temp document and leaving any older versions of the document read-only when the user clicks "Save". Still, I'd stick with my one-user lock scheme to avoid two users editing a document and having only the version saved last appear as the most current version.
I agree. This should really be the norm. This shouldn't operate ANY differently than a database does depending on isolation level. I like the idea of a collab-style "this use is also editing" but in the end, why the hell are two people editing the same thing at the same time other than for non-technical reasons. You can't solve ALL soft skill problems with technology.
So really there are a few options:
collaborative style notifications/shared workspace messaging (the whole "this user is editing" thing)
exclusive access at edit with an acceptable timeout (the process for determining the timeout is irrelevant)
dirty reads where similar to READ-UNCOMMITTED isolation level
I like a hybrid approach using exclusive access + collab along with some sort of versioning in place. I.E.
Suzie opens doc 1. Begins to edit.
Bobby opens doc 1. Notified that Suzie has exclusive access. Bobby is given the option of seeing the current saved version and allowed an option to edit a temporary copy.
In the event that Suzie is still editing when Bobby saves, Bobby's copy is stored as a Draft. At the point that Suzie finally commits her edit (after taking a 20 minute smoke break, hitting the head and getting lunch), Bobby is given the option to compare his changes to Suzie's changes.
At this point Bobby has exclusive access in case Frank decides to ALSO update the fucking record.
If Charles comes along to ALSO edit this obviously popular entry, he's prevented because these people obviously aren't talking to each other and need to get in the same fucking room and work some shit out.
Let me clarify about the codebase statement. I frequently reuse stuff I've written from project to project. So in essence, yes, my codebase is the same. It's the same logic, methods what not. While Nominum may not be using the EXACT bind9 codebase, if the same people who wrote it are writing code for the same type of product, the same code will be present in some form.
So as someone who works there, are you kindly going to tell this guy that he's a stupid git? The fact that he's slamming Bind and yet it's quite evident that the codebase for Nominum is most likely the Bind9 codebase is the same codebase for Nominum's product.
At BEST he's just basically said that the developers at the company can't write secure code (bind is insecure and dangerous for customers + the bind9 developers work for us = our code is insecure and dangerous for customers).
No. Private companies pay executives for the associated risks for being executives. If you think you're up to the job of managing a multinational corporation, then I implore you to give it a shot.
And as a side note, those jobs didn't belong to those tens of thousands of people. They belonged to the company.
Seriously. Ask someone at Google about mac address conflicts. When I interviewed with them I got the chance to ask a few questions so I said "What's the one thing you've seen at Google that you've never had to deal with anywhere else?"
The guy said "MAC address conflicts". Evidently because they buy SO many servers in bulk that they've frequently gotten batches of servers with the same MAC address.
I'm a fan of the asset tag solution myself. Considering most asset tags are generated by the printer specifically for your company, you don't really run the risk of conflicts.
At a financial company I worked for we actually did SSN masking for QA and Development (we restored those databases from production backups occasionally) and it wasn't a problem because we didn't use the SSN as a unique identifier. Anyone who wants to use an SSN as PK or unique identifier is just lazy.
Yes they did. I can only think of a handful of products that were Microsoft "originals" and even those were generic products. SQL Server is based on Sybase. In fairness, SQL Server 7 and beyond are supposed codebase rewrites.
"Clearly, the networked monitors with alarms sounding so frequently no one believed they meant anything is a serious design problem"
This isn't just applicable to this system. I can't tell you how many places I've been were network and system alarms were ignored and the answer was "that's one that we don't worry about". It leads to a really bad place. It always ends up that a real problem got missed because "app02 always has an alarm".
It does when the provider uses a home-made cable as justification for line errors. Stranger things have happened.
Re:Postgres is looking better than ever
on
Oracle Buys Sun
·
· Score: 1
And indices in DB2 are case-sensitive. What's your point? As the person who replied before me said, there's a big difference between "I helped my uncle Jack of a horse" and "I helped my uncle jack off a horse".
Case is important, yo! Why is it the database's job to divine what you meant to query instead of what you actually queried? Did you vote in Florida in 2000?
You're absolutely wrong. I've already pointed this out in another thread and it's been covered in multiple places since Boxee was banned from Hulu.
The cable/sat companies are the driving force behind this issue. The advertising revenue is actually DOWN right now. The cable companies paying for content actually make up the largest part of revenue for the content creators.
Content providers actually get to double dip right now. They cable/sat companies pay for the content and they get, if the ad is national (some ads are local and NBC corp doesn't get that revenue), they get another chunk.
While that's an extreme circumstance, there's probably a process in place to track those changes. Additionally, the checks to make sure that you don't cause any problems on the switch might mean that the work has to be done after hours.
It sounds like a simple change but considering you (the requestor) isn't the one on the hook if god forbid something does go wrong, I think I (the person responsible) has the right to do some due diligence.
Right, because an MBA was never stupid enough to say they needed something online in a week when it would take at least 2 weeks to get something as simple as hardware.
Or maybe there's this whole thing called SOX that requires us BY LAW to implement change control processes and auditing so we can stay in business.
The next time an MBA is willing to take a phone call at 4AM because something went wrong with that "simple change", then we can have a conversation.
You're absolutely right. The revenue was nowhere NEAR what spots on television were. However it could have been. I still think Boxee is/was the killer app for Hulu.
To answer the question, there are at least 4 parties in the mix for buying a TV spot (let's assume this is a local spot).
buyer (Joe's House of Shoes) ad agency broker (Katz for instance) station
That's a whole lot of moving parts. There are software businesses built around connecting all these people. A lot of people have fingers in the pie. Hulu may or may not have cut out part of that process. Probably not but the end cost was probably much cheaper to the person who drives all this, the buyer. Sure the cost would have gone up down the road as popularity for Hulu soared but right now, Hulu was probably the cheapest route to get your ads out there.
Slashdot Mirror
Wow. I haven't thought about WindowMaker for years. I always enjoyed that wm. If I could be arsed to tear myself away from the joy of openbox + tint2 + conky, I would ;)
If I could easily find the bug reports in multiple distros I would. This happened on multiple Broadcom cards against various AP vendors. The NIC doesn't disassociate unless the scan takes longer than a given period. In any given environment, I can count 10-15 networks in range and scanning each of those takes time. While scanning I suffer reduced throughput and if the scan takes too long, the connection is terminated and has to be reestablished.
I don't claim to be a hardware engineer. I don't know what the IETF says about how background scanning should be handled but the inability to disable the background scanning, the fact that it doesn't scale to larger numbers of available networks AND the fact that it causes reduced throughput while scanning is bothersome to me. The same problem doesn't happen under Windows on the same machines.
I've been running linux for a LONG time (Yggdrassil anyone?). The fact that something works under Windows and doesn't under Linux only has one thing to blame - the software. That can be driver or user space but unless it's something like the winmodems of yore, it's quite obvious that background scanning does work with the hardware.
Mainly because my main environment is a wired desktop. For that, NetworkManager is fine. I can deal with a single component in an entire distro being problematic because I can easily replace that single component where it makes sense.
And in the interest of full disclosure, I don't actually run the stock ubuntu desktop. I switched to openbox+tint2 about 4 or 5 months back and prefer that combination.
NetworkManager is a piece of shit w.r.t wireless. I've read every fucking thread out on various mailing lists and the author simply says "It's the driver's fault" despite the same problem happening across the board to multiple users of different cards.
The biggest problem is the stupid fucking background scanning it does. What happens is that when NetworkManager gets a wild hair up its ass and decides it time to scan for more networks, your wireless NIC will disassociate from the current AP until the scanning is over. God forbid there happens to be one shitty AP somewhere at the edge of your range and it takes too long to respond. Your connection is toast and you have to re-associate but meanwhile you've just lost connectivity for 2 minutes. Hope you didn't need that download anytime soon or that you remembered to screen that SSH session to a production server. Any machine I use that has wireless, is running WICD now instead of NetworkManager ( http://wicd.sourceforge.net/ )
I love Ubuntu. Honestly the only problems I've ever had were with the switch to PulseAudio. I grew out of tinkering with my distros a LONG time ago. I need my machine to work so I can work. I did a fresh install of Karmic and moved my home partition stuff around this time. The ONLY problem I had was with PulseAudio and my Audigy card ( https://bugs.launchpad.net/bugs/467732 ).
Unfortunately that's not a wifi model. The last I read was that a WiFi android phone would be hitting verizon Q1 next year.
You don't understand how the BB Storm is designed.
The Storm2 ships with three "memory" areas (yes I'm using memory incorrectly):
256MB OS pool
1GB Internal Storage
16GB External Storage (microsd)
The 256MB is where the OS runs, applications run and messages are stored. The original storm had 128MB. It would take 50MB just after powering on. In this case, the Storm2 is just delaying the problem instead of fixing it. We'll have to see how much better memory management is in OS5.
The 1GB internal storage is where pictures and songs and ringtones and such are stored by default. You can change this.
The 16GB microsd is the data storage area.
The biggest problem that RIM has is they have too many fucking devices. In the interest of developer sanity, they can't do cool shit like move the message store outside the 256MB area to the microsd because that might not work on another model that doesn't have a microsd slot.
Having said that, the storm2 from a data storage perspective is on par with the iphone.
That's the thing I never understood. There was NOTHING preventing RIM from putting more than the measly 128M memory pool in there. It was a cheap shitty move. Even bumping up to 1GB may not solve the problems...just push out the time needed to reboot. RIM was dragged kicking and screaming out of the pager market and the hardware occasionally shows that.
We'll see. I've not loaded OS5 on my current storm yet so I have no idea how much better or worse the memory management is.
That's the point though. What's the point of adding Wifi to the Storm 2? Nothing will really take advantage of it. Why does the BB *HAVE* to use a BIS in this day and age? I can SOMEWHAT understand BES but honestly with ActiveSync and EWS, there's no point to BES anymore.
The biggest reason I like my Storm is the feedback I get from the screen while typing. But that's slowly becoming less of a reason for me to keep it.
This is the biggest thing I've come to hate about my Storm (and the Storm 2). ANY push functionality has to go via RIM. The wifi is next to pointless on the Storm 2 the more that I look at it because I can't DO anything with it other than browse the web faster than over 3G. Oh Podtrapper will get my podcasts faster too. Whoop dee!
I was strongly considering upgrade my Storm to the Storm 2 but I think I may hold off until Verizon gets the Wifi android phone next year. The whole BIS/BES + your PIN/physical phone hardware is who you are (versus me being who I am) is dated.
I agree. This should really be the norm. This shouldn't operate ANY differently than a database does depending on isolation level. I like the idea of a collab-style "this use is also editing" but in the end, why the hell are two people editing the same thing at the same time other than for non-technical reasons. You can't solve ALL soft skill problems with technology.
So really there are a few options:
I like a hybrid approach using exclusive access + collab along with some sort of versioning in place. I.E.
Hate speech is still speech and should still be protected. The first amendment doesn't exist to protect popular speech but unpopular speech.
Hate crimes/hate speech are stupid concepts anyway. It's basically the modern equivalent of thoughtcrime.
Let me clarify about the codebase statement. I frequently reuse stuff I've written from project to project. So in essence, yes, my codebase is the same. It's the same logic, methods what not. While Nominum may not be using the EXACT bind9 codebase, if the same people who wrote it are writing code for the same type of product, the same code will be present in some form.
So as someone who works there, are you kindly going to tell this guy that he's a stupid git? The fact that he's slamming Bind and yet it's quite evident that the codebase for Nominum is most likely the Bind9 codebase is the same codebase for Nominum's product.
At BEST he's just basically said that the developers at the company can't write secure code (bind is insecure and dangerous for customers + the bind9 developers work for us = our code is insecure and dangerous for customers).
No. Private companies pay executives for the associated risks for being executives. If you think you're up to the job of managing a multinational corporation, then I implore you to give it a shot.
And as a side note, those jobs didn't belong to those tens of thousands of people. They belonged to the company.
Seriously. Ask someone at Google about mac address conflicts. When I interviewed with them I got the chance to ask a few questions so I said "What's the one thing you've seen at Google that you've never had to deal with anywhere else?"
The guy said "MAC address conflicts". Evidently because they buy SO many servers in bulk that they've frequently gotten batches of servers with the same MAC address.
I'm a fan of the asset tag solution myself. Considering most asset tags are generated by the printer specifically for your company, you don't really run the risk of conflicts.
At a financial company I worked for we actually did SSN masking for QA and Development (we restored those databases from production backups occasionally) and it wasn't a problem because we didn't use the SSN as a unique identifier. Anyone who wants to use an SSN as PK or unique identifier is just lazy.
Yes they did. I can only think of a handful of products that were Microsoft "originals" and even those were generic products. SQL Server is based on Sybase. In fairness, SQL Server 7 and beyond are supposed codebase rewrites.
"Clearly, the networked monitors with alarms sounding so frequently no one believed they meant anything is a serious design problem"
This isn't just applicable to this system. I can't tell you how many places I've been were network and system alarms were ignored and the answer was "that's one that we don't worry about". It leads to a really bad place. It always ends up that a real problem got missed because "app02 always has an alarm".
It does when the provider uses a home-made cable as justification for line errors. Stranger things have happened.
And indices in DB2 are case-sensitive. What's your point? As the person who replied before me said, there's a big difference between "I helped my uncle Jack of a horse" and "I helped my uncle jack off a horse".
Case is important, yo! Why is it the database's job to divine what you meant to query instead of what you actually queried? Did you vote in Florida in 2000?
You're absolutely wrong. I've already pointed this out in another thread and it's been covered in multiple places since Boxee was banned from Hulu.
The cable/sat companies are the driving force behind this issue. The advertising revenue is actually DOWN right now. The cable companies paying for content actually make up the largest part of revenue for the content creators.
Content providers actually get to double dip right now. They cable/sat companies pay for the content and they get, if the ad is national (some ads are local and NBC corp doesn't get that revenue), they get another chunk.
the law is on Microsoft's side. Now, what they can do is have a state IT policy that says the standard is XP and be done with it.
While that's an extreme circumstance, there's probably a process in place to track those changes. Additionally, the checks to make sure that you don't cause any problems on the switch might mean that the work has to be done after hours.
It sounds like a simple change but considering you (the requestor) isn't the one on the hook if god forbid something does go wrong, I think I (the person responsible) has the right to do some due diligence.
Right, because an MBA was never stupid enough to say they needed something online in a week when it would take at least 2 weeks to get something as simple as hardware.
Or maybe there's this whole thing called SOX that requires us BY LAW to implement change control processes and auditing so we can stay in business.
The next time an MBA is willing to take a phone call at 4AM because something went wrong with that "simple change", then we can have a conversation.
You're absolutely right. The revenue was nowhere NEAR what spots on television were. However it could have been. I still think Boxee is/was the killer app for Hulu.
To answer the question, there are at least 4 parties in the mix for buying a TV spot (let's assume this is a local spot).
buyer (Joe's House of Shoes)
ad agency
broker (Katz for instance)
station
That's a whole lot of moving parts. There are software businesses built around connecting all these people. A lot of people have fingers in the pie. Hulu may or may not have cut out part of that process. Probably not but the end cost was probably much cheaper to the person who drives all this, the buyer. Sure the cost would have gone up down the road as popularity for Hulu soared but right now, Hulu was probably the cheapest route to get your ads out there.