if android really takes off (and it seems to be starting that roll around now, given the number of devices coming out), that may change, as the dev environment is java based and so can be run on top of just about anything that can handle a java VM.
Which still makes a Mac a better platform for a developer who wants to write code on multiple platforms. JavaVM? great. OS X / OS X mobile? done. Windows.NET? Virtualbox / Fusion to the rescue. Web/php/python/mysql/whatever? It's all available.
How do browsers behave when 1 of the server goes down?
Half the DNS lookups will still point at the failed server. Since most browsers cache the dns lookup, they will not re-request the IP address, and will just assume the site is down.
If you rely on DNS round-robin records, you need to either ensure that they are always up (i.e. each one is an HA cluster) or that you can remove them quickly enough to cause your users as little pain as possible. (TTL should probably be 1 minute, and you'll want an automated method of removing the bad entry from the zone file)
They want me to put credit card information into a standard HTTP page? I'm absolutely SURE I trust them not to do something stupid like store that information unencrypted in a database.
Yeah, that's what I don't get. Amazon's Elastic IPs are free as long as there's a host running that they are assigned to. Plus they have an API for re-assigning them on the fly. Just assign them to the front end pound box and be done with it.
I've actually just completed migrating the host OS on my vmware servers from Windows to Ubuntu Server.
The primary motivator behind the move was a desire to throw more physical RAM at the boxes, and Windows won't handle over 4GB in it's "cheap" version.
What I discovered is that VMWare server on Linux handles memory allocation very differently than the Windows version. The Windows version allocates all the memory for the VM immediately and permanently. The Linux version seems to only allocate the memory when it's actually used. When I had placed enough virtual machines on the box to swamp it under windows, Ubuntu still reported several gigs available.
The biggest difference in my mind between them is that if you install Fedora, and then decide you want paid support, you have to migrate to RHEL, which may not contain the same stuff as the latest version of Fedora.
With Ubuntu, you install the edition you want (Ubuntu / Kubuntu / Edubuntu / Server / Whatever) and if you decide you'd like to purchase some support for it, you just pay Canonical and start asking questions.
having a pay service but giving free access to individuals minus redistribution rights, and no way of enforcing it or even checking it,
I'm pretty sure they had a way to check it. All you have to do is throw in some bad data and watch for it to show up in odd places. Pollstar does it all the time.
Isn't there any way to obtain this information in an "open-source" manner?
The best possible "open source" solution would be to create some sort of wiki for tv listing data.
Right. Cause what I want most in the world is to be the guy that discovers that the Battlestar Galactica entry got replaced by Discovery Network's real life surgery series (or CSPAN) right when my Myth TV went to update it.
The data can't be peer-reviewed, as there no way to verify it until it happens. It needs to come from an authoritative source. Especially when NBC starts shifting or super-sizing their sitcoms...
The SMTP procotol provides for reliable delivery, but MTAs which fail to adhere to the standard can lose email.
"Reliable" in the land of SMTP means that the sender will get notified if the message can't be delivered. Since it's so easy to spoof any sender's address, the majority of MTA's MUST break this standard or you'd get flooded under a pile of "you tried to deliver a message to aaaaa@somedomain.com and it failed" message everytime someone spoofs your email and starts a dictionary attack. Without reliable authentication, you cannot have reliable delivery.
This is a moderately hard problem to solve, but it's more an issue with most users being unable or unwilling to use things like PGP/GnuPGP or S/MIME.
No, that's a client side answer to a protocol-based problem. The servers that talk to each other need to be able to verify the identity of the sender, not the end users. SPF and Email caller ID are attempts to implement this, but because they will most likely never get enough of a foothold to deny email that doesn't contain such a verification layer, they are essentially useless.
This problem (in SMTP anyway) initiates from the fact that end users can easily pretend to be servers. Clients should only be allowed to talk to the servers that are authoritative for their email domain. Receiving servers should only accept email from servers that are authoritative for the domain of the sender. If you try to implement this is SMTP, it breaks many many things (forwarding, send on behalf of, smarthosts) but all those things are the reason that spam exists today. Imagine if spammers had to register a new domain for every campaign they sent out?
The overwhelming majority of email is delivered in under 10 minutes, but if someone's mailserver or Internet connection is down, then mail ain't going to be delivered until the problem gets rectified. I just checked my mailserver logs and out of the 500 emails delivered today, the average delay was about 5 seconds, and the longest delay was 111 seconds.
Just because that happens at the moment, doesn't mean that's how it's supposed to happen. The only deliverability guidelines in the SMTP RFC are that it should make it there within 3-4 days. The only methods provided to ensure it does rely on email flowing backwards to the sender, who never gets authenticated.
This is the #1 reason why SMTP is a broken protocol. The only people who know and understand how unreliable it is as a service are the people administering it, not the majority of the end user population.
It NEEDS to be torn down and rebuilt from the ground to provide (among other things): * reliable delivery (Is it there yet?) * reliable authentication (Is the sender whom they claim to be?) * timely delivery (Preferably under 10 minutes to any destination, or the original sender is notified, preferably NOT via email)
These things won't happen until the system breaks so badly that there's no alternative but to rebuild it. On the plus side, GoodMail seems as good a way to break the current system as any.
Typically printers that don't work in OS X are a lot like WinModems... They use a simple raster engine and do advanced formatting in the (Windows-only) driver.
Just as typically, these types of printers tend to work poorly under Linux as well. I had this issue with a cheap Lexmark laser printer that was advertised as Windows-only. Bought it before the MacBook it was going to be hooked up to.
I seem to remember them doing a win32 port of XBMC a while back that was less than stellar.
Also, I'm not a coder, but SDL doesn't appear to grant any interface to IR Remotes, which is one of the main reasons I still haven't ditched my xbox in favor of streaming video to my PS3. I still can't believe they didn't put an IR receiver in that thing.
As a longtime XBMC user: What XBMC brings to the table is an excellent 15-foot interface, seemless integration with a LARGE number of formats (including SMB shares and many types of streaming media) and a very active skinning / modding community. The fact that it's currently limited to the original xBox hardware is it's only weak point, in my mind. This keeps it from being able to play 720p content very well, simply because the hardware lacks the power.
A typical XBMC installation package (which is generally illegal, due to licensing restrictions on the compiler) runs between 25 and 50MB, so it's not that big.
I think the biggest challenge they will face with the linux port is hardware support. The xbox platform allowed them to focus more on the interface than trying to support a ton of different ir controllers / video cards / etc.
I wish them the best of luck, as I'd love to go get a MacMini and run XBMC (LBMC? LMC? XBMC4L?) on it.
Slashdot Mirror
Half-Life 1 was ported to Source awhile ago. So was Counter-Strike
Sure, it does _now_.
by the government of a famous athlete
I'm sure you had a joke in there that you were dying to get out, but this makes less than no sense.
Spectacularly. Thanks!
RMS knows where you can buy a completely FOSS computer
+1 good post, would mod again.
if android really takes off (and it seems to be starting that roll around now, given the number of devices coming out), that may change, as the dev environment is java based and so can be run on top of just about anything that can handle a java VM.
Which still makes a Mac a better platform for a developer who wants to write code on multiple platforms. JavaVM? great. OS X / OS X mobile? done. Windows .NET? Virtualbox / Fusion to the rescue. Web/php/python/mysql/whatever? It's all available.
Yep, every single thing costs money.
And there's absolutely no compatibility with any linux software at all.
You're 100% correct.
How do browsers behave when 1 of the server goes down?
Half the DNS lookups will still point at the failed server. Since most browsers cache the dns lookup, they will not re-request the IP address, and will just assume the site is down.
If you rely on DNS round-robin records, you need to either ensure that they are always up (i.e. each one is an HA cluster) or that you can remove them quickly enough to cause your users as little pain as possible. (TTL should probably be 1 minute, and you'll want an automated method of removing the bad entry from the zone file)
The dongle is installed inside the computer. It plugs into one of the USB headers on the system board, not a USB port. You shouldn't be removing it.
<form name="checkout_confirmation" action="http://www.efixusa.net/checkout_process.php" method="post">
So.... "no."
They want me to put credit card information into a standard HTTP page? I'm absolutely SURE I trust them not to do something stupid like store that information unencrypted in a database.
FAIL.
The trouble with EC2 and CNAMEs is that you cannot CNAME a base domain.
So you can easily cname www.example.com to whatever you need, but you cannot cname example.com at all.
The better solution (as noted below) is Amazon's Elastic IP offering. Which is free if it attached to a running EC2 instance.
Yeah, that's what I don't get. Amazon's Elastic IPs are free as long as there's a host running that they are assigned to. Plus they have an API for re-assigning them on the fly. Just assign them to the front end pound box and be done with it.
FYI, besides being out of touch with TPM, Bushnell has zero to do with Atari these days.
Atari is now a wholly owned subsidiary of Infogrammes.
Set the network to host only (assuming you are doing this on a workstation version of VMWare) and they'll never see the machine on the net.
//backdoor or //keylogger
At least you didn't name the machines
I've actually just completed migrating the host OS on my vmware servers from Windows to Ubuntu Server.
The primary motivator behind the move was a desire to throw more physical RAM at the boxes, and Windows won't handle over 4GB in it's "cheap" version.
What I discovered is that VMWare server on Linux handles memory allocation very differently than the Windows version. The Windows version allocates all the memory for the VM immediately and permanently. The Linux version seems to only allocate the memory when it's actually used. When I had placed enough virtual machines on the box to swamp it under windows, Ubuntu still reported several gigs available.
The biggest difference in my mind between them is that if you install Fedora, and then decide you want paid support, you have to migrate to RHEL, which may not contain the same stuff as the latest version of Fedora.
With Ubuntu, you install the edition you want (Ubuntu / Kubuntu / Edubuntu / Server / Whatever) and if you decide you'd like to purchase some support for it, you just pay Canonical and start asking questions.
having a pay service but giving free access to individuals minus redistribution rights, and no way of enforcing it or even checking it,
I'm pretty sure they had a way to check it. All you have to do is throw in some bad data and watch for it to show up in odd places. Pollstar does it all the time.
Isn't there any way to obtain this information in an "open-source" manner?
The best possible "open source" solution would be to create some sort of wiki for tv listing data.
Right. Cause what I want most in the world is to be the guy that discovers that the Battlestar Galactica entry got replaced by Discovery Network's real life surgery series (or CSPAN) right when my Myth TV went to update it.
The data can't be peer-reviewed, as there no way to verify it until it happens. It needs to come from an authoritative source. Especially when NBC starts shifting or super-sizing their sitcoms...
The SMTP procotol provides for reliable delivery, but MTAs which fail to adhere to the standard can lose email.
"Reliable" in the land of SMTP means that the sender will get notified if the message can't be delivered. Since it's so easy to spoof any sender's address, the majority of MTA's MUST break this standard or you'd get flooded under a pile of "you tried to deliver a message to aaaaa@somedomain.com and it failed" message everytime someone spoofs your email and starts a dictionary attack. Without reliable authentication, you cannot have reliable delivery.
This is a moderately hard problem to solve, but it's more an issue with most users being unable or unwilling to use things like PGP/GnuPGP or S/MIME.
No, that's a client side answer to a protocol-based problem. The servers that talk to each other need to be able to verify the identity of the sender, not the end users. SPF and Email caller ID are attempts to implement this, but because they will most likely never get enough of a foothold to deny email that doesn't contain such a verification layer, they are essentially useless.
This problem (in SMTP anyway) initiates from the fact that end users can easily pretend to be servers. Clients should only be allowed to talk to the servers that are authoritative for their email domain. Receiving servers should only accept email from servers that are authoritative for the domain of the sender. If you try to implement this is SMTP, it breaks many many things (forwarding, send on behalf of, smarthosts) but all those things are the reason that spam exists today. Imagine if spammers had to register a new domain for every campaign they sent out?
The overwhelming majority of email is delivered in under 10 minutes, but if someone's mailserver or Internet connection is down, then mail ain't going to be delivered until the problem gets rectified. I just checked my mailserver logs and out of the 500 emails delivered today, the average delay was about 5 seconds, and the longest delay was 111 seconds.
Just because that happens at the moment, doesn't mean that's how it's supposed to happen. The only deliverability guidelines in the SMTP RFC are that it should make it there within 3-4 days. The only methods provided to ensure it does rely on email flowing backwards to the sender, who never gets authenticated.
This is the #1 reason why SMTP is a broken protocol. The only people who know and understand how unreliable it is as a service are the people administering it, not the majority of the end user population.
It NEEDS to be torn down and rebuilt from the ground to provide (among other things):
* reliable delivery (Is it there yet?)
* reliable authentication (Is the sender whom they claim to be?)
* timely delivery (Preferably under 10 minutes to any destination, or the original sender is notified, preferably NOT via email)
These things won't happen until the system breaks so badly that there's no alternative but to rebuild it. On the plus side, GoodMail seems as good a way to break the current system as any.
Typically printers that don't work in OS X are a lot like WinModems... They use a simple raster engine and do advanced formatting in the (Windows-only) driver.
Just as typically, these types of printers tend to work poorly under Linux as well. I had this issue with a cheap Lexmark laser printer that was advertised as Windows-only. Bought it before the MacBook it was going to be hooked up to.
I seem to remember them doing a win32 port of XBMC a while back that was less than stellar.
Also, I'm not a coder, but SDL doesn't appear to grant any interface to IR Remotes, which is one of the main reasons I still haven't ditched my xbox in favor of streaming video to my PS3. I still can't believe they didn't put an IR receiver in that thing.
As a longtime XBMC user: What XBMC brings to the table is an excellent 15-foot interface, seemless integration with a LARGE number of formats (including SMB shares and many types of streaming media) and a very active skinning / modding community. The fact that it's currently limited to the original xBox hardware is it's only weak point, in my mind. This keeps it from being able to play 720p content very well, simply because the hardware lacks the power.
A typical XBMC installation package (which is generally illegal, due to licensing restrictions on the compiler) runs between 25 and 50MB, so it's not that big.
I think the biggest challenge they will face with the linux port is hardware support. The xbox platform allowed them to focus more on the interface than trying to support a ton of different ir controllers / video cards / etc.
I wish them the best of luck, as I'd love to go get a MacMini and run XBMC (LBMC? LMC? XBMC4L?) on it.