Doesn't the MultiProcessing package permit you to work around the GIL issue by using processes instead of threads? I maintain a Python application that does just that.
It makes communication between parts of the program difficult, but otherwise seems to work fine. And it's part of 2.6 - not a non-standard, native module.
I haven't decided. The intent is actually to map network traffic (i.e., one part captures traffic, stores the packets locally, and sends summary information to a server where another part of the app can access it to develop interactive maps and charts to evaluate the information). It currently has no concept of routers, switches, etc. but is focused on the traffic itself.
I have a SourceForge page for it (search Google for Flower NFA), but haven't updated the code there in many months. The code I have locally works well to generate flow maps and volume histograms.
It's a bit fragile, and still needs a lot of work, but it's coming together. If you'd like to try it out, e-mail me at email@justinthomas.name. You'll need a Debian box for the capture portion (it's written in Python) and a Glassfish v3 server for the Analysis and Visualization components (Java). I run all portions virtualized in two servers (Debian and Windows 2008) on a single XenServer.
I've tested with an application that I'm developing that generates complex SVG network maps (that validate as SVG 1.1 with the W3C validator with no errors).
Linear gradients don't work at all, stroke and fill colors appear to be sporadic. JavaScript doesn't work (but I didn't expect it to as it's targeted to Chrome and Safari primarily right now).
I expect that MS will add more functionality as the preview progresses. They have a lot of work to do, regardless.
You're doing something very wrong. I use an iMac and a Macbook Pro (both far weaker than the machine you're using) and routinely copy around gigabytes of data without the types of delays you're talking about.
I just did a quick test with dd and copy:
imac:~ justin$ date Mon Feb 1 11:46:09 PST 2010 imac:~ justin$ dd if=/dev/zero of=./test bs=2048k count=20 20+0 records in 20+0 records out 41943040 bytes transferred in 0.562810 secs (74524341 bytes/sec) imac:~ justin$ date Mon Feb 1 11:46:19 PST 2010 imac:~ justin$ sudo cp test / Password: imac:~ justin$ date Mon Feb 1 11:46:39 PST 2010 imac:~ justin$ ls -lah/test -rw-r--r-- 1 root admin 40M Feb 1 11:46/test imac:~ justin$
Even with pausing to think about what I was typing and to enter my sudo password, it took 30 seconds to create a 40MiB file and move it to another folder. Honestly, it was probably more like 10 seconds, considering that most of that time was me being slow and mis-typing.
Some folks prefer to separate the tasks - rip on Windows, convert on Linux.
I haven't looked at megui in a while, but most windows apps I've seen integrate different utilities by passing command line parameters. As I understand it, HandBrake tries to avoid that by linking the libraries directly. As a result, the HB package comes with all of the necessary parts (for what it does) and compiles those dependent tools into the overall package.
Different strokes, I guess. My experience with other Windows utilities has always involved installing dubious codec packages and frontends that may or may not actually work with the versions of other tools that you have to install separately. I gave up on that nonsense a long time ago, though, so I suppose it could be better now.
eac3to breaks open the EVO and splits in to elementary streams. Tsremux allows you to recombine the streams into a m2ts file. Handbrake converts the intermediate m2ts to an h.264/ac3 file in an mp4 container.
Tsremux can also do some of this conversion without eac3to.
AnyDVD HD handles the HDDVD and Blu-Ray decryption.
Despite the Troll rating on my above comment, I'm sticking by my statement. H.264 blows DivX away. I wasted too many years on DivX, XviD, etc. Once you move away from it (for encoding and playback) you absolutely will not miss it.
Heck, h.264 is what's used on many Blu-Ray disks (VC-1 on some).
From my perspective, I use it solely on Linux from the CLI. My main computer is a Mac, but I have more power in my Linux server (four cores, 8Gb RAM), I've used other programs, but none handle the task as simply as HandBrake. The developers can be a little snarky, but you take the good with the bad.
I do use it for Apple products (mainly Apple TV). But having used both DivX (in my pre-Mac days and for a bit into them) and h.264, I can't imagine switching back to DivX.
True enough. I've never tried using x264 and ffmpeg by themselves. HB makes using a bunch of different programs to accomplish a single task pretty easy.
HandBrake is the de-facto standard for creating h.264 files on Mac, Linux and Windows systems. You should get to know it; you won't miss that crappy, proprietary DivX.
Chrome does a really fantastic job with SVG JavaScript DOM manipulation; it's crazy fast. Safari 4 does as well.
Firefox is okay, but slowish and the latest 3.5 build breaks some things with SVG transforms that worked fine in 3.0
I'm not sure that you can categorize it as not being "fast enough". The progress in WebKit based browsers is really quite remarkable. I've been writing an SVG application for the last year or so that relies on the ability to manipulate a dynamically generated network map with JavaScript (move it around, scale, etc., found at http://sourceforge.net/projects/flower-nfa/) Chrome and Safari do a great job. Firefox did in 3.0, but is badly broken in 3.5.
Also, check out Google's SVGWeb project. It looks like it might be promising if MS decides to never build support for SVG.
I have a dual dual-core Xeon system built on a Tyan Tempest (i5000VF) motherboard with 8GB of RAM that runs XenServer 5. Right now I have it running 2 Windows 2008 domain controllers, an XP instance, an OpenSolaris instance, and several Linux VMs.
From NewEgg, that RAM cost me about $160 total. The 5 500GB drives (at the time I bought them) were $150 a piece. The processors were $150/each and the motherboard was $340. I picked up a 3Ware 9550SX PCI-Express RAID controller from E-bay for about $200.
It is server-class hardware, but can be built in stages (e.g., start with one processor, 4GB of RAM and 1 drive). I'd recommend not skimping; you'll appreciate the stability in the long term. I've been using this setup for about 2 years and am just now looking at starting again with new hardware (I'd like to build a shared-storage setup with OpenSolaris and ZFS).
You should give it a try again. I've been using XenServer for the last couple of years on dual socket, dual core, old-ish Xeon 5030s (my personal server, I purchased them for about $150 a piece). I run Windows Server 2003, Windows Server 2008, Windows XP, Windows 7, Debian, Ubuntu, and OpenSolaris and have not had a single crash associated with XenServer. The PV drivers for Windows makes performance a non-issue. All of the *NIX platforms have PV kernels available.
I think your information is quite outdated. For me, it just works.
Incidentally, I chose XenServer over VMWare because it worked with my RAID controller (a 3ware device that ESXi does not - and will never, according to VMware - support).
And even if I wanted to move to a mountain top to avoid the "perks" of society, if that mountain top was within the borders of the USA I would still have to pay for the wasteful programs that our federal government throws around.
That is the epitome of the difficulty with increased socialization at a federal level. Folks who do not want and will not use the provided services still have to foot the bill.
"if you think otherwise, youre just a self centered, spoiled individual. just go to a mountain top, and live there, where you will not have to spend a dime for nobody else's interests, but also be away from the perks society brings."
No. I'm an American.
We aren't all self-centered and spoiled, but we do value our ability to make our own decisions and many of us resent any efforts by others to impose their will upon us.
Government is EVERYONE'S corporation. It is a corporation in which my interests may differ from the will of the majority. It is a corporation that can PUT ME IN PRISON for choosing not to participate.
I can choose to walk away from a private corporation; I can't easily escape my government.
A government program that forces its benefits and, more importantly, its costs on all people, regardless of the wants and needs of each individual (social security, public education, etc.) must represent the needs of individuals with sometimes radically divergent agendas. I have not yet seen a "participation mandatory" government program (in the US or elsewhere) that is able to represent those differing needs in anything but a mediocre manner.
In general, if private enterprise can provide a resource, private enterprise should. This empowers the individual to make his or her own decisions.
If the government simply wants to compete and will not force the costs of running broadband out to every remote area in the US on me, then I have no problem with it. If the opposite is true, then this plan is a travesty of freedom.
I set something up for myself for purposes similar to yours (although I'm not a PM). I've been doing a lot of Java and C/C++ development over the last year or so.
I use XenServer on a quad-core (2 dual-core) Xeon box with 4GB RAM and 1.5TB of disk space. I use a 3ware RAID card for the 5 disks (2 mirrored, 3 RAID 5). I run 2 Windows 2003 Servers, 2 Debian instances and a Windows XP Client (not always at the same time) on this one machine. My IDE runs on a Mac and the CLI.
I also rent a VM from linode.com. I have an IPSec VPN I use between my house and Linode.
So at home, I have 2 machines hosting 7 operating systems (the Mac also has a Windows VM I run on occasion). I run multiple databases (PostgreSQL) with no trouble - it's not the fastest, but it works for development. The flexibility you get by being able to copy/create/backup/restore VMs at your leisure definitely outweighs any performance hit.
Doesn't the MultiProcessing package permit you to work around the GIL issue by using processes instead of threads? I maintain a Python application that does just that.
It makes communication between parts of the program difficult, but otherwise seems to work fine. And it's part of 2.6 -
not a non-standard, native module.
I haven't decided. The intent is actually to map network traffic (i.e., one part captures traffic, stores the packets locally, and sends summary information to a server where another part of the app can access it to develop interactive maps and charts to evaluate the information). It currently has no concept of routers, switches, etc. but is focused on the traffic itself.
I have a SourceForge page for it (search Google for Flower NFA), but haven't updated the code there in many months. The code I have locally works well to generate flow maps and volume histograms.
It's a bit fragile, and still needs a lot of work, but it's coming together. If you'd like to try it out, e-mail me at email@justinthomas.name. You'll need a Debian box for the capture portion (it's written in Python) and a Glassfish v3 server for the Analysis and Visualization components (Java). I run all portions virtualized in two servers (Debian and Windows 2008) on a single XenServer.
I've tested with an application that I'm developing that generates complex SVG network maps (that validate as SVG 1.1 with the W3C validator with no errors).
Linear gradients don't work at all, stroke and fill colors appear to be sporadic. JavaScript doesn't work (but I didn't expect it to as it's targeted to Chrome and Safari primarily right now).
I expect that MS will add more functionality as the preview progresses. They have a lot of work to do, regardless.
Ah. My bad.
You're doing something very wrong. I use an iMac and a Macbook Pro (both far weaker than the machine you're using) and routinely copy around gigabytes of data without the types of delays you're talking about.
I just did a quick test with dd and copy:
imac:~ justin$ date /test /test
Mon Feb 1 11:46:09 PST 2010
imac:~ justin$ dd if=/dev/zero of=./test bs=2048k count=20
20+0 records in
20+0 records out
41943040 bytes transferred in 0.562810 secs (74524341 bytes/sec)
imac:~ justin$ date
Mon Feb 1 11:46:19 PST 2010
imac:~ justin$ sudo cp test /
Password:
imac:~ justin$ date
Mon Feb 1 11:46:39 PST 2010
imac:~ justin$ ls -lah
-rw-r--r-- 1 root admin 40M Feb 1 11:46
imac:~ justin$
Even with pausing to think about what I was typing and to enter my sudo password, it took 30 seconds to create a 40MiB file and move it to another folder. Honestly, it was probably more like 10 seconds, considering that most of that time was me being slow and mis-typing.
Some folks prefer to separate the tasks - rip on Windows, convert on Linux.
I haven't looked at megui in a while, but most windows apps I've seen integrate different utilities by passing command line parameters. As I understand it, HandBrake tries to avoid that by linking the libraries directly. As a result, the HB package comes with all of the necessary parts (for what it does) and compiles those dependent tools into the overall package.
Different strokes, I guess. My experience with other Windows utilities has always involved installing dubious codec packages and frontends that may or may not actually work with the versions of other tools that you have to install separately. I gave up on that nonsense a long time ago, though, so I suppose it could be better now.
eac3to breaks open the EVO and splits in to elementary streams. Tsremux allows you to recombine the streams into a m2ts file. Handbrake converts the intermediate m2ts to an h.264/ac3 file in an mp4 container.
Tsremux can also do some of this conversion without eac3to.
AnyDVD HD handles the HDDVD and Blu-Ray decryption.
Despite the Troll rating on my above comment, I'm sticking by my statement. H.264 blows DivX away. I wasted too many years on DivX, XviD, etc. Once you move away from it (for encoding and playback) you absolutely will not miss it.
Heck, h.264 is what's used on many Blu-Ray disks (VC-1 on some).
Maybe. It'll also convert from m2ts files (e.g., files remuxed from a Blu-Ray or HDDVD disk).
That's the extent of my knowledge as it's all I've seen it used for, but it could be more flexible than I realize.
Urk. I replied to the wrong comment - see my post to the grandparent.
From my perspective, I use it solely on Linux from the CLI. My main computer is a Mac, but I have more power in my Linux server (four cores, 8Gb RAM), I've used other programs, but none handle the task as simply as HandBrake. The developers can be a little snarky, but you take the good with the bad.
I do use it for Apple products (mainly Apple TV). But having used both DivX (in my pre-Mac days and for a bit into them) and h.264, I can't imagine switching back to DivX.
True enough. I've never tried using x264 and ffmpeg by themselves. HB makes using a bunch of different programs to accomplish a single task pretty easy.
HandBrake is the de-facto standard for creating h.264 files on Mac, Linux and Windows systems. You should get to know it; you won't miss that crappy, proprietary DivX.
and also for a talented developer (such as myself)
And humble, to boot!
"Or how about showing me an image that I picked out but will soon ignore after seeing that it never changes?"
That's kind of the point. If it changes, then that's an indication that someone might be trying to scam you.
Regardless, it's a weak mechanism.
You keep pointing to Safari's and Chrome's support of Canvas as a unique factor that points to greater adoption for Canvas.
Both browsers also support SVG quite well (as well as Firefox - at least up to 3.0); I'm not sure why you're trying to draw that distinction.
Chrome does a really fantastic job with SVG JavaScript DOM manipulation; it's crazy fast. Safari 4 does as well.
Firefox is okay, but slowish and the latest 3.5 build breaks some things with SVG transforms that worked fine in 3.0
I'm not sure that you can categorize it as not being "fast enough". The progress in WebKit based browsers is really quite remarkable. I've been writing an SVG application for the last year or so that relies on the ability to manipulate a dynamically generated network map with JavaScript (move it around, scale, etc., found at http://sourceforge.net/projects/flower-nfa/) Chrome and Safari do a great job. Firefox did in 3.0, but is badly broken in 3.5.
Also, check out Google's SVGWeb project. It looks like it might be promising if MS decides to never build support for SVG.
I have a dual dual-core Xeon system built on a Tyan Tempest (i5000VF) motherboard with 8GB of RAM that runs XenServer 5. Right now I have it running 2 Windows 2008 domain controllers, an XP instance, an OpenSolaris instance, and several Linux VMs.
From NewEgg, that RAM cost me about $160 total. The 5 500GB drives (at the time I bought them) were $150 a piece. The processors were $150/each and the motherboard was $340. I picked up a 3Ware 9550SX PCI-Express RAID controller from E-bay for about $200.
It is server-class hardware, but can be built in stages (e.g., start with one processor, 4GB of RAM and 1 drive). I'd recommend not skimping; you'll appreciate the stability in the long term. I've been using this setup for about 2 years and am just now looking at starting again with new hardware (I'd like to build a shared-storage setup with OpenSolaris and ZFS).
You should give it a try again. I've been using XenServer for the last couple of years on dual socket, dual core, old-ish Xeon 5030s (my personal server, I purchased them for about $150 a piece). I run Windows Server 2003, Windows Server 2008, Windows XP, Windows 7, Debian, Ubuntu, and OpenSolaris and have not had a single crash associated with XenServer. The PV drivers for Windows makes performance a non-issue. All of the *NIX platforms have PV kernels available.
I think your information is quite outdated. For me, it just works.
Incidentally, I chose XenServer over VMWare because it worked with my RAID controller (a 3ware device that ESXi does not - and will never, according to VMware - support).
It'll be interesting to see how human trials go. The last time I saw cytokines referenced, it was in relation to this drug:
http://en.wikipedia.org/wiki/TGN1412
Looked great in animal studies; not so great for the humans involved.
And even if I wanted to move to a mountain top to avoid the "perks" of society, if that mountain top was within the borders of the USA I would still have to pay for the wasteful programs that our federal government throws around.
That is the epitome of the difficulty with increased socialization at a federal level. Folks who do not want and will not use the provided services still have to foot the bill.
"if you think otherwise, youre just a self centered, spoiled individual. just go to a mountain top, and live there, where you will not have to spend a dime for nobody else's interests, but also be away from the perks society brings."
No. I'm an American.
We aren't all self-centered and spoiled, but we do value our ability to make our own decisions and many of us resent any efforts by others to impose their will upon us.
No.
Government is EVERYONE'S corporation. It is a corporation in which my interests may differ from the will of the majority. It is a corporation that can PUT ME IN PRISON for choosing not to participate.
I can choose to walk away from a private corporation; I can't easily escape my government.
Bask in your ignorance.
A government program that forces its benefits and, more importantly, its costs on all people, regardless of the wants and needs of each individual (social security, public education, etc.) must represent the needs of individuals with sometimes radically divergent agendas. I have not yet seen a "participation mandatory" government program (in the US or elsewhere) that is able to represent those differing needs in anything but a mediocre manner.
In general, if private enterprise can provide a resource, private enterprise should. This empowers the individual to make his or her own decisions.
If the government simply wants to compete and will not force the costs of running broadband out to every remote area in the US on me, then I have no problem with it. If the opposite is true, then this plan is a travesty of freedom.
I set something up for myself for purposes similar to yours (although I'm not a PM). I've been doing a lot of Java and C/C++ development over the last year or so.
I use XenServer on a quad-core (2 dual-core) Xeon box with 4GB RAM and 1.5TB of disk space. I use a 3ware RAID card for the 5 disks (2 mirrored, 3 RAID 5). I run 2 Windows 2003 Servers, 2 Debian instances and a Windows XP Client (not always at the same time) on this one machine. My IDE runs on a Mac and the CLI.
I also rent a VM from linode.com. I have an IPSec VPN I use between my house and Linode.
So at home, I have 2 machines hosting 7 operating systems (the Mac also has a Windows VM I run on occasion). I run multiple databases (PostgreSQL) with no trouble - it's not the fastest, but it works for development. The flexibility you get by being able to copy/create/backup/restore VMs at your leisure definitely outweighs any performance hit.
I'm definitely an advocate of the VM route.