Main memory of today's computers cannot be considered "private" & "secure" enough as Intel IME and similar garbage can directly read from it.
Assumption that you have total control of memory is false.
also, Google have own OS (android) which reducing Apple smartphone business income, and Google is buying custom build servers (not hyper pricey IBM gear).
from Draft:
The IP addresses and ports can change in the
middle of a connection.
ad/js/css blocking will be much more difficult, as browser probably dont have proper API for plugins to kill those.
blocking will probably need to move to some proxy (which need to be done from scratch)
also browser should be put in separate linux namespace to avoid ability to go-around proxy.
iptables support for this protocol woud be much more complicated as there is now mixing of data/tranasport/security features...
IMHO.. garbage (just to gain few ms in latency)
nope. you are not only one. i'm using fvwm since late '90is:) and dont have plans to change that. i tryed few others and returned back.
power of window manager + no othern clutter. also no eye candy who waste resources.
- parallel read/write to multiple disk in RAID5 in btrfs (speed!, specially when multiple reading/writing userspace processes on btrfs which contains 10+ disks)
- pull back multi-path mgmt at FibreChannel driver level in kernel (WTF userspace need to know that scsi device have multiple path, userspace just get device and use it, rest is kernel work to handle, IMHO)
- throw out interface teaming (bonding is more that good, long time used in production, with VLAN tagging. 0 problems:)
why? i'm quite more affraid of 1000+ ready ICBMs in US/Russian arsenal. with american threat to use nukes i dont see any difference to small country which can produce at most few of them, with limited range and precision.
Slashdot Mirror
simply add possession of HAM radio&License as negative to SocialScore. problem solved :)
Main memory of today's computers cannot be considered "private" & "secure" enough as Intel IME and similar garbage can directly read from it. Assumption that you have total control of memory is false.
also, Google have own OS (android) which reducing Apple smartphone business income, and Google is buying custom build servers (not hyper pricey IBM gear).
from Draft: The IP addresses and ports can change in the middle of a connection. ad/js/css blocking will be much more difficult, as browser probably dont have proper API for plugins to kill those. blocking will probably need to move to some proxy (which need to be done from scratch) also browser should be put in separate linux namespace to avoid ability to go-around proxy. iptables support for this protocol woud be much more complicated as there is now mixing of data/tranasport/security features ...
IMHO .. garbage (just to gain few ms in latency)
nope. you are not only one. i'm using fvwm since late '90is :) and dont have plans to change that. i tryed few others and returned back.
power of window manager + no othern clutter. also no eye candy who waste resources.
- parallel read/write to multiple disk in RAID5 in btrfs (speed!, specially when multiple reading/writing userspace processes on btrfs which contains 10+ disks)
- pull back multi-path mgmt at FibreChannel driver level in kernel (WTF userspace need to know that scsi device have multiple path, userspace just get device and use it, rest is kernel work to handle, IMHO)
- throw out interface teaming (bonding is more that good, long time used in production, with VLAN tagging. 0 problems :)
p.s. you ask for Linux, not GNU Linux. Right? :)
dd if=/dev/zero of=/dev/sda bs=512 count=1 faster, not real damage, much to learn during recover :)
why? i'm quite more affraid of 1000+ ready ICBMs in US/Russian arsenal. with american threat to use nukes i dont see any difference to small country which can produce at most few of them, with limited range and precision.