My belief is that the EU and US recession from 2008 which is now 7 years (writing this in 2015) will last for at least another 13 years. The developed world is not changing, there is minimal infrastructure, and it is still cutting back to recover from the recession.
The factories and product development is still moving to Brazil, India, China and Nairobi.
The developed country Japan is a great example of this theory, Japanese lost (two) decades since 1995 of growth
. I would argue that the situation may not change. Tokyo pioneered quantitative easing (pumping money into the economy, to increase liquidity). This down values the currency, and causes inflation. Europe (EU, London) and USA have been doing the same. They can't increase interest rates, because the economy is flat lining.
I would expect London to still be sitting on a low interest rate, 0.5% in 2020, or even in 2030. Likewise the EU and USA too.
Only developing countries will have higher base rates in my opinion.
A lot of countries income is based on export of oil (eg Russia, Oman), they have plenty, so won't run out, but demand fluctuates.
Will my hunch be proved wrong?
POSIX functions such as open() and read() return -1 in case of error, and then set a thread-specific errno global. Makes code checking awkward, better to just return the error code directly.
But what about functions such as opendir() and readdir() which return pointers? They should be modified to take params, and return int!
int opendir_s(const char *name, DIR * dir);
int readdir_s(DIR * dirp, struct dirent * entry);
Note, there is readdir_r which does return int already, but there isn't a matching opendir_s
regular opendir() and readdir() should be depreciated in favour of safer versions.
Signed values considered wasteful
POSIX APIs are quite wasteful.
on a 32bit system "int" in C programming language is a signed twos compliment number. Most POSIX APIs return -1 in case of error, this means that the largest number representable is 2^31 -1. eg lseek off_t (which is also 32bit int). This is wasteful, because there only needs to be one error value, or even allowing up to 1024 error codes would have been fine.
so off_t could have been unsigned: 0 ... 4294967296
4294967296: Error code, check errno
0...4294967295: Normal range.
This isn't as bad as the 7bit email limitaiton, which means 1/8th of everyone's bandwidth is wasted. Its time we had SMTP2 rolled out.. with PGP signing default too. These things will come... eventualy.
So the POSIX standard would need to have:
#define EERROR32 (4294967296)
#define EERROR64 (18446744073709551616)
#define ERROR EERROR32
#define EERROR EERROR64
Of course, 32bit machines still have lseek64, so would need to use EERROR64
What is the best solution?
Of course, the POSIX standard won't be changed, its all too well established. This is a legacy hangover.
I find the fact that PGP email subject lines are not encrypted or signed surprising. Massive loophole!
JCDecaux advert displays running GNU+Linux
London Victoria station
This company twice signed me up to their spam service and deducted cash via EE direct debit.
The only way this fraud can be stopped is if we report to mobile provider and PhonePayPlus regulator.
Mobitrans are now refusing to promptly refund the fraud.
They are hiding their address in The Netherlands on their website and domain name.
Call Mobitrans support center on 02087790598 Mon-Fri 7am-9pm or email supportUK@mobileacademy.com
Time format in latest Android app has been broken for last 6 months.
National Rail can you sort out?