C/C++ Programming Guidelines

BoFox · Posted: Sun May 13, 2012 2:07 pm

jaydip wrote:

First thing that comes to mind that my compiler may be faulty,how on earth can be "-1" greater than "1"?This is one of the very weird rules of C/C++. if u mix an unsigned and signed type the result is of the unsigned type.If u convert "-1" to a positive integer it becomes a huge number which is definitely greater than "1". The third "If...else" shows what is happening behind the scenes by means of a cast.Moral of the story ,never mix an unsigned and signed type.We tend to declare unsigned types for some variables which can't be negative for e.g. Age,but this is a bad choice.We should only declare unsigned types for bitfields (more on this later). Now cppcheck throws a waring when u compile this code,.

Ok, how does converting '-1' to a positive integer make it a huge number??? :lol:

jaydip · Posted: Mon May 14, 2012 12:44 am

BoFox wrote:

jaydip wrote:

First thing that comes to mind that my compiler may be faulty,how on earth can be "-1" greater than "1"?This is one of the very weird rules of C/C++. if u mix an unsigned and signed type the result is of the unsigned type.If u convert "-1" to a positive integer it becomes a huge number which is definitely greater than "1". The third "If...else" shows what is happening behind the scenes by means of a cast.Moral of the story ,never mix an unsigned and signed type.We tend to declare unsigned types for some variables which can't be negative for e.g. Age,but this is a bad choice.We should only declare unsigned types for bitfields (more on this later). Now cppcheck throws a waring when u compile this code,.

Ok, how does converting '-1' to a positive integer make it a huge number??? :lol:

Ok lets find the solution shall we :grin:

For signed numbers the last digit carry the sign bit,"0" for +ve numbers ,"1" for -ve numbers.Now in case of unsigned numbers the last digit is not a sign bit any more,because sign bit has no meaning in case of unsigned numbers.So "-1" which is all "1"s represent a very large +ve integer.Is that clear now?Do u want me to add this into the original article?

grstanford · **Joined:** Sat Apr 24, 2010 11:19 am **Posts:** 5013

Fewer and fewer people get this sort of thing nowadays, because they have never dealt with assembly code and manipulated binary numbers.

C is pretty unforgiving about errors too. It's mostly up to the programmer not to make errors. This is why there are so many software bugs out there.

jaydip · Posted: Mon May 14, 2012 2:38 am

grstanford wrote:

Fewer and fewer people get this sort of thing nowadays, because they have never dealt with assembly code and manipulated binary numbers.

C is pretty unforgiving about errors too. It's mostly up to the programmer not to make errors. This is why there are so many software bugs out there.

I agree completely.Now a days unless u are doing systems programming it is very fool hardy to choose C/C++.There are many other languages which takes the burden away from the programmer and simplifies their job.But we have lots of legacy C/C++ code which needs maintaining unfortunately :grin:

.Also if u are into CUDA C is the pretty obvious choice.

BoFox · Posted: Tue May 15, 2012 12:20 am

jaydip wrote:

BoFox wrote:

jaydip wrote:

First thing that comes to mind that my compiler may be faulty,how on earth can be "-1" greater than "1"?This is one of the very weird rules of C/C++. if u mix an unsigned and signed type the result is of the unsigned type.If u convert "-1" to a positive integer it becomes a huge number which is definitely greater than "1". The third "If...else" shows what is happening behind the scenes by means of a cast.Moral of the story ,never mix an unsigned and signed type.We tend to declare unsigned types for some variables which can't be negative for e.g. Age,but this is a bad choice.We should only declare unsigned types for bitfields (more on this later). Now cppcheck throws a waring when u compile this code,.

Ok, how does converting '-1' to a positive integer make it a huge number??? :lol:

Ok lets find the solution shall we :grin:

For signed numbers the last digit carry the sign bit,"0" for +ve numbers ,"1" for -ve numbers.Now in case of unsigned numbers the last digit is not a sign bit any more,because sign bit has no meaning in case of unsigned numbers.So "-1" which is all "1"s represent a very large +ve integer.Is that clear now?Do u want me to add this into the original article?

Not completely clear for me yet, sorry.. yeah, add it to the original article.. thanks!!

By the way, just a guess.. you're in Mumbai rather than Calcutta/Delhi/Hyderabad/etc..?

jaydip · Posted: Tue May 15, 2012 1:33 am

BoFox wrote:

jaydip wrote:

BoFox wrote:

jaydip wrote:

First thing that comes to mind that my compiler may be faulty,how on earth can be "-1" greater than "1"?This is one of the very weird rules of C/C++. if u mix an unsigned and signed type the result is of the unsigned type.If u convert "-1" to a positive integer it becomes a huge number which is definitely greater than "1". The third "If...else" shows what is happening behind the scenes by means of a cast.Moral of the story ,never mix an unsigned and signed type.We tend to declare unsigned types for some variables which can't be negative for e.g. Age,but this is a bad choice.We should only declare unsigned types for bitfields (more on this later). Now cppcheck throws a waring when u compile this code,.

Ok, how does converting '-1' to a positive integer make it a huge number??? :lol:

Ok lets find the solution shall we :grin:

For signed numbers the last digit carry the sign bit,"0" for +ve numbers ,"1" for -ve numbers.Now in case of unsigned numbers the last digit is not a sign bit any more,because sign bit has no meaning in case of unsigned numbers.So "-1" which is all "1"s represent a very large +ve integer.Is that clear now?Do u want me to add this into the original article?

Not completely clear for me yet, sorry.. yeah, add it to the original article.. thanks!!

By the way, just a guess.. you're in Mumbai rather than Calcutta/Delhi/Hyderabad/etc..?

OK sorry,I should have made it much clearer.I will explain it by using a Nibble(4 bits)
8 4 2 1
--------
0 0 0 0 ---> Decimal 0(Add all the digits under 8 4 2 1)
0 0 0 1 ---> Decimal +1
0 1 1 1 ---> Decimal +7
1 1 1 1 ---> (-7) in signed integer (+15) as an unsigned integer

Now u can see if we have all ones that represent a big +ve integer if we use an unsigned integer type.If u have any questions please let me know and thanks for asking me to clarify :good:

.I am currently @ Bangalore.

BoFox · Posted: Mon May 21, 2012 5:50 am

Bangalore.. ahh, cool! For some reason, I thought Mumbai was the tech hub of India, but I was wrong. From Wiki:

Quote:

Bangalore is well known as a hub for India's information technology sector. It is among the top 10 preferred entrepreneurial locations in the world.[8]

Today as a large city and growing metropolis, Bangalore is home to many well-recognized colleges and research institutions in India. Numerous public sector heavy industries, software companies, aerospace, telecommunications, and defence organisations are located in the city. Bangalore is known as the Silicon Valley of India because of its position as the nation's leading IT exporter.[9][10][11] A demographically diverse city, Bangalore is a major economic and cultural hub and the second fastest growing major metropolis in India.[12]

http://en.wikipedia.org/wiki/Bangalore

jaydip · Posted: Sat Oct 27, 2012 9:24 am

It has been a long time since I updated this and this was my fault for being too lazy

.I was thinking about doing some neat stuff and then I thought why not CUDA?What do you guys think?

apoppin · Posted: Sat Oct 27, 2012 9:27 am

Why not?

jaydip · Posted: Sat Oct 27, 2012 10:22 am

Thanks I will start with setting up the cuda environment and some basic gpu codes first.

C/C++ Programming Guidelines

Who is online