ISP: Assignment 2

The assignment, except problem 2, is due at the beginning of class on Monday, October 11. Problem 2 is due October 13. Please follow the instructions in submitting your assignment.

1. Create a test program that you can run on an unknown architecture to find out how it handles bit and byte alignment. You should test your program on Linux and Solaris (SPARC), at least.

   The test program should be able to find out:

   • Is the machine big endian or little endian?
   • How big are the C data types char, short, int, long and long long?
   • How is the following structure laid out in memory:

     struct {
       int a:1;
       int b:9;
       int c:6;
       short d;
       short e;
       int f; 
     }
     

     To present your solution, you may have the program generate ASCII graphics such as

     msb
     |...;...|...;...|
     eeeee___ffff
     

     or indicate the byte and bit offset numerically (e.g., "a: byte 3, bit 7"). For uniformity, number bits by their power-of-two-exponent, i.e., the least significant bit has the number 0, and the most significant bit in a byte has the number 7. Note that the & operator cannot be used for bit-fields, so you have to find another way of finding their location in memory.
   • Does the space allocation differ for bit fields on char, short, int or long? I.e., is

     int x:4;
     int y:4;
     

     different from

     char x:4;
     char y:4;
     

   • Do big-endian and little-endian machines order fields in structs the same way, e.g.,:

     union {
       struct {
         unsigned int x:10;
         unsigned int y:6;  
       } s;
       unsigned short z;
     } u;
     

     Does x contain the least significant or the most significant bits of z?
2. In this problem, you will explore various aspects of character sets and internationalization (I18N). Among the platforms used in class, only Solaris 2.7 currently supports this functionality.
   • Using the iconv() function (not the /usr/bin/iconv program), convert text passed as a command-line argument between ISO 8859-1 and UCS-4. UCS-4 is the full 32-bit 'universal character code' or ISO 10646. You can find tables for UCS-4 at unicode.org.

     Hint: you may need to do the conversion in two steps. Available conversions can be deduced from /usr/lib/iconv/.

     Print the output to standard out using wprintf(). Hint: ANSI C allows "wide" character and string constants using the

     L"some string"
     L'x'
     

     notation.
   • Using the strfmon function, display 1234.567 currency units as british pounds, as U.S. dollars and German Marks, using both the national and international currency symbols. Your output should look something like

     USD 1,234.57 = $1,234.57
     GBP 1,234.57 = £1,234.57
     DEM 1.234,57 = DM1,234.57
     

     (Note the rounding taking place.)

   You can use the sample 8859-1 and UTF-8 files for testing.

   You can also convert between character sets using the iconv utility, e.g.,

   iconv -f 8859-1 -t UTF-8 8859.txt > utf8.txt
   

   converts from an ISO 8859-1 to a UTF-8 file. (Note: If you've already done the collation routine, you do not have to do the modified version. However, collation doesn't seem to be working yet on Solaris.)
3. Write a test program that checks if your operating system is fully Y2K-compliant. To determine this, check whether it handles the leap year computation for the year 2000 correctly. Use the Unix time and date routines to compute the weekday for February 29, 2000. Check what happens if you try to compute the weekday for February 29, 1999.
4. Compute file system statistics for your home directory and recursively the files in the directory tree below that. Compute the following statistics:
   • size of file; show as a histogram (table or graph), with logarithmically scaled bin sizes (i.e., bin boundaries of 0 bytes, 4 bytes, 16 bytes, 64 bytes, 256 bytes, 1024 bytes, ...) indicating the frequency of files that fall within a given range of sizes;
   • distribution of the age of files (log. histogram, i.e., younger than 1 hour, 1 day, 1 week, 1 month or 1 year) since modification, access or last i-node change;
   • frequency of different types of file (plain file, block or character special, directory, socket, etc).
   If your CS directory just got setup, it is probably more interesting to tally results for your machine at work or your Cunix account. Sorry, no ftw() allowed.
5. Stevens, Problem 4.15. Write a program that prints out the date when the last email arrived and when the user last read mail. On CS systems, mail boxes are stored in /home/user/.mailspool.
6. Stevens, Problem 4.18.