Protocol Design – FAQ

Last change Tuesday, 04-Nov-2008 09:40:29 CET

We really try to keep this page up-to-date, and use the follwing markers:

Item is new
Item has changed lately
Item is around for a while

Questions

Answers

How can I access the machines remotely?

Just use ssh to log into any of the machines of the course. The machines are named: adder, boa, catsnake, cobra, copperhead, keelback, kingsnake, mudsnake, oilsnake, python, seakrait, seasnake, treesnake; the domain is called net.t-labs.tu-berlin.de

An example of a full hostname would be: treesnake.net.t-labs.tu-berlin.de

I want to make my homedirectory readable to my partner, so that we can share code

We are using AFS for our homedirectories. Most notably this means that most of the standard access bits of files are ignored. But you can change permission for entire directories using the fs command. Example: If your parter has the login prak_john then you can give him read access to your home by typing cd $HOME ; fs setacl . prak_john read

What are the requirements to the submissions?

The solutions

must run on the computers of the lab-course. We don't care whether the programs run on your own machine, in the central computing pool or on the hacked computer of the CIA. Only our machines count!
must consist of comprehensible code. We judge whether something is comprehensible or not. The minimal requirements are self-explanatory variable names, in-line comments for the crucial parts of the code and sensible indenting. We will deduct points for not adhering to these standards.
must be submitted at the deadline using the provided submission script. You can submit multiple times, only the last submission will count.
must not contain excessive data. Especially originally provided raw data must not be contained in the submission.
must be explained by both partners in the debriefing. "My partner did this part of the solution" is never a valid excuse.
should contain the source code of your programs, the resulting output of the program, and a textfile explaining how your program works, plus anything required by the assignment. Do not include unprocessed data in the submission unless explictly asked.
must not be copied from another team, even though helping each other with generic information or single commands is OK. We usually find out, when people just copied a solution, that's another reason for the debriefings ...

How do I submit a solution?

Log into one of the lab course machines
Put all files belonging to your solution into one directory
Call labcourse-submit.sh <directory_with_solution> <assignment_number>

Example: Given that all the files of your solution to assignement 1 are in the directory sheet1. In this case you just call labcourse-submit.sh sheet1 1 . In case you forgot something or want to correct your submission, just run the same line again. Only the last submission counts. If you are running out of time before the deadline, you might want to submit at least one solution early enough, and try to squeeze in a better one just before the deadline.

How do these two-people teams work?

You work in teams of two for several reasons. First you should learn how to write code with a partner, and all the benefits that come from having a partner to help you debugging code. This means that just sharing the questions and each partner doing half of the work is not really intended, though possible. Remember to explain to your partner all of the code in any case.
We do not allow teams of three, because experience shows that in such teams almost inevitably two people will do all the work while the third stays back.
Both partners usually get the same tentative score for the submission, but it will be individually adjusted in the debriefing to yield the final score of the assignement.
Usually team partners cannot be changed after the first submission deadline.
It is not nescessary for each partner to submit a solution. Quite the contrary, only the last submission of any partner will be rated for both partners.

What is a file like this: `name.gz`?

It is a compressed file that has been produced using the gzip command. To dump it to standard output use one of these commands:

```
zcat name.gz
```
```
gzip -cd name.gz
```

It is usually not advisible to just uncompress such files to form an uncompressed file on the disc, since they can grow really large.

Perl allows you to directly open such files using

open(DATEIHANDLE, "zcat name.gz|")

which will uncompress the file on the fly.

How to deal with data that is dumped to standard output?

There is no need to write programs all the time. Sometimes it is possible to get important information with just one line of shell code.

You can use so called pipes. A pipe connects the standard output of one program to the standard input of another program. Many Unix tools are able to process data from standard input and return the results on standard output. If you want to view the contents of a compressed file, you can pipe the output of the decompression tool to a pager such as less: zcat file.gz | less



      You can even connect multiple pipes, which can become really
      mighty. One example: 

      
zcat name.gz | grep "http" | sort -n -r | uniq | head >output.txt


      
        
          This reads a compressed file from disc, and outputs the 
	  uncompressed contents to standard output,
          (zcat),
        

        
          passes through only those lines that contain http 
	  (grep),
        

        
          sorts (sort) these lines by their numeric
	  value (in the first column) (-n) in reverse order
	  (-r),
        

        
          removes all duplicate lines
          (uniq),
        

        
          returns only the first few lines of this possibly still 
	  very long list (head)
        

        
          and redirects this output in a file (>output.txt)
        
      

      Note this is still just one line to pass to the shell, which
      solves a sophisticated task without any programming.

      How to process such
      a file using perl?

      
      The following combined perl/shell script returns how often each HTTP
      return value appears in the logfile.
	

      zcat logfile.gz | perl -e 'while(<>){s/\s+/ /g;@l=split;@f=split("/",$l[3]);$p{$f[1]}++;}foreach $i (sort keys %p) {print "$i:$p{$i}\n";}'
  

      Obviously, this one-liner is barely readable (and hence does not comply to
      the submission rules). Thus here again in a commented form:

      
        
          
            zcat logfile.gz

            uncompress logfile to stdout
          

          
            |

            redirect stdout to stdin
          

          
            perl -e

            run perl with the quoted (very long) argument as
	    executeable script
          

          
            while() {}

            while loop
          

          
            <>

            read the next line from stdin to $_
          

          
            while(<>) {}

            iterate over all lines of stdin
          

          
            s/\s+/ /g;

            globally replace multiple whitespaces with only one
	    whitespace
          

          
            @l=split;

            split $_ using space as seperator, and store
	    the individual words in l.
            
          

          
            @f=split ("/", $l[3]);

            split the forth word (using "/") 
	    as seperator
          

          
            $p{$f[1]};

            use the second entry of f as key in the hashmap
            p
          

          
            $p{$f[1]}++;
	    
	    increase the value of the hashmap entry by one
           
          

          
            keys %p

            
              list of all keys of the hashmap p

              
            
          

          
            sort keys %p

            sort the list of keys
          

          
            foreach $i (sort keys %p) {}

            iterate over all keys of the hashtable, in proper order
          

          
            print "$i:$p{$i}\n";

            Print the value of i and the value of the hash entry at i
	    to stdout
          
        
      

      As you see, Perl can be used on the fly. But such one-liners 
      tend to be really obfuscated, hence proper documentation is needed in
      the submission. Some more advanced tricks of how to call perl can be
      found using man perlrun
      What should the server respond, if
	 the requested file exists, but is not a reguar file?
       The server should reply with the code 200 only if the file is a
      regular file, with a 55x else.

`zcat logfile.gz`	uncompress logfile to stdout
`\|`	redirect stdout to stdin
`perl -e`	run perl with the quoted (very long) argument as executeable script
`while() {}`	while loop
<>	read the next line from stdin to `$_`
`while(<>) {}`	iterate over all lines of stdin
`s/\s+/ /g;`	globally replace multiple whitespaces with only one whitespace
`@l=split;`	split `$_` using space as seperator, and store the individual words in `l`.
`@f=split ("/", $l[3]);`	split the forth word (using "`/`") as seperator
`$p{$f[1]};`	use the second entry of f as key in the hashmap `p`
`$p{$f[1]}++;`	increase the value of the hashmap entry by one
`keys %p`	list of all keys of the hashmap `p`
`sort keys %p`	sort the list of keys
`foreach $i (sort keys %p) {}`	iterate over all keys of the hashtable, in proper order
`print "$i:$p{$i}\n";`	Print the value of `i` and the value of the hash entry at `i` to stdout

Protocol Design – FAQ

Questions

General

First Sheet

Second Sheet

Answers

How can I access the machines remotely?

I want to make my homedirectory readable to my partner, so that we can share code

What are the requirements to the submissions?

How do I submit a solution?

How do these two-people teams work?

What is a file like this: `name.gz`?

How to deal with data that is dumped to standard output?

How to process such a file using perl?

What should the server respond, if the requested file exists, but is not a reguar file?

News / Ankündigungen

Navigation