[tahoe-dev] lines of code and test coverage (was: Re: Don't put capabilities in argv?)

[zooko]

Dear Rob Meijer:

On Jul 16, 2008, at 21:41 PM, Rob Meijer wrote:

> Hi, I havn't posted on this list before, I joined the list to not  
> lose track
> of the tahoe-dev part of this discussion, and from an interest in  
> Tahoe as
> it is in its concepts (aparently not its implementation) very much  
> simular
> to my MinorFs project.

Welcome!

> (MinorFs is absolutely dwarved by Tahoe as in codebase, I don't  
> have the
> number of lines that tahoe does as source files, what seems absolutely
> enormous)

That's interesting -- I think of Tahoe as having relatively few lines  
of code for what it does.

HACK Wonwin-McBrootles-Computer:~/playground/allmydata/tahoe/trunk$  
for D in contrib misc mac windows src/allmydata src/allmydata/test ;  
do echo `find $D -name '*.py' -print0 | xargs -0 cat | sort -u | wc - 
l` $D ; done
2518 contrib
1114 misc
1556 mac
210 windows
24884 src/allmydata
8605 src/allmydata/test

HACK Wonwin-McBrootles-Computer:~/playground/allmydata/tahoe/trunk$  
time find src/allmydata -name '*.py' -and -not -path 'src/allmydata/ 
test/*' -print0 | xargs -0 cat | sort -u | wc -l
    16616

Well, 16,000 non-test lines is significant, but not unreasonable  
considering that it implements a functional distributed, secure,  
fault-tolerant file system.  :-)

Here are some numbers that make me feel good -- the code coverage  
statistics:

http://allmydata.org/tahoe-figleaf/current/

Figleaf considers the core to have 14,095 lines of code, of which  
13,195 are executed by the unit tests.

Way to go on writing good tests, Brian!


I'll reply to some of your other questions in a separate message, Rob.

Regards,

Zooko