output inaccuracies

[toomanycats]

Hi,

I'm really grateful for your program. I haven't touched C in a long time and couldn't have
done such a good job from scratch.

However, I've found some issues with the stream_counter() function and have rewritten it using a different logic. I also added what I consider to be good default behavior for unkown file extensions, which is to still count lines but return 0's for everything else.

My rewritten stream_counter() could use some more refactoring, but it is readable. The logic is simple and exhaustive. The verbosity can makes it a little hard to keep track of.

I've never done a pull request before but will figure out how to do that if you're interested.

Cheers !

[velentr]

What kind of output inaccuracies have you found? It hasn't been tested exhaustively, but I haven't noticed anything wrong.

The reason that I skip unknown file types is to avoid wasting time counting binary files etc. When run in a directory with a lot of code that may contain other documentation (PDFs, images, etc), only counting lines in files I care about saves a lot of time. I only care about the languages that I've already defined, so this behavior works for me. Maybe include a command-line switch to toggle this behavior?

I've been meaning to rewrite the stream counter, but it's always worked for me, so I always put it off. Please send your pull request so I can look at what you've written.

[toomanycats]

Thanks for replying...

I'll have to check the "inaccuracies again", it's possible I'm
wrong....As I recall, the problem I found occurred when comments and
code appear in the same line, using the block style.

For instance in C code:

int foo=0; /* initialize a counter here */

I'll work on sending the pull request. You'll probably want to cherry
pick my changes if you decide that they are useful.

Cheers !

  Daniel P Cuneo
  [email protected]

On Sun, Jul 19, 2015, at 10:04 AM, Brian Kubisiak wrote:

What kind of output inaccuracies have you found? It hasn't been tested exhaustively, but I haven't noticed anything wrong.

The reason that I skip unknown file types is to avoid wasting time counting binary files etc. When run in a directory with a lot of code that may contain other documentation (PDFs, images, etc), only counting lines in files I care about saves a lot of time. I only care about the languages that I've already defined, so this behavior works for me. Maybe include a command-line switch to toggle this behavior?

I've been meaning to rewrite the stream counter, but it's always worked for me, so I always put it off. Please send your pull request so I can look at what you've written.

—
Reply to this email directly or view it on GitHub[1].

Links:

1. output inaccuracies #1 (comment)

[toomanycats]

Hi again,

I've done some reading on "forking" with Github. Since I didn't use the
Github fork tool and I just cloned your repo.

I think the easiest thing is for you to add a remote url that points to
my cloned copy on Github. Then, do a git diff and check out what I did.
The more I look at my refactor the less I think you'll be impressed. If
you have the time I'd appreciate trying to follow through since I've
never done this before.

I've since added a "sloc" repo to my Github account from Bitbucket:
https://github.com/toomanycats/sloc.git

Cheers !

  Daniel P Cuneo
  [email protected]

On Sun, Jul 19, 2015, at 12:36 PM, Daniel P Cuneo wrote:

Thanks for replying...

I'll have to check the "inaccuracies again", it's possible I'm wrong....As I recall, the problem I found occurred when comments and code appear in the same line, using the block style.

For instance in C code:

int foo=0; /* initialize a counter here */

I'll work on sending the pull request. You'll probably want to cherry pick my changes if you decide that they are useful.

Cheers !

  Daniel P Cuneo
  [email protected]

On Sun, Jul 19, 2015, at 10:04 AM, Brian Kubisiak wrote:

What kind of output inaccuracies have you found? It hasn't been tested exhaustively, but I haven't noticed anything wrong.

The reason that I skip unknown file types is to avoid wasting time counting binary files etc. When run in a directory with a lot of code that may contain other documentation (PDFs, images, etc), only counting lines in files I care about saves a lot of time. I only care about the languages that I've already defined, so this behavior works for me. Maybe include a command-line switch to toggle this behavior?

I've been meaning to rewrite the stream counter, but it's always worked for me, so I always put it off. Please send your pull request so I can look at what you've written.

—
Reply to this email directly or view it on GitHub[1].

Links:

1. output inaccuracies #1 (comment)