Forums : Ohloh General Discussion

Dear Open Hub Users,

We’re excited to announce that we will be moving the Open Hub Forum to https://community.synopsys.com/s/black-duck-open-hub. Beginning immediately, users can head over, register, get technical help and discuss issue pertinent to the Open Hub. Registered users can also subscribe to Open Hub announcements here.


On May 1, 2020, we will be freezing https://www.openhub.net/forums and users will not be able to create new discussions. If you have any questions and concerns, please email us at info@openhub.net

git branches not appearing in statistics?

Hello,

It appears that while you can list a git branch in the enlistment page, the results are not included in things like the contributor pages.

You can see this on our (Thousand Parsec http://www.thousandparsec.net/) project pages. For example 'Tyler Shaub' has only ever committed on the 'rfts' branch of tpserver-cpp ( http://git.thousandparsec.net/gitweb/gitweb.cgi?p=tpserver-cpp.git;a=shortlog;h=rfts ) which is enlisted at http://www.ohloh.net/projects/3679/enlistments?page=2&sort=name but he does not show up on the Contributors page ( http://www.ohloh.net/projects/3679/analyses/latest/contributors ).

Thanks for your help!

Tim Ansell

Tim 'mithro... about 15 years ago
 

Hi Tim,

Yes, we had some serious problems here, which I believe are now fixed. I re-ran the rfts branch and Tyler Shaub's commits are now visible.

It will take a while, but I will re-run all of the Thousand Parsec repositories which are not on the master branch, and that should improve the report considerably. Thanks for reporting this problem; I'm glad to have it fixed.

Thanks,
Robin

Robin Luckey about 15 years ago
 

Thanks for your speedy fix!

Might I also suggest auto converting URLs in posts to clickable links?

Tim

Tim 'mithro... about 15 years ago
 

Just checking through the stats and there are a few things more you might want to check.

Donata Małecka has committed to the rewrite branch of our metaserver-lite.
These are not appearing in the stats.

The libtpproto-java repository seems to have failed to download as well, you can see the info below,

git://git.thousandparsec.net/git/libtpproto-java.git

Step 1 of 3: Downloading source code history (Failed)

Lastly, the stats are reporting that our code is 'shell script - 29%', however our own stats 'shell script - 8.33%' (you can find out info http://www.thousandparsec.net/tp/dev/sloc.php ). We use a tool call sloccount.

Tim 'mithro... about 15 years ago
 

It has taken a day or two for the dust to settle after making the changes to our Git branch handling, and updates are still happening. After the weekend, the affected reports should finally start to look right.

I've used sloccount many times to sanity-check our numbers. Once the Thousand Parsec report is fully up-to-date and tracking the right branches, I'll take a few minutes to drill into the report and see what's going on with the line counts. For the languages we process, we're usually very accurate, so there may be something funny going on there.

Robin Luckey about 15 years ago
 

Well the stats seem to be all fine now.

Any info on the shell script issue?

Tim 'mithro... about 15 years ago
 

Hi mithro,

There's a lot going on here, and it would take a very long time to solve this exactly.

First, the list of repositories entered on Ohloh does not match the list of repositories on the link you included. There are many additional branches on the Ohloh report, so the numbers on the Ohloh report are going to be higher.

I ran sloccount locally against all 34 repositories entered on Ohloh (no easy feat), and came up with the following totals:

Totals grouped by language (dominant language first):

 python:       47556 (37.20%)
 cpp:          35426 (27.71%)
 java:         15801 (12.36%)
 ansic:        14060 (11.00%)
 sh:            9282 (7.26%)
 php:           4160 (3.25%)
 perl:           729 (0.57%)
 ruby:           648 (0.51%)
 lisp:           165 (0.13%)

These numbers are much lower than the Ohloh counts:

 python     | 72853 | 0.28870853329423280402
 cncpp      | 71546 | 0.28352903412445856995
 shell      | 54119 | 0.21446772423030740149
 html       | 18926 | 0.07500168422888076056
 xml        |  9452 | 0.03745725030811481289
 php        |  6869 | 0.02722110160457476193
 perl       |  2751 | 0.01090191447287598924
 javascript |  1693 | 0.00670917528265323511
 java       |  1371 | 0.00543312422475935341
 css        |  1116 | 0.00442258689630301853
 ruby       |   720 | 0.00285328186858259260
 scheme     |   527 | 0.00208844381214309209
 sql        |   105 | 0.00041610360583496142

Some of this might be simply due to the extra branches on Ohloh, which lead to double-counting, but I don't think this accounts for an order of magnitude increase. Also, it's curious that Ohloh is not seeing the Java.

Because this project has so many repositories and files, it's kind of a needle-in-a-haystack problem to figure out where we are different from sloccount. In general, Ohloh recognizes many types of files that sloccount does not, so our numbers are almost always higher, but I'm concerned about the missing Java.

It will take a lot of hours of manual counting to get to the bottom of it. If there are bugs in the Ohloh counter we definitely want to fix them, but no promises on how long it will take to get to this.

Robin Luckey about 15 years ago
 

Thanks for working on this.

What I find the most interesting is that sloccount reports 7% for shell (around 9282 lines), while Ohloh counts 21% for shell (around 54119). That is a huge difference which I don't think can be explain just using branches (the increase in C++ and Python are probably right for double counting).

Feel free to send me an email directly if there is something I can do which you think might help you figure out the problem.

Tim 'mithro... about 15 years ago
 

Any update on this?

Tim 'mithro... about 15 years ago