Posts

Ohloh General Discussion : Kudos rank calculation flawed?

I agree with Max that there are some significant flaws in the Kudos algorithms. The flaws are perhaps in the assumptions that are being made about what is an important measure. In my opinion, kudos ... [More] is effectively something that is earned by work performed. It shouldn't matter how many projects a developer contributes to but the quantity and quality of those contributions. For projects, every active contributor to the project is a user of the project. To suggest that a project with with 250 lifetime contributors and 40 active contributors but only a dozen "I use it" votes should have less Kudos than a project with a single contributor and 50 "I use it" votes simply makes no sense to me. Someone that is willing to contribute code, documentation, etc. to a project contributes more love to that project than a user. When project kudos is shared by the contributors, it should be shared based upon the relative contributions to the project and not simply 1/n. Number of commits is one measure of contribution but those commits should be weighted by the complexity of the commits. Length of active participation in the project should also matter. A project's kudos should also be based upon code base size and quality, number of contributions, age of project, consistency of contributions, etc. Just some random thoughts. I hope they are helpful. [Less]

Views
Author
Posted
7678
over 6 years ago

Suggestions for Ohloh 2.0 : Sort projects by lines of code

On the https://www.ohloh.net/p page please add the ability to sort projects by lines of code.

Views
Author
Posted
672
over 6 years ago

Technical Issue Help : git repository commits are sorted/grouped by the author date not the commit date

A git repository maintains two dates for each commit: The AuthorDate which is displayed in standard "git log" output The CommitDate which is displayed in "git log --pretty=fuller" output Here is ... [More] an example from the OpenAFS project: commit f199ac666195771a02e3ebb040c6e5fe47c58c58 Author: Andrew Deason AuthorDate: Wed Nov 2 16:55:49 2011 -0500 Commit: Derrick Brashear CommitDate: Sat Nov 12 06:55:52 2011 -0800 afs: Do not use separate array for srvAddrs This use of the AuthorDate instead of the CommitDate results in two problems: The commit list displayed by Ohloh is not ordered the same as "git log" output which results in confusion since the most recent commits will not always be displayed on the top of the list. Monthly totals, graphs, statistics, etc are incorrect. The AuthorDate is the date that the Author created the first version of the patchset in the her/his local repository. The CommitDate is the date that the patchset was accepted by the project and committed to the upstream repository. Can this please be fixed? Thank you. [Less]

Views
Author
Posted
447
over 6 years ago

Suggestions for Ohloh 2.0 : Permit projects to configure the Cocomo cost per hour

The hourly cost of a developer varies based upon the skill set involved. Please permit projects to configure the default Cocomo hour wage that is used for computing the estimated cost of the project.

Views
Author
Posted
544
over 6 years ago

Suggestions for Ohloh 2.0 : Historical Cocomo analysis

Cocomo analysis provides a point in time analysis of a code base which is intended to represent what it would take to reproduce the code base if coding was started from scratch. It does not provide ... [More] any indication of how much effort was spent over the history of a project. I think it would be helpful for ohloh to develop an estimated costing model which could be used to show to overall investment in project over its lifetime or on a monthly basis. [Less]

Views
Author
Posted
587
over 6 years ago

Suggestions for Ohloh 2.0 : Analysis by directory

Large projects such as OpenAFS are often broken up into subcomponents (processes, libraries, packaging, documentation, etc.) These components often have very different attributes. They may be ... [More] developed with different languages or different licenses. They are often developed by different sub-groups of the project development team. It may also be true that while the project as a whole may be active specific subcomponents may be rotting. I believe it would be quite useful if the Code Analysis, Commits and Estimated Cost tools could be restricted to particular directories or subtrees. Thanks. [Less]

Views
Author
Posted
460
over 6 years ago