Forum Posts

@fazend, some of what you're looking for is available on the ohloh site. Click on "Tools" and you'll find the capability to compare projects, languages, and repositories for the projects that ohloh ... [More] scans. This is a hard problem however. For instance: o there are no databases that scan every last FOSS project out there. ohloh looks at over 250,000 code repositories, projects, and forums. That is a lot but it isn't all of them! o it is hard to count contributors, since it can be difficult to know that two particular contributor IDs on different projects are actually the same person. o there are many different definitions of "active". Are active projects those with commits? Contributors? Forum posts? Downloads? How many commits constitute "active"? Does that number vary depending on the overall size of the code base and the number of committers? Hard question! o how can you reliably tell when a project forks, and when it just has multiple development threads? You get the idea - you're asking some really interesting and challenging questions! Please check out the Tools page and let us know what you think. As for other sources of info, you might have a look at the statistics published on some of the code forge sites. Anyone else have some suggestions for @fazend's research? [Less]

Ohloh Community, We're about 1/2 way through our first agile sprint on fixes and improvements to the Ohloh site since Black Duck acquired Ohloh 2 weeks ago. This is a short, 2-week sprint to get us ... [More] up to speed on how to operate and make changes to the site, now that we're its owners. We'd like you to know what we're focused on first, and how it is going. First off - we're going through the forums and pulling out the requests for unsticking stuck enlistments, fixing language and license recognition and counting problems, etc., and if these are easily fixed without a lot of research or coding, we're tackling them straight away. You're seeing the results of this in our forum posts in response to your issues, as we fix them. Some of these issues will have to wait until the Ohloh site hardware is physically moved into its new home as a part of Black Duck's IT infrastructure. We're planning to move the site to its new location in the next few months. During the actual move, we'll have a backup version of the site available read-only, for a few days. We'll post something a week in advance so you'll know what is happening. We're planning this process in conjunction with a specialist firm in data center moves, and we expect that the site will be read-only for about 48 hours. As we get closer to the move date, watch for more detailed information. Beyond the immediate fixes and IT move planning, we're looking at the following additional issues for fixing in Sprint 1: JSON API not filtering private fields Adding enlistment yields blank page (areaandvolum project) Allow port numbers in project URLs and links Project has no code detected (qataki project) Contributors whose positions have expired still show as contributors in summary (https://www.ohloh.net/forums/10/topics/5611) UI: paging through projects shows duplicate last and first items 50 projects/user hard limit (https://www.ohloh.net/forums/10/topics/5394) Documentation bug for API (reported on IRC) Support cvs+ssh (https://www.ohloh.net/forums/10/topics/5434) Incorrect license identification: LGPL 2.1 identified as LGPL 3 Change sort order of projects in a developer's experience list Add support for lp: syntax in bzr Issues we're working on for Sprint 1 but that will likely roll over to subsequent sprints: Unique constraint error on people table Visiting commits page results in error (flexible10n project) "I am this person" button yields blank page (https://www.ohloh.net/forums/10/topics/5447) Scrapes entire repository every week (https://www.ohloh.net/forums/10/topics/5490) Comment ratio analysis incorrect for some projects Negative line counts (NumPY project and others) Support a method for excluding directories in code analysis Of these, two merit a little additional explanation: The negative line counts are a symptom of an underlying data model mismatch between Ohloh and distributed SCMs. Ohloh was originally designed before distributed SCMs became common, and has an internal model that matches up to CVS's and SVN's linear commit history. If two branches in a git repository both delete the same large chunk of code, and then those two branches are merged, this data model mismatch can cause negative line counts for the whole project. This same issue is distorting line count statistics for all projects hosted in distributed SCMs, and may also be responsible for the comment ratio analysis bug as well. The fix for this is complex, and may take us awhile to implement. Support for excluding directories in code analysis is a frequently requested enhancement to Ohloh. This might improve many projects' statistics, as project managers would have a better ability to control how Ohloh looks at things. We're evaluating several possible ways to implement this and will have more info on it in a subsequent post. This is also a complex feature to implement, and will likely run into future sprints. As of last night, we've made 46 fixes and improvements to the site since Black Duck acquired Ohloh. We're continuing to put time and effort into addressing the most troublesome problems, while fixing the easy stuff right away. We're committed to being transparent and responsive to your needs - please comment here and elsewhere on the forum as you see issues, have ideas for improvements, or want to comment on future direction. The Ohloh Team [Less]

Andrey, This is something others have asked for in the past - there is a longer thread on it at https://www.ohloh.net/forums/3491/topics/1293 - I'm going there now to reactivate the thread and get more input on this idea. Rich Sands

We hear you on this enhancement! As Robin said in his reply, adding OpenID support is something we'd like to explore. So I'm re-activating this thread to gather more input from you all on how you'd ... [More] want OpenID to work on ohloh. We have a few questions: Should OpenID be the only way to log in to ohloh, or should it be an alternative to standard name/password? Which should be the default? How would you want ohloh to handle people who have multiple OpenIDs that they use on different developer sites? How widespread is that case? More generally - how would you like to see ohloh identities map to identities on SCMs and forges and projects? How well do SCMs support OpenID now? How widely is OpenID supported across developer communities now? Are there other broadly adopted standards for identities used by popular projects/forges? Just making ohloh use OpenID as an alternative to name/password is probably a doable task. Making people's identities across different sites match up, and using that match to improve ohloh's attribution of contributions etc. may be a much bigger task, and the developer ecosystem may not be up to it yet. Do you think that OpenID is adopted widely enough to be the answer? Is this sort of unified ID across developer sites a good idea? Are there privacy implications? Please comment - we're listening! Rich Sands [Less]

Chris, Thank you for taking our poll! The Ohloh Team is currently working on cleaning out the spam in the forums. We will also be looking into the possibility of a more convenient spam reporting mechanism in the near future. Ohloh Team

Everyone, As we said in the announcement, we're keen to get your input on where to focus first in improving Ohloh. We've gone through the suggestions and issues forums and have found several themes ... [More] that keep on popping up. Here is a quick, one question poll on what you'd like to see us focus on right away. Please vote now! http://www.zoomerang.com/Survey/WEB22BAFGLQCCZ We'll be posting this sort of poll frequently. We're sincere about wanting your input! Please comment on the poll here, and post new threads or comment on existing ones in the forums. If we're not asking the right questions, let us know that too! Thanks! The Ohloh Team [Less]

nomeata, Thanks for your question. Alas, the mechanics of adding darcs support did not change with the change of ownership. Robin's process and list of requirements for adding support for a particular ... [More] SCM still likely represent the best outline of how to get it done. That said, we're looking for the "low hanging fruit" for making some substantive updates to ohloh. We're evaluating what would be most valuable and doable. Are there members of the darcs team ready to tackle this? Rich Sands [Less]

Tom, Thanks for the suggestion! Auto-tagging and taxonomy might be a real challenge, but perhaps there might be a way to let project managers add metadata such as tags to their projects to be ... [More] crawled. Have you tried koders.com? It is a free code search site that Black Duck owns, which lets you search the full-text of source code. If you know the kinds of keywords you might find in comments, you can use technology like koders.com to zoom in on what you want. We're interested in how developers might use a mashup of koders.com and ohloh.net technologies to find and select code. Please keep those ideas coming! Rich Sands [Less]

John, we're certainly aware of this issue. Based on our experience, we know that automatic scanning will always produce some mistakes. We have some better license scanning technology, and hope to ... [More] improve the quality of data over time. But we may need to also provide a mechanism so that the community can crowdsource corrections for the inevitable errors. This combination should solve the problem, but probably not immediately. [Less]

Chris, as Black Duck gets started on improvements to ohloh we'll be looking into Ohcount. I'll make sure your request gets noticed. Rich Sands