No new commits, still pending alias additions. Can this be fixed, please?
Hi Alexey,
I have rescheduled this project for update.
I had to remove a few repositories which do not appear to exist anymore.
This is a large project with a lot of repositories, so it might take a few days to complete. I'll keep an eye on it.
Thanks, Robin
All repositories that use http:// can safely be removed. Is there any way to automaticaly update repositories for this project?
It's a linux distribution with one repository per source package (about 8000 source packages). We would like to replace all current enlistments with direct links to package repositories and hope that this would not break anything.
@Vitaly,
We have successfully grabbed all of the code, but we are experiencing some trouble with our line counter. There is a new bug lurking in our code that prevents this project's analysis from completing. We are investigating, and should have it fixed within a few days.
@Alexey,
We do not currently have any way to automate the list of repositories for a project. This is something that has come up a few times in the past for very large projects, so it's something we're considering, but there are currently no plans.
However, current enlistments is a big mistake. We have set of repositories (one repo per package) that corresponds to source packages. Listing personal developer sandboxes is not that good.
I have list of git uls in a plain text file, can someone help up update repository list for this project? It will take ages of mouse-clicking.
Hi Alexey,
Don't take it personally, but I am marking all of these new enlistments as "failed" as they are added to our system.
Our server cluster is not large enough to handle this sudden influx of new repositories. You've single-handedly grown our data set by 5% in just a few hours.
All of the new repositories are monopolizing our resources, and preventing updates to other projects.
So I have marked all of the enlistments as failed, and I will reschedule them as our server resources allow. I warn you that this is so much new code that it could literally take months for us to count it all.
You may want to reconsider whether all 50,000 Gimp commits are really part of the ALT Linux project.
This is going to take a long time, please be patient with our little system.
Thanks, Robin
I guess we can live with rescheduling, thanks for your patience.
And what's wrong with tracking upstream sources in our packages? Other projects also use this scheme (for example debian - https://www.ohloh.net/p/debian/enlistments/23868). This is just how we maintain and build packages for our distribution.
I'm not terribly surprised that there is a timeout. We intended the aliases UI to be used with a dozen repositories, not thousands, so the queries aren't optimized.
I'll see if there's something we can do to fix it. However, if the total number of names in all of these repositories is large (and I'm guessing that there are several thousand), then the web UI will probably be unusable anyway, because the drop-down lists of names will be so long.
Something like "contributor filter" (only count commits those author matches patterns) feature would be nice.
Not yet. This will take months, not days.
I'm not even sure our system will scale to hold 2500 repositories in a single project -- there's a chance it may not work without some changes on our end.
Thanks Robin. Just for the record, I've mostly tried to file relevant repos containing commits by ALT Linux team members (both as original development and when they're also upstream developers), so if repolist versionized somehow, it might make sense to roll back with s/http/git/g and postpone full-blown DoS until it's reasonable, with committer domain filters or anything else needed for practical usability of the results (let alone cluster stress).
I'm afraid claiming 389-server as an ALT Linux project is wrong at best. I've hand-picked exactly relevant repos back then, even if s/http/git/g would be nice.
2 raorn: let's create a mask list for your script (e.g. alterator, installer, mkimage*) and ask ohloh folks to hopefully revert this attack or drop everything and let us start from, well, scratch?