Forums

"Project SeqAn is stuck and not updating"

The project SeqAn is not updating. It is stuck 5 months ago:

http://www.ohloh.net/p/seqan

Manuel Holtgrewe

9 months ago

Manuel,

Rescheduled the hung update which finished OK (948 added commits). Analysis is now at August 16, 2012 on code collected on August 16, 2012 and last commit is at August 16, 2012 - about 1 hour ago.

Please remind me here if the project goes astray again.

Thanks!

ssnow-blackduck

9 months ago

Thank you for the prompt fix.

One more thing: It appears to me that the commit count is incorrect. In Feburary 2011, it reports 1.8k commits.

Also the most current commits were done by "anonymous", according to the statistics.

See here: http://www.ohloh.net/p/seqan/commits?page=200

Manuel Holtgrewe

9 months ago

Manuel,

A great many entries in the subversion (2138) look like this:

r12626 | (no author) | (no date) | 1 line

Since I can't find a web interface to this subversion I can't investigate further.

Commits for February, 2012 were r11162 through r11362 (200)

Commits for February, 2011 were r8700 through r8857 (157)

ssnow-blackduck

9 months ago

Manuel,

I did some additional checking on the ohloh side. See: https://www.ohloh.net/forums/10/topics/7917

ssnow-blackduck

9 months ago

Thank you for your response.

Revision 12626 is a revision in an area that is not publically visible. We have certain parts of the repository that are worked on by our university groups' members which contain yet unpublished projects.

Could this be the problem?

Thanks, Manuel

Manuel Holtgrewe

9 months ago

Manuel,

It's possible if the revisions that look like it to us eventually become public. If so, that could explain the discrepancy between the numbers of "(no author) | (no date) | 1 line" entries now versus when they were collected. Our system will most certainly be confused by this. I'm going to check if the new fetch has finished and what I can see from it.

Thanks!

ssnow-blackduck

9 months ago

Manuel,

Fetch is at 20% but can't tell if this is first or second pass. Once this stage is done, three steps remain (import, line counts and analysis.)

Thanks!

ssnow-blackduck

9 months ago

Manuel,

While we're waiting, it might be a good idea to look at an ignore order to cover these "invisible" projects. If they are limited to areas of the repository that are predictable, you might include a blanket ignore order that covers one or all of them. That could cut down on the confusion. Then later when they become public, they could move to a different place in the hierarchy which would be picked up by ohloh automatically. Just thinking out loud here...

Thanks!

ssnow-blackduck

9 months ago

Manuel,

For example, if the invisible commits were taking place in a directory called "secret" at the root of the repository then:

Disallow: secret/

would maybe serve to make the invisible also not confusing. However, if the existence of this directory's name was important to keep under wraps, you may want to reconsider. We advise that anything you enter in the enlistments area can become public and managers should consider that...

ssnow-blackduck

9 months ago

Hi,

I added the folders with the hidden files to the ignore list.

Also, our Subversion URL is a subdirectory of the repository root. Could this be a problem?

Manuel Holtgrewe

9 months ago

Manuel,

We'll know once the current update finishes. Any ignore orders which refer to incorrect paths are themselves ignored. We'll see how many items are ignored then refine the order as needed.

"Ohloh ignored xxx files during processing."

Thanks!

ssnow-blackduck

9 months ago

Manuel,

I adjusted the ignore order to put the "Disallow: " into the order for all lines. If they aren't seen in the enlistment as it is currently, they will have no effect. I'm trying to be deliberately unclear about the paths but it will be easy to determine if it worked once the analysis is finished. If you want to discuss the specifics I'll be glad to do that via an e-mail.

Thanks!

P.S. The best way to keep this information semi-private is to make the project "Managers Only can edit" as it is now. This is no assurance, though.

ssnow-blackduck

9 months ago

Manuel,

Now 80% - probably first pass. Second pass is faster. Once this stage is done, three steps remain (import, line counts and analysis.)

Thanks!

ssnow-blackduck

9 months ago

Manuel,

The re-fetch is finished now. We retrieved 8715 commits. The weird commit bump is gone. February 2011 is now down to 63 commits from 1,881 in the previous. We only ignored 32 files during processing and there are still 2138 commits by "[anonymous]" in the last 30 days. August 2012 will show 2294 commits when the month is done. The weird commit bump is now just in a different place. It's clear that we need to refine the ignore order and then when we are confident that we have it right, I will do another re-fetch.

I can't help much from here since I can't see the directory structure. Perhaps give it some thought and let me know here when you are ready to try again.

Thanks!