Resolving "unknown" status

kiwirob · Post by **kiwirob** » Wed Aug 03, 2005 1:06 am

I have just upgraded to 3.0.7

Is there anyway to turn off the automatic downloading of baseline versions when it resolves the "unknown" status while idle?

Its causing me all sorts of grief because vault is trying to download mb of files I don't want. I don't even have time to cloak the folders before it starts or cancel the operation.

lbauer · Post by **lbauer** » Wed Aug 03, 2005 7:35 am

Resolving the status of Unknown files should be a one-time operation. Plus, if the version of a file in the working folder matches the version of the file in the Vault repository, nothing is downloaded. So it shouldn't be a continual performance hit.

Are you running into a problem with Unknown file status on a regular basis or just during the upgrade?

kiwirob · Post by **kiwirob** » Wed Aug 03, 2005 4:49 pm

On a regular basis. Most of our devs are in the UK, i'm the only remote one here in NZ. I'm only an adsl connection and i've only got 1GB bandwidth per month!

We have many developer branches of our main source (approx 200MB), a new branch is created each time a developer does a significant piece of work this can be 2-3 per dev per month.

We have a review process where another developer will code check others work. This involves getting their branch, building, checking code etc. Once the code is checked its merged back into the main dev branch.

Previously when reviewing I would just make a local copy of an existing branch, set my working folder for the branch and just download the files required (from our bug tracking system). I know this isn't strictly correct but its was close enough for code review.

Now however vault starts its auto download everything because its all unknown status.

Yesterday vault got stuck in an endless loop during a get and chewed up 300MB.

I really don't like applications that just use bandwidth when not told too

kiwirob · Post by **kiwirob** » Wed Aug 03, 2005 7:41 pm

Some other nice features would be

1. The ability to do a get on multiple files in the history, currently you have to do it file by file

2. The ability to do a get from the search dialog when doing a filename search.

arden_dan · Post by **arden_dan** » Thu Aug 04, 2005 4:45 am

kiwirob needs to buy more bandwidth

kiwirob · Post by **kiwirob** » Thu Aug 04, 2005 4:49 am

arden_dan needs to do less surfing

lbauer · Post by **lbauer** » Thu Aug 04, 2005 4:40 pm

Its causing me all sorts of grief because vault is trying to download mb of files I don't want. I don't even have time to cloak the folders before it starts or cancel the operation.

I experimented with this a little. If you cloak the folders ahead of time, before you change the working directory, then files aren't downloaded. But as soon as you do any file operation, such as a diff, Vault resolves the Unknown status.

I'm not sure how else we can make this more efficient for you. How did you deal with all the Unknown status files before? Was the status irrelevant?

Some other nice features would be

1. The ability to do a get on multiple files in the history, currently you have to do it file by file

2. The ability to do a get from the search dialog when doing a filename search.

We've had other requests for these. I'll add your "vote."

kiwirob · Post by **kiwirob** » Thu Aug 04, 2005 5:21 pm

Thanks Linda

The unknown status is irrelevant i'm only interested in the files that have been modified and these are listed in our bug tracking system.

If I do a diff on a file will it resolve unknowns of other files or just the file selected?

If I cloak the top level folder before setting the working folder will this stop all the unknown status resolutions of sub-folders?

Ideally an option or registry setting to completely disable it would be nice

lbauer · Post by **lbauer** » Fri Aug 05, 2005 8:20 am

The cloak stops the initial download. However once you do any file operationi, the file list is refreshed and the status of all Unknown files are resolved. So the cloak would only help with directories you don't need to deal with.

I can log a feature request to provide options for dealing with Unknown files. But most users want Unknown file status to be resolved.

I'm wondering if there's a better way to do your comparison than by switching working folders. That's why I asked about your process. I can't quite visualize the steps you're using.

kiwirob · Post by **kiwirob** » Sun Aug 07, 2005 5:59 pm

I'm not switching working folders i'm normally setting it for the first time but the folder already has all the files in it from another branch download.

I think I can get around it by setting the working folder to an empty folder then just downloading the files I need and then copying them into another build folder.

Most files will have Missing status and therefore will not get downloaded, is this correct?

While we are at it, when doing a branch of a large project it always times out. The branch is committed ok, the timeout seems to occur on the folder structure get. The server timeout has been checked and is ok. I then have to undo the branch and the branch remains.

Once the branch has been completed all files have to be downloaded, is there anyway for it to get the files from the original project as immediately after the branch they are of course the same.

Thanks

lbauer · Post by **lbauer** » Mon Aug 08, 2005 7:20 am

I'm not switching working folders i'm normally setting it for the first time but the folder already has all the files in it from another branch download.

I think I can get around it by setting the working folder to an empty folder then just downloading the files I need and then copying them into another build folder.

Most files will have Missing status and therefore will not get downloaded, is this correct?

Yes, it would be better to set the working folder to an empty folder, then do a get. As long as you're not doing a get from the folder level, then you can just get the individual files you need from the file list(s).

I then have to undo the branch and the branch remains.

I'm a little unclear on this -- the branch is completed but the branch action remains in the pending change set?

Once the branch has been completed all files have to be downloaded, is there anyway for it to get the files from the original project as immediately after the branch they are of course the same.

Not really. If you copy the files from the origin directory working folder, then you'll end up with status Unknown again.

It might help to understand a little more about how the client works. To improve performance, Vault relies on cached information to track the status of files and incremental file changes. Much of the cached info is created as a result of file operations using a specific working directory. If the working directory contains files that were not retrieved to the working directory by the Vault Client, then the cache files and file status info are out of sync.

More info here:

http://support.sourcegear.com/viewtopic.php?t=562

GregM · Post by **GregM** » Mon Aug 08, 2005 9:12 am

lbauer wrote:
I then have to undo the branch and the branch remains.
I'm a little unclear on this -- the branch is completed but the branch action remains in the pending change set?

I've seen this too. I've also had local files remain in the pending chance set with status "needs merge" after a checkin because the get of the new baseline couldn't complete.

kiwirob · Post by **kiwirob** » Mon Aug 08, 2005 5:10 pm

I'm a little unclear on this -- the branch is completed but the branch action remains in the pending change set?

Yep, the server logs say the branch has completed. Its the download that fails, so I just undo the branch from the pending change set and normally reconnect to refresh everything.

Not really. If you copy the files from the origin directory working folder, then you'll end up with status Unknown again.

It might help to understand a little more about how the client works. To improve performance, Vault relies on cached information to track the status of files and incremental file changes. Much of the cached info is created as a result of file operations using a specific working directory. If the working directory contains files that were not retrieved to the working directory by the Vault Client, then the cache files and file status info are out of sync.

Just would be nice if in certain cases it could copy files from other branch caches, its just a waste to re-download 200MB when I know the exact same version is already cached in another branch