There is a gold support ticket on this issue already, however since we have been able to reproduce it regularly I thought I'd post as well (here is the existing ticket http://support.sourcegear.com/viewtopic ... st+version)
Our setup is similar to his environment We have in our solution:
43 projects total
10 web projects
31 class libraries
2 windows services
1993 total classes
2031 files
Basically if we do a "get latest version" from the solution level OR sometimes from selecting multiple projects......the IDE hangs for a few moments while it thinks, then the IDE just goes away. No error message, no "crash analysis notification", no logging, etc.
Also, because of the sheer size and scope of this solution, it is not feasible to send you a repro solution. My guess is that since the only people having this problem have massive solutions, that is where the problem lies. A smaller solution example most likely would not be able to repro the problem.
VS.NET 2005 crash during Get Latest Version
Moderator: SourceGear
If there is no logging at all (assuming you have Vault IDE logging working) before the Get Latest, then the crash would appear to be happening on the Visual Studio side. If you haven't gotten IDE logging running, that would be the first step in diagnosing.
I wonder if this is related to RAM availability. Does there seem to be a connection between when this happens and how much RAM is being used by VS 2005 (and how much is available on the machine running)?
We've had at least one other user who had problems with VS2005 chewing up all available RAM and crashing. Vault certainly uses more RAM than say VSS does (because it requires the .Net Framework to be loaded), but we've not had reports that it leaks memory, whereas there does appear to be problems in this area with VS2005.
The customer with these problems said they got a hotfix from MS about the memory leak issue, so if it looks like this might be the issue, there might also be a workaround.
If the RAM issue looks like a dead-end, it actually would be very helpful to get a copy of the full solution. Often just creating large dummy solutions don't have the same complexity that real solutions do, and often don't help in reproducing the problems.
I wonder if this is related to RAM availability. Does there seem to be a connection between when this happens and how much RAM is being used by VS 2005 (and how much is available on the machine running)?
We've had at least one other user who had problems with VS2005 chewing up all available RAM and crashing. Vault certainly uses more RAM than say VSS does (because it requires the .Net Framework to be loaded), but we've not had reports that it leaks memory, whereas there does appear to be problems in this area with VS2005.
The customer with these problems said they got a hotfix from MS about the memory leak issue, so if it looks like this might be the issue, there might also be a workaround.
If the RAM issue looks like a dead-end, it actually would be very helpful to get a copy of the full solution. Often just creating large dummy solutions don't have the same complexity that real solutions do, and often don't help in reproducing the problems.