Community technical support mailing list was retired 2010 and replaced with a professional technical support team. For assistance please contact: Pre-sales Technical support via email to sales@march-hare.com.
On Tue, 13 Jul 2004 13:16:32 -0400, "Chuck Kirschman" <Chuck.Kirschman at bentley.com> wrote: >Were you using a sufficiently large source base? I think it's relevant. >The particular repository I'm working in has just under 21,000 files and >clocks in at 335 Mb. Typically I'm modifying files in the sandbox that >are about 10,000 or so files into the update. I think that the lag is >important; I couldn't reproduce on a trivial 150 file repository. I >couldn't modify the files fast enough after the update started to get it >to occur. It just seems like the client gets lost when files are >changed. If you modify the files *during* the update it'll definately mess things up. Don't do that. >Are there 2 passes though the source tree? The fact that all the files >that aren't in CVS appear at the top of the list makes me think that. >I'm wondering if things changing between these 2 passes is the issue? There aren't two passes. The state of every modified file is sent to the server, which then chooses what to send back to the client and sends it. The client is pretty dumb and just does what the server tells it to - which is why it's important not to touch the sandbox during the update as it'll just wipe any changes that are made in the interim (that's if you're lucky.. ). Tony