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 Thu, 05 May 2005 09:49:13 -0700, Glen Starrett <grstarrett at cox.net> wrote: >I'm not sure cvsmailer is set up for that either, since it uses the CVS >pid to keep track of which mail is which and triggering mult lines might >cause a problem. > >Tony -- Please confirm this is the way it is working now, because it is >a significant change from before. > >Bo -- Will this work with cvsmailer like this? > The email consolidation is done by using the CVSPID to get the process ID and then the start time of that process, from which a temp folder is calculated into which data from all different loginfo calls from that cvs invocation is collected. Once a timeout has passed with no action then the consolidated email is sent out. I pretty much rely on the loginfo call coming only once from each directory (I should have tested with a trigger line plus the ALL directive, but I haven't..). The fuss about the temp folder name calculation is done in order to separate cvs processes that might have gotten the same PID from Windows within the timeout before sending the email out. I soon discovered that Windows quite early will reuse an abandoned PID for a new process, so I decided to get the process start time (in nanoseconds!) from the win api and use that as the identifier instead. I think that the possibility of a false duplicate here is asronomically small. In any case I think this sounds like bad news for those using the loginfo to progressively widen the scope until a match is found (like when using errorlevel in batch files in the old MSDOS days)... /Bo (Bo Berglund, developer in Sweden)