[cvsnt] Does cvs start more than once for multi-module commits?

Bo Berglund bo.berglund at telia.com
Sun Dec 1 17:12:15 GMT 2002


Tony,
I assumed that if the user issues the 'cvs commit' command when he has
several files edited in a number of subdirectories then cvs on the
server would iterate through these and process them one by one.

But it seems like this does not happen, I have entered code in my
mailing program to pick up the PID of the process that launches my
program from the loginfo script and the logs from all executions show
different parent PID:s all the time.

So is it like this?:
cvs on the client runs as one command and scans the sandbox folders.
For each subfolder with changes inside it sends a call to the server
for this submodule and the server thus spawns a new cvs process for
all of them.

Or is there a parent cvs.exe on the server that spawns new threads
with different PID:s for each submodule to process?

Or is there a separte thread created for each processing of the script
files?

Grateful for hints on this.

If this does not work then I have another idea:
According to my limited knowledge cvs on the server creates a
subdirectory inside the temp folder where it keeps intermediate files
during a commit.
The name of this subfolder is unique(?) and if I could get a way to
pick up this name when running the script process then I could use it
as an identifier for all cvs executions that belong together.
I seem to remember that I saw directories named as numbers in the temp
folder earlier on, but these flders disappear so fast that they are
impossible to see using Explorer...

Appreciate a bit of info here. Thanks.



/Bo
(Bo Berglund, developer in Sweden)



More information about the cvsnt mailing list