Commit uses two-phase locking principles, which means it locks everything
it needs, does the processing, then unlocks.  Larry's change removes two-
phase locking from rtag, which means that rtag can be interrupted by other
write operations.  So rtag won't interfere with commit, but commit can
interfere with a running rtag.

After checking the manual again, I see that the "cvs rtag" command can
take -r and -D options.  It doesn't say that both can be given in the same
command.  Hopefully they can.

It seems to me that "-r HEAD -D now" are reasonable defaults, and using
them gives the user what he wants.  At present, "-r HEAD" is the default,
so supplying the datestamp doesn't seem like a big leap.

--- Forwarded mail from [EMAIL PROTECTED]

Would it be a safe workaround to use something like -D"10 minutes ago " and
assume that something currently being checked in will not be in the tagging
process initiated with a rtag. 
( I am assuming I wait for 10 minutes after a proper file as needed by me
was checked in ).

>From: [EMAIL PROTECTED] (Larry Jones)

>Paul Sander writes:
>> 
>> Another serious issue is when someone commits while an rtag is in
>> progress, and the new data are erroneously tagged.

>Anyone who does an rtag without specifying an explicit revision to tag
>gets exactly what they deserve.

--- End of forwarded message from [EMAIL PROTECTED]



_______________________________________________
Info-cvs mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/info-cvs

Reply via email to