--- Mark <[EMAIL PROTECTED]> wrote: > When I tryed to have a loginfo script read the > writers file, it seems that the > administrative database isn't really rebuilt till > after the loginfo scripts > have completed runing. The writers file read by the > loginfo script always sees > a writers file without the changes I just commitied. > (The loginfo script was > reading the actual writers file in the repository). > I had to have it use the > workarea copy to read the accurate contents. Is this > the correct operation of > the admin files in CVSROOT on commits?
Since loginfo scripts (or at least loginfo itself) are CVSROOT files themselves, it makes sense that CVSROOT files are updated after loginfo scripts execute. > Is this the same with all other files in CVS? One > thing that has cause issues > for me is that it seems cvs maintains locks after a > commit while the loginfo > scripts are running (I had found out about these > lock issues trying to add to > the commited file's log message in a loginfo > script). I had to fork the script > and exit the parent to get cvs to release the lock, > though I am not sure what > is really happening or what I really did to get > things working. Can anyone > elaborate on how cvs deals with locks after a commit > and into the running of > loginfo scripts? If you do this, you introduce a race condition. For example, more than one loginfo script may now be executing. If you need to use CVS commands within loginfo, try either using RCS commands, or create a temporary module who's archive files are symlinked to the real module's archive files. Do not fork unless you're positive race conditions won't affect you. Noel __________________________________________________ Do You Yahoo!? Yahoo! Movies - coverage of the 74th Academy Awards® http://movies.yahoo.com/ _______________________________________________ Info-cvs mailing list [EMAIL PROTECTED] http://mail.gnu.org/mailman/listinfo/info-cvs