Message ID | 20110223221240.103EEAE1E8@tobiano.tor.corp.google.com |
---|---|
State | New |
Headers | show |
On Wed, Feb 23, 2011 at 2:12 PM, Diego Novillo <dnovillo@google.com> wrote: > > When we merge into the google/* branches, we would like to preserve > the original log messages to facilitate patch archaelogy. However, this > causes problems with bugzilla spam. > > This patch prevents this by removing branches/google/* from the announce > message sent by the svn server. > > OK to commit to hooks? I think you should just use git and move away from using subversion if you want to patch archaelogy. Currently you are adding more than just spam, you are also adding lots of revisions which really are not needed. There are better ways of doing a merge, like creating a new branch and applying your patches one at a time. This will create a better patch archaelogy than the way you are doing it. -- Pinski
On Wed, Feb 23, 2011 at 2:44 PM, Andrew Pinski <pinskia@gmail.com> wrote: > On Wed, Feb 23, 2011 at 2:12 PM, Diego Novillo <dnovillo@google.com> wrote: >> >> When we merge into the google/* branches, we would like to preserve >> the original log messages to facilitate patch archaelogy. However, this >> causes problems with bugzilla spam. >> >> This patch prevents this by removing branches/google/* from the announce >> message sent by the svn server. >> >> OK to commit to hooks? > > I think you should just use git and move away from using subversion if > you want to patch archaelogy. Currently you are adding more than just > spam, you are also adding lots of revisions which really are not > needed. There are better ways of doing a merge, like creating a new > branch and applying your patches one at a time. This will create a > better patch archaelogy than the way you are doing it. > That is what I have been doing with x32 work.
On Wed, 23 Feb 2011, Diego Novillo wrote: > When we merge into the google/* branches, we would like to preserve > the original log messages to facilitate patch archaelogy. However, this > causes problems with bugzilla spam. As previously discussed, we want all commits to all branches to go to gcc-cvs, even if it *might* be desirable for some not to result in Bugzilla messages. So anything that suppresses gcc-cvs mails for some branches is inappropriate.
On Thu, Feb 24, 2011 at 1:33 AM, Joseph S. Myers <joseph@codesourcery.com> wrote: > On Wed, 23 Feb 2011, Diego Novillo wrote: > >> When we merge into the google/* branches, we would like to preserve >> the original log messages to facilitate patch archaelogy. However, this >> causes problems with bugzilla spam. > > As previously discussed, we want all commits to all branches to go to > gcc-cvs, even if it *might* be desirable for some not to result in > Bugzilla messages. So anything that suppresses gcc-cvs mails for some > branches is inappropriate. OTOH we don't necessarily want bugzilla to pick up an non-FSF branch commits (yeah, that implies bugzilla shouldn't normally be used for bugs on random branches). Richard.
On Thu, Feb 24, 2011 at 05:06, Richard Guenther <richard.guenther@gmail.com> wrote: > On Thu, Feb 24, 2011 at 1:33 AM, Joseph S. Myers > <joseph@codesourcery.com> wrote: >> >> As previously discussed, we want all commits to all branches to go to >> gcc-cvs, even if it *might* be desirable for some not to result in >> Bugzilla messages. So anything that suppresses gcc-cvs mails for some >> branches is inappropriate. > > OTOH we don't necessarily want bugzilla to pick up an non-FSF branch > commits (yeah, that implies bugzilla shouldn't normally be used for > bugs on random branches). Yeah. Does anyone know how svnmailer works? Any way to tell it not to trigger bugzilla mail for some branches? I'm fine with gcc-cvs mail, but I really would like to avoid bugzilla mail. Thanks. Diego.
Diego Novillo <dnovillo@google.com> writes: > Yeah. Does anyone know how svnmailer works? Any way to tell it not > to trigger bugzilla mail for some branches? I'm fine with gcc-cvs > mail, but I really would like to avoid bugzilla mail. svnmailer is controlled by //gcc.gnu.org/svn/gcc/hooks/svnmailer.conf. You can set for_paths and exclude_paths in the [gccdefault] section to control whether bugzilla_to_addr is set. If bugzilla_to_addr is not set for some paths, then commits to those paths will not be sent to bugzilla. Docs can be found at http://opensource.perlig.de/svnmailer/ . We are currently running version 1.1.0 on gcc.gnu.org. Ian
On Thu, Feb 24, 2011 at 08:48, Ian Lance Taylor <iant@google.com> wrote: > Diego Novillo <dnovillo@google.com> writes: > >> Yeah. Does anyone know how svnmailer works? Any way to tell it not >> to trigger bugzilla mail for some branches? I'm fine with gcc-cvs >> mail, but I really would like to avoid bugzilla mail. > > svnmailer is controlled by //gcc.gnu.org/svn/gcc/hooks/svnmailer.conf. > You can set for_paths and exclude_paths in the [gccdefault] section to > control whether bugzilla_to_addr is set. If bugzilla_to_addr is not set > for some paths, then commits to those paths will not be sent to > bugzilla. Docs can be found at http://opensource.perlig.de/svnmailer/ . > We are currently running version 1.1.0 on gcc.gnu.org. That's what I thought my patch did. Apparently, it doesn't. I'm not quite sure how to separate the [gccdefault] section. Can you look at my patch and see if I missed anything obvious?
Index: svnmailer.conf =================================================================== --- svnmailer.conf (revision 170444) +++ svnmailer.conf (working copy) @@ -87,3 +87,4 @@ to_addr = java-cvs@gcc.gnu.org to_addr = gcc-cvs@gcc.gnu.org #to_addr = dberlin@dberlin.org bugzilla_to_addr = gcc-bugzilla@gcc.gnu.org +exclude_paths = .*/branches/google/.*