Re: Problems with commit feature negotiation and write-through proxies

2018-01-10 Thread Philip Martin
Evgeny Kotkov writes: > I would be fine with this approach as well, but perhaps with a few tweaks: Right, I'll incorporate those changes, test it and commit. -- Philip

Re: Problems with commit feature negotiation and write-through proxies

2018-01-10 Thread Evgeny Kotkov
Evgeny Kotkov writes: >> Index: subversion/mod_dav_svn/version.c >> === >> --- subversion/mod_dav_svn/version.c(revision 1820704) >> +++ subversion/mod_dav_svn/version.c(working copy) >> @@

Re: Problems with commit feature negotiation and write-through proxies

2018-01-10 Thread Evgeny Kotkov
Philip Martin writes: > We already have system for handling create-txn and create-txn-with-props > so I was thinking of extending that code: > > Index: subversion/mod_dav_svn/version.c > === > ---

Re: Problems with commit feature negotiation and write-through proxies

2018-01-10 Thread Philip Martin
Evgeny Kotkov writes: > Philip Martin writes: > >> In the past we supported different versions on the master and slave and >> we introduced the SVNMasterVersion directive to configure it. >> Unfortunately that information is not immediately

Re: Problems with commit feature negotiation and write-through proxies

2018-01-10 Thread Evgeny Kotkov
Philip Martin writes: > In the past we supported different versions on the master and slave and > we introduced the SVNMasterVersion directive to configure it. > Unfortunately that information is not immediately available in > get_vsn_options() where the server

Problems with commit feature negotiation and write-through proxies

2018-01-09 Thread Philip Martin
I run a write-through proxy for svn.apache.org using mod_dav_svn's SVNMasterURI directive. Commits now fail when using trunk on the client and slave because they negotiate svndiff2 which the s.a.o master does not support. I can reproduce locally using a trunk slave and 1.9 master: $ svn