RE: svnsync crash on empty repo

2015-09-06 Thread Bert Huijben
I'll write a larger reply tomorrow (after looking through the code a bit more), but I think this is about how I feel about this. Great summary! Bert -Original Message- From: "Julian Foad" Sent: ‎6-‎9-‎2015 15:15 To: "Bert Huijben" Cc: "dev" Subject:

Re: svnsync crash on empty repo

2015-09-06 Thread Julian Foad
I have had an idea about what may be bothering you. Is it that you don't like us to have any interface where a valid revnum value could be greater than HEAD, because it's counter-intuitive and we rarely do it and in most cases a valid revnum cannot be greater than head? If that's the case, I have

Re: svnsync crash on empty repo

2015-09-06 Thread Julian Foad
Bert Huijben wrote: > Julian Foad wrote: >> Bert wrote: >>> I don't think we should fix this with a 'revision+1' to explicitly allow >>> many bad ranges, >> >> Why? >> >> An empty range isn't inherently "bad". Allowing an empty range isn't >> bad. Being able to specify an empty range in many differ

RE: svnsync crash on empty repo

2015-09-05 Thread Bert Huijben
> -Original Message- > From: Julian Foad [mailto:julianf...@gmail.com] > Sent: zaterdag 5 september 2015 15:00 > To: Bert Huijben > Cc: dev > Subject: Re: svnsync crash on empty repo > > Bert wrote: > > I don't think we should fix this with a '

Re: svnsync crash on empty repo

2015-09-05 Thread Julian Foad
Bert wrote: > I don't think we should fix this with a 'revision+1' to explicitly allow > many bad ranges, Why? An empty range isn't inherently "bad". Allowing an empty range isn't bad. Being able to specify an empty range in many different ways isn't bad. Changing a released API to make a previou

RE: svnsync crash on empty repo

2015-09-05 Thread bert
Huijben Subject: svnsync crash on empty repo Hi, Bert. On trunk (released versions are unaffected), "svnsync sync" with an empty source repo fails an assertion in svn_ra_replay_range(): subversion/libsvn_ra/ra_loader.c' line 1198: assertion failed (SVN_IS_VALID_REVNUM(start_re

Re: svnsync crash on empty repo

2015-09-05 Thread Julian Foad
I (Julian Foad) wrote: > On trunk (released versions are unaffected), "svnsync sync" with an > empty source repo fails an assertion in svn_ra_replay_range(): In fact this problem is not specific to an empty repository, but to any repository that is already fully synchronized. An empty repository j

svnsync crash on empty repo

2015-09-05 Thread Julian Foad
Hi, Bert. On trunk (released versions are unaffected), "svnsync sync" with an empty source repo fails an assertion in svn_ra_replay_range(): subversion/libsvn_ra/ra_loader.c' line 1198: assertion failed (SVN_IS_VALID_REVNUM(start_revision) && SVN_IS_VALID_REVNUM(end_revision) && start_revision <=