Apologies for the late answer. Eric S. Raymond wrote on Thu, Jan 21, 2021 at 23:34:24 -0500: > Daniel Shahaf <d...@daniel.shahaf.name>: > > > > It makes me twitchy that there's any possible future in which that > > > > format might be unsuppoted or inaccessible. > > > > We _are_ committed to providing an upgrade path from 1.x to 2.x, you > > know. Always have been. You can stop twitching. > > It's not 1.x I'm worried about being dropped, it's 2.x. And only > because Julian said he couldn't think of any good reason to use it.
"1.x" and "2.x" are version numbers (as in `svn --version -q`), not dump format numbers. > Bitter experience tells me that when devs start thinking that about an > export feature it tends to get dropped in a later release. Again, we're committed to our compatibility promises. Basically, since we promise compatibility within a major line and an upgrade path from 1.x to 2.x, we aren't going to lose the ability to consume anything that any 1.x produced any time soon, and possibly ever. > But probably the bigger issue for me right now is a whether Subversion devs > understand that Version 2 has real uses and dropping it because you > think v3 is better would cause a lot of unpleasant ripple effcts. I think there are good arguments for both sides here, but in any case, dropping the ability to either produce or consume v2 dump streams isn't on the table at this point in time. If you're concerned about it happening down the road, your best bet is to get that concern documented in a regression test or at least a milestone=2.0 issue.
