On Fri, 2026-08-28 15:25:48 +0200, Rhialto <rhialto%falu.nl@localhost> wrote: > On Thu 27 Aug 2026 at 17:46:02 +0000, Benny Siegert wrote: > > I have a box with a checkout of NetBSD src from Mercurial. It was last > > updated in February. However, hg pull fails in one of the "manifests", as > > below. The error is reproducible, with an identical message. What can I do? > > > > # hg pull > > pulling from https://anonhg.netbsd.org/src-public > > searching for changes > > adding changesets > > adding manifests > > transaction abort! > > rollback completed > > abort: stream ended unexpectedly (got 24594 bytes, expected 32768) > > I had the same thing. In my case I was told that because of the file > system failure and subsequent re-creation of the hg repository, > essentially everything in the history was changed, and it would be > faster to move the repo aside and clone it again, fresh. So that's probably the same root cause why the Github mirror hasn't been updated since ... some two weeks. For my internal builds (doing CI mass builds coined towards reproducibility, including comparing NetBSD- and Linux-based builds), having a GIT repo is the easiest way in my setup. As I'm only interested in `current`, I'm really interested in seeing the NetBSD GIT conversion running again. (Or as an alternative, if it's clear that it'll take quite some more time to get the whole setup up'n'running again, I'd just create an own repo, simply importing some current `cvs update`d tree each time.) In the end, it would be quite helpful (at least to me) to hear about the GIT conversion future, possible obstacles and if there's something I'd help with. MfG, JBG --
Attachment:
signature.asc
Description: PGP signature