On Mon, 16 Apr 2018 08:25:07 -0400, Matt Harbison wrote:
>
> > On Apr 16, 2018, at 7:35 AM, Yuya Nishihara <y...@tcha.org> wrote:
> >
> >> On Sun, 15 Apr 2018 02:44:08 -0400, Matt Harbison wrote:
> >> # HG changeset patch
> >> # User Matt Harbison <matt_harbi...@yahoo.com>
> >> # Date 1523752111 14400
> >> # Sat Apr 14 20:28:31 2018 -0400
> >> # Node ID 3b0c3d4939b56ca038dbbba17da424699a6b339d
> >> # Parent 691a7d3f2df80908e52a170897a4492a528c3533
> >> verify: invoke the file prefetch hook
> >>
> >> It's unfortunate that verify wants to download everything. Maybe we can
> >> create
> >> a custom transfer handler for LFS. But it shouldn't be painful in the
> >> meantime,
> >> and it's likely that blobs will be served up from more than just hgweb.
> >>
> >> diff --git a/mercurial/verify.py b/mercurial/verify.py
> >> --- a/mercurial/verify.py
> >> +++ b/mercurial/verify.py
> >> @@ -25,6 +25,7 @@ from . import (
> >>
> >> def verify(repo):
> >> with repo.lock():
> >> + scmutil.fileprefetchhooks(repo, repo.set('all()'))
> >> return verifier(repo).verify()
> >
> > I don't think "hg verify" should go that abstraction level because the
> > repository might be inconsistent state.
>
> Do you mean using a revset, or prefetching at all? Grabbing one file at a
> time will likely be painfully slow.
>
> If the revset bit is the bad part, what is the alternative with a corrupt
> repo?
I think we should avoid any write operation on damaged repo. Is there any way
to disable fetching at all?
_______________________________________________
Mercurial-devel mailing list
Mercurial-devel@mercurial-scm.org
https://www.mercurial-scm.org/mailman/listinfo/mercurial-devel
