bliss@tigris.org writes:
> --- trunk/subversion/libsvn_client/copy.c (original)
> +++ trunk/subversion/libsvn_client/copy.c Sat Feb 19 17:47:38 2005
> @@ -132,8 +132,9 @@
> }
> else
> {
> - SVN_ERR (svn_wc_adm_open2 (&adm_access, NULL, dst_parent,
> - TRUE, 0, pool));
> + SVN_ERR (svn_wc_adm_open3 (&adm_access, NULL, dst_parent,
> + TRUE, 0, ctx->cancel_func,
> + ctx->cancel_baton,pool));
> }
> }
I don't think it's worth another commit to fix, but watch out for
spacing (after the comma above).
> --- trunk/subversion/libsvn_client/prop_commands.c (original)
> +++ trunk/subversion/libsvn_client/prop_commands.c Sat Feb 19 17:47:38 2005
> @@ -203,8 +203,8 @@
> return svn_error_createf (SVN_ERR_CLIENT_PROPERTY_NAME, NULL,
> _("Bad property name: '%s'"), propname);
>
> - SVN_ERR (svn_wc_adm_probe_open2 (&adm_access, NULL, target, TRUE,
> - recurse ? -1 : 0, pool));
> + SVN_ERR (svn_wc_adm_probe_open3 (&adm_access, NULL, target, TRUE,
> + recurse ? -1 : 0, NULL, NULL, pool));
I was going to ask, why NULLs here, but then when I went to look at
the latest trunk code, I see that they are now a ctx->cancel_func and
ctx->cancel_baton. So just pretend I never mentioned it.
> SVN_ERR (svn_wc_entry (&node, target, adm_access, FALSE, pool));
> if (!node)
> return svn_error_createf (SVN_ERR_UNVERSIONED_RESOURCE, NULL,
> @@ -429,8 +429,8 @@
> else
> pdir = target;
>
> - SVN_ERR (svn_wc_adm_open2 (&adm_access, NULL, pdir, FALSE,
> - 0, pool));
> + SVN_ERR (svn_wc_adm_open3 (&adm_access, NULL, pdir, FALSE,
> + 0, NULL, NULL, pool));
> SVN_ERR (svn_wc_entry (&entry, target, adm_access, FALSE, pool));
> if (! entry)
> return svn_error_createf (SVN_ERR_UNVERSIONED_RESOURCE, NULL,
Any reason not to make maybe_convert_to_url() take a ctx object (or
take the cancellation parameters explicitly), and pass them through to
this call to svn_wc_adm_open3()? Is it that this is a one-time
resolution that isn't happening in a loop, so response time to
cancellation wouldn't be appreciably reduced anyway?
> --- trunk/subversion/libsvn_client/revisions.c (original)
> +++ trunk/subversion/libsvn_client/revisions.c Sat Feb 19 17:47:38 2005
> @@ -78,8 +78,8 @@
> return svn_error_create
> (SVN_ERR_CLIENT_VERSIONED_PATH_REQUIRED, NULL, NULL);
>
> - SVN_ERR (svn_wc_adm_probe_open2 (&adm_access, NULL, path, FALSE,
> - 0, pool));
> + SVN_ERR (svn_wc_adm_probe_open3 (&adm_access, NULL, path, FALSE,
> + 0, NULL, NULL, pool));
> SVN_ERR (svn_wc_entry (&ent, path, adm_access, FALSE, pool));
> SVN_ERR (svn_wc_adm_close (adm_access));
Same question here, I guess, but the answer is starting to emerge from
the mist now: it's what I said, that there's no point passing the
cancellation stuff in a non-loop call. Which seems reasonable to me,
so I'll stop mentioning it now.
Nice change!
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Wed Feb 23 00:36:03 2005