You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by kf...@collab.net on 2004/10/18 05:05:59 UTC
Re: svn commit: r11439 - trunk/subversion/libsvn_client
lundblad@tigris.org writes:
> --- trunk/subversion/libsvn_client/status.c (original)
> +++ trunk/subversion/libsvn_client/status.c Sat Oct 16 15:42:30 2004
> @@ -119,27 +124,52 @@
> return svn_error_createf (SVN_ERR_UNVERSIONED_RESOURCE, NULL,
> _("'%s' is not under version control"), path);
>
> - /* Close up our ADM area. We'll be re-opening soon. */
> - SVN_ERR (svn_wc_adm_close (adm_access));
> + /* Need to lock the tree. We lock the anchor first, to not lock too much
> + if we have a target. We need to lock the target and immediate children
> + if the status is non-recursive. Else, we lock the whole hierarchy
> + under target. If we are contacting the repository, we need a recursive
> + lock under anchor so that auth data can be stored. */
So "that auth data can be stored"? What sort of auth data?
Btw, I looked at the rest of the commit, but wasn't able to fully
follow it (which is not a statement about your code, more about my not
having worked in libsvn_wc in too long). I'll try to review it with
Ben Collins-Sussman tomorrow.
Best,
-Karl
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: svn commit: r11439 - trunk/subversion/libsvn_client
Posted by Philip Martin <ph...@codematters.co.uk>.
"Peter N. Lundblad" <pe...@famlundblad.se> writes:
> If someone with deeper knowledge can confirm that locking the whole tree
> just for authentication isn't necessary anymore, then I could elminitate
> this.
It appears that the old comment came from r4311 so it refers to the
old auth system and is no longer relevant.
--
Philip Martin
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: svn commit: r11439 - trunk/subversion/libsvn_client
Posted by "Peter N. Lundblad" <pe...@famlundblad.se>.
On Thu, 28 Oct 2004, Julian Foad wrote:
> Peter N. Lundblad wrote:
> > ------------------------------------------------------------------------
> > r11439 | lundblad | 2004-10-16 21:42:30 +0100 (Sat, 16 Oct 2004) | 6 lines
> >
> > Fix slow "svn st" where the targets are files by not locking the whole anchor
> > tree.
> >
> > * libsvn_client/status.c (svn_client_status): Don't lock the whole directory
> > hierarchy if not necessary.
>
> In the log message we give the path to each file relative to the root of the
> source tree - thus it should be "subversion/libsvn_client/status.c".
>
Hmmm, you're right. Interesting no one has complained before...
Also, this is what svn-log-message in Emacs svn-dev.el does for me. Anyone
knows why. I ahve my soruces in ~/src/svn/svn. Maybe others use another
convention?
> > + if (update)
> > + {
> [...]
> > + }
> > + else
> > + {
> [...]
> > + err = svn_wc_adm_open2 (&target_access, anchor_access, path,
> > + FALSE, (descend || update) ? -1 : 1, pool);
>
> The variable "update" is redundant in that test since it is known to be false.
>
True. I'll eliminate the whole special-case for the update case. It's not
necessary anymore according to earlier discussion.
Thanks for the review,
//Peter
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: svn commit: r11439 - trunk/subversion/libsvn_client
Posted by Julian Foad <ju...@btopenworld.com>.
Peter N. Lundblad wrote:
> ------------------------------------------------------------------------
> r11439 | lundblad | 2004-10-16 21:42:30 +0100 (Sat, 16 Oct 2004) | 6 lines
>
> Fix slow "svn st" where the targets are files by not locking the whole anchor
> tree.
>
> * libsvn_client/status.c (svn_client_status): Don't lock the whole directory
> hierarchy if not necessary.
In the log message we give the path to each file relative to the root of the
source tree - thus it should be "subversion/libsvn_client/status.c".
> + if (update)
> + {
[...]
> + }
> + else
> + {
[...]
> + err = svn_wc_adm_open2 (&target_access, anchor_access, path,
> + FALSE, (descend || update) ? -1 : 1, pool);
The variable "update" is redundant in that test since it is known to be false.
- Julian
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: svn commit: r11439 - trunk/subversion/libsvn_client
Posted by "Peter N. Lundblad" <pe...@famlundblad.se>.
On Mon, 18 Oct 2004, Tobias Ringström wrote:
> kfogel@collab.net wrote:
>
> >So "that auth data can be stored"? What sort of auth data?
> >
> >
> It looks like a copied comment from the old code, and I think Peter just
> copied the functionality from the old code.
>
That's exactly the case:-)
> Peter, a long time ago, Suversion stored the authentication data in the
> .svn directories, but it was moved to e.g. ~/.subversion/auth before you
> got here. It's possible that the code no longer really requires that
> recursive lock because of that change. I should warn you that I have
> not looked at the code in question, so I cannot tell you for certain.
>
So even more optimization is possible (and code decomplexification). This
is good, but I don't think the code is wrong as-is.
If someone with deeper knowledge can confirm that locking the whole tree
just for authentication isn't necessary anymore, then I could elminitate
this.
Thanks,
//Peter
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: svn commit: r11439 - trunk/subversion/libsvn_client
Posted by Tobias Ringström <to...@ringstrom.mine.nu>.
kfogel@collab.net wrote:
>So "that auth data can be stored"? What sort of auth data?
>
>
It looks like a copied comment from the old code, and I think Peter just
copied the functionality from the old code.
Peter, a long time ago, Suversion stored the authentication data in the
.svn directories, but it was moved to e.g. ~/.subversion/auth before you
got here. It's possible that the code no longer really requires that
recursive lock because of that change. I should warn you that I have
not looked at the code in question, so I cannot tell you for certain.
/Tobias
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org