You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Brian Denny <br...@briandenny.net> on 2003/06/09 02:25:41 UTC
[PATCH] Followup to Issue #1075 / r5814
This patch tries to address some of the concerns expressed by Philip and
Mike about r5814. (It does *not* address all of Philip's concerns,
because Mike shot down my last try at a server-side solution and I just
don't want to go there again, at least not without some consensus.)
So at this point I'm basically looking for: Am i doing anything stupid, or
is this patch a reasonable improvement on r5814?
(I personally feel concerned about the number of different places in
libsvn_wc that I had to write very similar code to acheive what I
needed, both here and in r5814. Is this a reflection on my solution, or
a reflection on the state of libsvn_wc?)
-brian
Log:
Followup to r5814 (Issue #1075 - update receiving delete for missing
directory). Make sure to clean up missing directories any time we bump
a revision number (really, when we remove the 'incomplete' flag).
Also, don't delete schedule-add entries.
* libsvn_wc/log.c
(log_do_delete_entry): Don't delete schedule-add entries.
* libsvn_wc/adm_ops.c
(recursively_tweak_entries): Don't delete schedule-add entries.
* libsvn_wc/update_editor.c
(maybe_bump_dir_info): For each directory that we actually bump,
remove any entries for missing or 'deleted' children. (But don't
delete schedule-add entries).
Index: subversion/libsvn_wc/log.c
===================================================================
--- subversion/libsvn_wc/log.c (revision 6165)
+++ subversion/libsvn_wc/log.c (working copy)
@@ -536,14 +536,18 @@
{
if (svn_wc__adm_missing (adm_access, full_path))
{
- /* The directory is already missing, so don't try to recurse --
- just delete the entry in the parent directory. */
- apr_hash_t *entries;
- SVN_ERR (svn_wc_entries_read (&entries, loggy->adm_access, TRUE,
- loggy->pool));
- svn_wc__entry_remove (entries, name);
- SVN_ERR (svn_wc__entries_write (entries, loggy->adm_access,
- loggy->pool));
+ /* The directory is already missing, so don't try to recurse. */
+ if (entry->schedule != svn_wc_schedule_add)
+ {
+ /* Delete the parent directory's entry. */
+ apr_hash_t *entries;
+ SVN_ERR (svn_wc_entries_read (&entries, loggy->adm_access, TRUE,
+ loggy->pool));
+ svn_wc__entry_remove (entries, name);
+ SVN_ERR (svn_wc__entries_write (entries, loggy->adm_access,
+ loggy->pool));
+ }
+ /* Else, leave schedule-add entry untouched. */
}
else
{
Index: subversion/libsvn_wc/adm_ops.c
===================================================================
--- subversion/libsvn_wc/adm_ops.c (revision 6165)
+++ subversion/libsvn_wc/adm_ops.c (working copy)
@@ -117,14 +117,18 @@
if (remove_missing_dirs
&& svn_wc__adm_missing (dirpath, child_path))
{
- svn_wc__entry_remove (entries, name);
- if (notify_func)
- (* notify_func) (notify_baton, child_path,
- svn_wc_notify_delete,
- current_entry->kind, NULL,
- svn_wc_notify_state_unknown,
- svn_wc_notify_state_unknown,
- SVN_INVALID_REVNUM);
+ if (current_entry->schedule != svn_wc_schedule_add)
+ {
+ svn_wc__entry_remove (entries, name);
+ if (notify_func)
+ (* notify_func) (notify_baton, child_path,
+ svn_wc_notify_delete,
+ current_entry->kind, NULL,
+ svn_wc_notify_state_unknown,
+ svn_wc_notify_state_unknown,
+ SVN_INVALID_REVNUM);
+ }
+ /* Else if missing item is schedule-add, do nothing. */
}
/* Not missing or deleted, so recurse. */
Index: subversion/libsvn_wc/update_editor.c
===================================================================
--- subversion/libsvn_wc/update_editor.c (revision 6165)
+++ subversion/libsvn_wc/update_editor.c (working copy)
@@ -297,6 +297,9 @@
{
svn_wc_entry_t tmp_entry;
svn_wc_adm_access_t *adm_access;
+ apr_hash_t *entries, *dirents;
+ apr_hash_index_t *hi;
+ apr_pool_t *subpool = svn_pool_create (pool);
if (--bdi->ref_count > 0)
return SVN_NO_ERROR; /* directory isn't done yet */
@@ -309,6 +312,53 @@
&tmp_entry,
SVN_WC__ENTRY_MODIFY_INCOMPLETE,
TRUE /* immediate write */, pool));
+
+ /* Clean up deleted/missing directories. */
+ SVN_ERR (svn_wc_adm_retrieve (&adm_access, eb->adm_access, bdi->path,
+ pool));
+ SVN_ERR (svn_wc_entries_read (&entries, adm_access, TRUE, pool));
+ SVN_ERR (svn_io_get_dirents (&dirents, bdi->path, pool));
+ for (hi = apr_hash_first (pool, entries); hi; hi = apr_hash_next (hi))
+ {
+ svn_pool_clear (subpool);
+ const void *key;
+ apr_ssize_t klen;
+ void *val;
+ const svn_wc_entry_t *current_entry;
+
+ /* Get the next entry */
+ apr_hash_this (hi, &key, &klen, &val);
+ current_entry = val;
+
+ /* Skip THIS_DIR. */
+ if (! strcmp (key, SVN_WC_ENTRY_THIS_DIR))
+ continue;
+
+ /* If the item is a 'deleted' or missing dir, remove it
+ * (unless it is schedule-add). */
+ if (current_entry->kind == svn_node_dir
+ && (current_entry->deleted
+ || (! apr_hash_get (dirents, key, klen)))
+ && (current_entry->schedule != svn_wc_schedule_add))
+ {
+ svn_wc__entry_remove (entries, current_entry->name);
+ if (eb->notify_func)
+ (*eb->notify_func) (eb->notify_baton,
+ svn_path_join (bdi->path,
+ current_entry->name,
+ subpool),
+ svn_wc_notify_update_delete,
+ svn_node_dir, NULL,
+ svn_wc_notify_state_unknown,
+ svn_wc_notify_state_unknown,
+ SVN_INVALID_REVNUM);
+ }
+
+ } /* end entries loop */
+
+ svn_pool_destroy (subpool);
+
+ SVN_ERR (svn_wc__entries_write (entries, adm_access, pool));
}
/* we exited the for loop because there are no more parents */
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: [PATCH] Followup to Issue #1075 / r5814
Posted by cm...@collab.net.
Brian Denny <br...@briandenny.net> writes:
> (I personally feel concerned about the number of different places in
> libsvn_wc that I had to write very similar code to acheive what I
> needed, both here and in r5814. Is this a reflection on my solution, or
> a reflection on the state of libsvn_wc?)
Without examining your patch, let me assure you that it is quite
likely that the state of libsvn_wc is such that it would cause you
this kind of repetitive pain.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org