Mercurial > dovecot > original-hg > dovecot-1.2
changeset 3048:12fc3c22ffe8 HEAD
updated
author | Timo Sirainen <tss@iki.fi> |
---|---|
date | Fri, 07 Jan 2005 00:48:13 +0200 |
parents | 89c49afa57fd |
children | 7f503848533e |
files | TODO |
diffstat | 1 files changed, 51 insertions(+), 8 deletions(-) [+] |
line wrap: on
line diff
--- a/TODO Fri Jan 07 00:26:40 2005 +0200 +++ b/TODO Fri Jan 07 00:48:13 2005 +0200 @@ -1,4 +1,52 @@ - - lock_method=dotlock, mmap_disable=no crashes in cache locking + - message header search: we should ignore LWSP between two MIME blocks + - if indexes exist but dovecot-uidlist doesn't, it's not tried to be recreated + - maildir: copy is broken? + - copy doesn't print error message if source is expunged + + - keywords: + - add some limits to how many there can be + - don't return \* in PERMANENTFLAGS when we're full + - send FLAGS/PERMANENTFLAGS untagged replies when they change + - sync to mbox/maildir + - remove unused keywords? + - support specifying hex/base64 encoding in password scheme. for example + {plain-md5.base64} + - happens too often: mail-transaction-log-view.c: line 138 (mail_transaction_log_view_set): assertion failed: (min_file_seq != max_file_seq || min_file_offset <= max_file_offset) + - login_process_per_connection=no + auth_count > 1 + - dovecot-auth gives timeout errors + - failed requests are sent to all auth processes + - mbox file offsets still aren't 64bit aligned for some reason.. +line 493 (mbox_lock): assertion failed: (lock_type == F_RDLCK || ibox->mbox_lock_type != F_RDLCK) +#5 0x0806faa3 in mbox_lock (ibox=0x80d89b8, lock_type=1, lock_id_r=0xbea6b434) at mbox-lock.c:493 +#6 0x08075439 in mbox_sync (ibox=0x80d89b8, flags=MBOX_SYNC_REWRITE) at mbox-sync.c:1286 +#7 0x0806d38b in mbox_storage_close (box=0x80d89b8) at mbox-storage.c:793 +#8 0x0809666e in mailbox_close (box=0x80d89b8) at mail-storage.c:296 +#9 0x080587b6 in client_destroy (client=0x80cefe0) at client.c:69 +#10 0x080591c3 in client_output (context=0x80cefe0) at client.c:386 +#11 0x080b0086 in stream_send_io (context=0x80cf0a8) at ostream-file.c:339 + +mail-index-transaction.c: line 467 (mail_index_transaction_add_last): +assertion failed:(idx == size || data[idx].uid1 <= update.uid1) + + - flag merging code is too complex in index syncing. it should be moved into + mbox/maildir syncing where they're combined + - index_removal_timeout gets leaked in some conditions. how? + - non-mmaped cache: last block in file isn't cached. is invalidating still + not enough? if cache_offset is updated after our view is synced, we see the + new offset but not it's invalidation? -> don't re-read index file, sync it + as a view. + + - login_process_per_connection = no -> log throtting is beginning to be a + problem! make it per-file setting and disable in this case. + - rename foo foo.xyz -> infinite loop possible? + - pipe() failed: Too many open files - fine, but don't log it 1000 times + a second + - inetd startup doesn't work anymore + - auth protocol: make sure values can't have tabs/lfs + - auth cache: cache userdb data too. + - passdb extra data: if name starts with userdb_, user it for userdb. + add "passdb" userdb if passdb handles that. + - auth: remove system_user and allow returning multiple gids instead. - transaction log: when replacing log with a same sequence, we remove it from log's file list, but we don't do anything to existing log views. this can crash later in mail_transaction_log_view_set() because 'first' is from log @@ -29,9 +77,8 @@ - we can mail_cache_add() same header multiple times.. - configurable syslog prefix - SIGHUP restarts auth processes .. but does it wait until they've finished - with all requests? + with all requests? no. - SIGHUP rather shouldn't restart listening sockets if they didn't change.. - - dovecot-auth + ldap leaks? - subscriptions file should contain namespace prefixes. at least optionally. there's the subscriptions = yes setting now for namespaces.. do it so that if prefix = "" has subscriptions, it contains prefixes. otherwise not. @@ -40,6 +87,7 @@ - mbox: syncing existing indexes takes 4x longer than creating new one, why? - caching + - size.physical isn't cached - force bits should be used only for nonregistered fields - change envelope parsing not to use get_headers() so imap.envelope can actually be cached without all the headers.. @@ -77,8 +125,6 @@ - open("/var/run/dovecot//ssl-parameters.dat", O_RDONLY|O_LARGEFILE) =-1 ENOENT ^ loops forever - - don't return \* in PERMANENTFLAGS when we're full - - support read-only logins. user could with alternative password get only read-access to mails so mails could be read relatively safely with untrusted computers. Maybe always send [ALERT] about the previous @@ -230,9 +276,6 @@ - cleanups: - check if t_push()/t_pop() should be added somewhere - - try to fix @UNSAFE code to use buffer API instead - - subscription-file.c, custom_flags - - [io]stream-file.c? - grep for FIXME - auth / login: