[Dovecot] 20060114 snapshot compilation warnings
Timo,
I've noticed a couple of extra compilation errors in last night's snapshot compared to the snapshot I downloaded on Tuesday, probably due to the int -> bool change :- (gcc 3.3.2 on Solaris 8)
mbox-sync.c: In function `mbox_sync': mbox-sync.c:1568: warning: comparison is always false due to limited range of data type
index-search.c: In function `search_index_arg': index-search.c:131: warning: comparison is always false due to limited range of data type
quota-storage.c: In function quota_check': quota-storage.c:114: warning: passing arg 3 of quota_try_alloc' from
incompatible pointer type
together with a few "unused parameter/variable" errors which were there before.
I also needed to rewrite half my internal patches :) but that shouldn't have made any difference (they certainly don't involve those files).
Best Wishes, Chris
-- --+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+- Christopher Wakelin, c.d.wakelin@reading.ac.uk IT Services Centre, The University of Reading, Tel: +44 (0)118 378 8439 Whiteknights, Reading, RG6 2AF, UK Fax: +44 (0)118 975 3094
Talking of which, one of my patches is for what I think are minor bugs with LIST and namespaces:
- LIST "<namespace>" "" should return the delimeter and root name (returns delimeter and ""), I think because namespace_find modifies the "ref" argument
- LIST "" "<namespace>" probably should return LIST (\Noselect) "<delim>" "<namespace>" (this keeps Pine and possibly other clients happy, otherwise they try to "create" the directory)
i.e. for "/" delimeter and "~/mail/" prefix I'd want
. LIST "" "~/mail/"
- LIST (\Noselect) "/" "~/mail/" . LIST "~/mail/" ""
- LIST (\Noselect) "/" "~/"
I can supply the patches but the first is trivial and the second one is very ugly!
Best Wishes, Chris
Chris Wakelin wrote:
I also needed to rewrite half my internal patches :) but that shouldn't have made any difference (they certainly don't involve those files).
Best Wishes, Chris
-- --+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+- Christopher Wakelin, c.d.wakelin@reading.ac.uk IT Services Centre, The University of Reading, Tel: +44 (0)118 378 8439 Whiteknights, Reading, RG6 2AF, UK Fax: +44 (0)118 975 3094
On Sat, 2006-01-14 at 14:36 +0000, Chris Wakelin wrote:
Talking of which, one of my patches is for what I think are minor bugs with LIST and namespaces:
- LIST "<namespace>" "" should return the delimeter and root name (returns delimeter and ""), I think because namespace_find modifies the "ref" argument
This already works actually as long as the namespace isn't hidden. But looks like the logic was a bit strange for hidden namespaces. Now the hidden namespaces are ignored completely when looking for the root namespace.
- LIST "" "<namespace>" probably should return LIST (\Noselect) "<delim>" "<namespace>" (this keeps Pine and possibly other clients happy, otherwise they try to "create" the directory)
Fixed.
. LIST "~/mail/" ""
- LIST (\Noselect) "/" "~/"
I think UW-IMAP returns this because ~/ is a namespace. But it also has this weird behavior:
x LIST foo/bar ""
- LIST (\Noselect) "/" "foo/"
Using the same logic ~/mail/ also gets truncated to ~/. And now that Dovecot is ignoring the hidden namespaces, it fallbacks to looking "" namespace with ~/mail/ parameter, and ~/ is returned.
On Sat, 2006-01-14 at 14:21 +0000, Chris Wakelin wrote:
Timo,
I've noticed a couple of extra compilation errors in last night's snapshot compared to the snapshot I downloaded on Tuesday, probably due to the int -> bool change :- (gcc 3.3.2 on Solaris 8)
Looks like gcc 4.0 has gotten worse with detecting these. Fixed all of them.
Oops, looks like I missed one (this time I was trying gcc 2.95, which is the version "shipped" with Solaris 8)
mail-index-sync.c: In function `mail_index_sync_next': mail-index-sync.c:587: warning: comparison is always false due to limited range of data type
The others are now fine.
Here's all the errors (I did "make > make.log", so this is what came on stderr)
mail-index-transaction.c: In function mail_index_update_header_ext': mail-index-transaction.c:695: warning: unused parameter t'
mail-index-transaction.c:696: warning: unused parameter ext_id' mail-index-transaction.c:696: warning: unused parameter offset'
mail-index-transaction.c:697: warning: unused parameter data' mail-index-transaction.c:697: warning: unused parameter size'
mail-index-sync.c: In function mail_index_sync_next': mail-index-sync.c:587: warning: comparison is always false due to limited range of data type dbox-transaction.c: In function dbox_transaction_commit':
dbox-transaction.c:20: warning: unused parameter flags' dbox-transaction.c: In function dbox_transaction_rollback':
dbox-transaction.c:77: warning: unused variable dbox' auth-master-connection.c: In function master_input_die':
auth-master-connection.c:145: warning: unused parameter `conn'
Best Wishes, Chris
Timo Sirainen wrote:
On Sat, 2006-01-14 at 14:21 +0000, Chris Wakelin wrote:
Timo,
I've noticed a couple of extra compilation errors in last night's snapshot compared to the snapshot I downloaded on Tuesday, probably due to the int -> bool change :- (gcc 3.3.2 on Solaris 8)
Looks like gcc 4.0 has gotten worse with detecting these. Fixed all of them.
-- --+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+---+- Christopher Wakelin, c.d.wakelin@reading.ac.uk IT Services Centre, The University of Reading, Tel: +44 (0)118 378 8439 Whiteknights, Reading, RG6 2AF, UK Fax: +44 (0)118 975 3094
On 2006-01-14 17:19:58 +0200, Timo Sirainen wrote:
On Sat, 2006-01-14 at 14:21 +0000, Chris Wakelin wrote:
Timo,
I've noticed a couple of extra compilation errors in last night's snapshot compared to the snapshot I downloaded on Tuesday, probably due to the int -> bool change :- (gcc 3.3.2 on Solaris 8)
Looks like gcc 4.0 has gotten worse with detecting these. Fixed all of them.
gcc 4.1 is even more strict. :) *starts a rebuild*
darix
participants (3)
-
Chris Wakelin
-
Marcus Rueckert
-
Timo Sirainen