John McEleney / mailutils

Alain Magloire authored 2001-02-07 06:50:17 +0000

Alain Magloire authored 2001-02-06 04:30:13 +0000

Still yucky, much more things need to be put on the sanity checks.

Alain Magloire authored 2001-02-05 16:23:14 +0000

Alain Magloire authored 2001-02-05 05:26:50 +0000

Alain Magloire authored 2001-02-05 05:26:14 +0000

Alain Magloire authored 2001-02-05 05:25:43 +0000

* mailbox/imap4d/select.c : First draft implementation.
* mailbox/imap4d/util.c : Add util_msgset() and util_send().
Reuse of util_getcommand() for subcommands.

Alain Magloire authored 2001-02-05 05:25:24 +0000

to have UIDL POP3.

Alain Magloire authored 2001-02-05 05:24:52 +0000

message_set_get_uid().
* include/mailutils/mailbox.h: New prototypes mailbox_uidnext(),
mailbox_uidvalidity().

Alain Magloire authored 2001-02-05 05:23:37 +0000

the UID of the message (IMAP definition) and message_get_uidl() for
the POP3 definition.
* mailbox/attachment.c : Indentation fixes.
* mailbox/file_stream.c : Make sure that the FILE* is close before
opening.
* mailbox/folder_imap.c : Use strcasecmp() not strcmp().
* mailbox/mailbox.c : Added mailbox_uidnext() and mailbox_uidvalidity.
* mailbox/mapfile_stream.c : When munmap() failed we should check
against MAP_FAILED, not NULL.
* mailbox/mbx_imap.c : Implement imap_uidnext(), imap_uidvalidity()
imap_message_uid().
* mailbox/mbx_mbox.c : Implement mbox_uidnext(), mbox_uidvalidity()
mbox_message_uid().
* mailbox/mbx_mboxscan.c : Implement uidvalidity, it is save int the
header of the first message, "X-IMAPbase: 127673838 123", a la c-client.
Save UID, it is save in "X-UID".
* mailbox/mbx_pop.c : Implement pop_messages_recent(),
pop_message_unseen() and pop_uid().
* mailbox/trans_stream.c : Indentation fixes.

Alain Magloire authored 2001-02-05 05:21:43 +0000

is known, we can use memcpy() and friends for string manipulations.
memXXX() functions are builtin in GNU C (gcc) and are much faster
then the strXXX() counterparts. Small change in the loop to take
advantage of this.

* pop3d/quit.c : The rfc1939 insist that after a QUIT, we must
release any resources and close the connection:
"Whether the removal was successful or not, the server
then releases any exclusive-access lock on the maildrop
and closes the TCP connection."  This was not done if error
occured while expunging, no we will close the connection and
notify the client of the error(ERR_FILE).

Alain Magloire authored 2001-02-02 04:00:10 +0000

not found.
* mailbox/mbx_mbox.c: When expunging check to see if the registar
the path_record too.

Alain Magloire authored 2001-02-02 03:59:35 +0000

Alain Magloire authored 2001-02-02 03:59:01 +0000

Alain Magloire authored 2001-02-01 04:38:27 +0000

header_get_fvalue()
mailutils/mailbox.h  : rename mailbox_recent_count() to
mailbox_unseen_count()

Alain Magloire authored 2001-02-01 04:23:26 +0000

header.c mbx_mboxscan.c mbx_mbox.c : an other attempt for speed
when parsing, the most common use headers (From, To, Date, CC,..)
are save in the structure, so when retrieve them it will be
faster.  Downside the tight parsing loop is not so thight and
is frankly ugly.  More memory is being use per messages.
It is hard to guess at this stage the memory patterns
and where to optimise, probably my last attempt at this
before the code become truly unmaintable.

mbx_imap.c mailbox.c : I did not understand the difference between
"recent" and "unseen" english words.  I think that "unseen"
means new mail, an "recent" mail that was not read yet.
So I had to rename some functions mailbox_recent_count()
to mailbox_unseen_count(), but I maybe wrong ... sigh
somebody with better english skill should explain this
to me.

mbx_pop.c include/header0.h include/mailbox0.h :
Introduce header_set_fill() to let the mailboxes fill
up a buffer that will contains the entire headers.

Alain Magloire authored 2001-02-01 04:21:52 +0000

Alain Magloire authored 2001-01-29 06:17:23 +0000

When being very quiet just freopen(/dev/null, w, stdout).

Alain Magloire authored 2001-01-29 06:15:56 +0000

Alain Magloire authored 2001-01-29 06:13:23 +0000

and update Copyrigth.
* pop3d/pop3d.c : Remove mbox_record register, only need path_record.
* pop3d/retr.c : When retrieve a message should be mark as read.
since we do not strip out the header "Status:".
* pop3d/top.c : Adjust stream_readline().

Alain Magloire authored 2001-01-29 06:13:03 +0000

frm/Makefile.am: s/from/frm/

Jakob Kaivo authored 2001-01-26 07:55:35 +0000

shape and ready.  The biggest thing will be the
awfull parsing done for FETCH/SEARCH etc ..
"This will be for another day" -- Scarlet O'hara

Alain Magloire authored 2001-01-26 04:55:24 +0000

This was usefull to implement SELECT/EXAMINE for imap4

Alain Magloire authored 2001-01-26 04:53:17 +0000

Alain Magloire authored 2001-01-26 04:50:58 +0000

the display.

Alain Magloire authored 2001-01-26 04:50:35 +0000

failed ofile == NULL we should not attempt to fprintf(ofile, ..) nor
fflush().  Buglet discover when working on imap4d.
pop3d.[ch]/user.c : indentation and comments.

Alain Magloire authored 2001-01-25 05:12:42 +0000

Alain Magloire authored 2001-01-25 05:09:03 +0000

change, this will help to know if the message_is_modified() and speed things
up when expunging.
file_stream.c : Don't disable the stream buffering and let the OS choose the
best BUFSIZ i.e. remove setbuf() call.
message.c registrar.c mailer.c folder.c header.c list.c mailbox.c :
Copyright and indent.
mapfile_stream.c mbx_imap.c mbx_mbox.c mbx_pop.c : go easy on the stack
for array of strings.
stream.c : Put code for doing our own simple littel buffering.

Alain Magloire authored 2001-01-25 05:08:41 +0000

was doing the equivalent of:

char buffer[1024];
memset (buffer, 0, 1024);
...
read (popfd, buffer, 1024);
...
ret = malloc (strlen (buffer) +1);

According to the rfc a command line should be no longer then
255 but a malicious client could send a big buffer that could fit
1024 then buffer would not be null terminated, strlen(buffer) may
be supceptible to overflow.  A simple fix would be to
read (fd, buffer, 1023); /* leave space for a null */
I've put a different fix that does not need the call to memset().
And at the same time reduce the size of the buffer to  go
easy on the stack 512 is sufficient.

Alain Magloire authored 2001-01-23 17:13:22 +0000

Jakob Kaivo authored 2001-01-23 10:33:59 +0000

Jakob Kaivo authored 2001-01-23 10:26:09 +0000

Jeff Bailey authored 2001-01-20 04:52:47 +0000

pop: RETR and TOP if buffer is not a full line i.e terminated by a '\n' realloc
the buffer and try again.
pop: STAT remove excess comment.
pop: user use the /dev/null scheme if mailbox does not exist.
mailbox: buglet if the mailbox format we were reading less then expected.

Alain Magloire authored 2001-01-20 04:34:39 +0000

Alain Magloire authored 2001-01-19 05:05:59 +0000

to the "frm" that comes with elm package.  It was written in less then 15
minutes, cool.

Alain Magloire authored 2001-01-19 05:03:13 +0000

Alain Magloire authored 2001-01-17 03:43:06 +0000

Alain Magloire authored 2001-01-17 00:48:18 +0000

Alain Magloire authored 2001-01-16 06:24:53 +0000

include multiple times.  Compiler like Watcomm C, will scream loudly
about definition of those variables multiple times.  Gcc is very silent
about this.  So in pop3d.h we declare those variables extern and move the
definition in pop3d.c.
MAXHOSTNAMELEN is define in netinet/in.h on Solaris so we guard with
#ifdef's.
We were calling syslog("%s ", NULL); On solaris passing a NULL to  a printf
function will crash and  burn, this was a FIXME I did not see it earlier
the NULL was the place holder for the mailbox name.

Alain Magloire authored 2001-01-16 06:24:08 +0000

address.c mime.c misc.c: Always cast to (unsigned char) when using the ctype
functions like isspace();
mbx_default.c: If _PATH_SENDMAIL not set via paths.h use "/usr/spool/mail"
message.c: uidl is now <md5 . time . sequene>
tcp.c: INADDR_NONE is not define on solaris define it to -1.

Alain Magloire authored 2001-01-16 06:14:52 +0000