- 20 May, 2001 7 commits
-
-
- 17 May, 2001 1 commit
-
-
- 14 May, 2001 2 commits
-
-
- 13 May, 2001 1 commit
-
-
- 11 May, 2001 1 commit
-
-
- 10 May, 2001 1 commit
-
-
- 09 May, 2001 2 commits
-
-
- 07 May, 2001 3 commits
-
-
- 04 May, 2001 3 commits
-
-
- 02 May, 2001 2 commits
-
-
- 25 Apr, 2001 1 commit
-
-
- 24 Apr, 2001 5 commits
-
-
message into mum->message member. But mbox_get_message() relies on 1-based message numbers, whereas `mum' pointer is obtained using 0-based indexing:
Alain Magloire authored -
-
-
-
* pop3d/*.[ch]: Rename all the pop function pop3d_xx() instead of pop3_xx() to be consistent with imap4d/*. * mailbox/mbx_pop.c (pop_user): This is sudden death; for many pop servers, it is important to let them time to remove locks or move the .user.pop files. This happen when we do BUSY_CHECK(). For example, the user does not want to read the entire file, and wants start to read a new message, closing the connection and immediately contact the server again, and we'll end up having "-ERR Mail Lock busy" or something similar. To prevent this race condition we sleep 2 seconds. 2001-04-23 Sergey Poznyakoff It is often convenient to separte log outputs from POP and SMTP servers. --with-log-facility flag which allows to specify to which log facility the loggin output should be directed. System administrators often prefere to have more information about unsuccessfull authentications. I have added more verbose logging to pop3d/user.c. Both failed attempts and possible account probes (USER immediately followed by QUIT) are logged. Made pop3d to be less verbose about its WELCOME prompt. When the symbol TERSE_MODE is defined, pop3d introduces itself just as +OK POP3 ready <apop_hash> insead of divulging its type and version. This is a bit paranoid, but it is better to be on the safe side...
Alain Magloire authored
-
- 23 Apr, 2001 3 commits
-
-
-
* configure.in: Check for sigaction(). * configuire.in: AC_DEFINE(_REENTRANT). * doc/Makefile.am: add rfc2060-errata. * doc/rfc2060-errata: New File. * imap4d/copy.c: First Implementation. * imap4d/status.c: First implementation. * mailbox/file_stream.c (_file_open): For CREAT, close the fd if we failed. * mailbox/folder_imap.c: Remove the _REENTRANT. * mailbox/mbx_mbox.c: Remove the _REENTRANT. * mailbox/monitor.c: Remove the _REENTRANT. * pop3d/pop3d.c (main): maxchildren boosted to 20. (pop3_daemon_init): Use sigaction() for SIGCHLD, the sematics of signal() is unreliable on some platforms. * pop3d/signal.c: Save errno.
Alain Magloire authored -
if _REENTRANT is set or not. So for enable thread we take the approach of always defining _REENTRANT. He also noted that when in standalone the child was not exiting.
Alain Magloire authored
-
- 22 Apr, 2001 1 commit
-
-
* imap4d/delete.c: First implementation. * imap4d/list.c: First implementation. * imap4d/rename.c: First implementation. * imap4d/util.c: New functions. (util_tilde_expansion): expand ~. (util_unquote): Remove surrounding double quotes. (util_getfullpath): Expand the path to absolute. * imap4d/fetch.c: Did not cycle throught the items. (fetch_send_address): Patch from Sam Roberts, did not use the index. * mailbox/attachement.c (message_unencapsulate): Little buglet strncasemp() passing the wrong size.
Alain Magloire authored
-
- 20 Apr, 2001 4 commits
-
-
-
-
-
we must maintain the offset. * imap4d/list.c: partially implemented the LIST command. * imap4d/imap4d.c: homedir new global variable. * imap4d/login.c: same ${HOME} in homedir. * imap4d/util.c (util_getword): Be aware of the quotes. * lib/fnmatch.c: New replacement file. * lib/fnmatch.h: New replacement file. * configure.in: Check for fnmatch().
Alain Magloire authored
-
- 18 Apr, 2001 1 commit
-
-
an email excerpt: states is the valid states for a command (eg. LOGIN is only valid in non-authenticated state, SELECT is valid in authenticated or selected). success is the state to enter when the command completes successfully (STATE_NONE means no change) and failure is the state to enter when the command fails. The fetch code is getting close to completion! ye!
Alain Magloire authored
-
- 17 Apr, 2001 1 commit
-
-
- 16 Apr, 2001 1 commit
-
-
for -lpthread since in libc the thread functions are defined but all return ENOSYS, you need to explicitely link with -lphtread. the ctype functions is*() arguments should be explicitely cast since Solaris use them as indexes. __REENTRANT as to be define if compile with support for threads. Never realize this but setenv() is a BSD/GNU thing, so took a variant from libiberty to cope.
Alain Magloire authored
-