instead of a single hard-coded branch, use a generic method to split
ranges as needed.
this is of course entirely over-engineered as of now, but subsequent
commits will make good use of it.
turns out the comment advising against it was bogus - unlike for
memcmp(), the standard does indeed prescribe that the memchr()
implementation may not read past the first occurrence of the searched
char.
that's what the sources already assumed anyway. size_t is total
overkill, as No Email Ever (TM) will exceed 2GiB.
this also fixes a harmless format string warning in 32 bit builds.
if AuthMechs includes more than just LOGIN and the server announces any
AUTH= mechanism, we try SASL. but that can still fail to find any
suitable authentication mechanism, and we must not error out in that
case if we are supposed to fall back to LOGIN.
specifically, if AuthMechs included more than just LOGIN (which would be
the case for '*') and the server announced any AUTH= mechanism, we'd
immediately error out upon seeing it, thus failing to actually try
LOGIN.
the number was chosen to make queries more comprehensible when the
server sends no UIDNEXT, but it appears that such insanely large UIDs
actually show up in the wild. so send 32-bit INT_MAX instead.
note that this is again making an assumption: that no server uses
unsigned ints for UIDs. but we can't sent UINT_MAX, as that would break
with servers which use signed ints. also, *we* use signed ints (which is
actually a clear violation of the spec).
it would be possible to special-case the range [1,inf] to 1:*, thus
entirely removing arbitrary limits. however, when the range doesn't
start at 1, we may actually get a single message instead of none due to
the imap uid range limits being unordered. this gets really nasty when
we need to issue multiple queries, as we may list the same message
twice.
a reliable way around this would be issuing a separate query to find the
actual value of UID '*', to make up for the server not sending UIDNEXT
in the first place. this would obviously imply an additional round-trip
per mailbox ...
trashing many messages at once inevitably overtaxes m$ exchange, and the
connection breaks. without any progress tracking, it would restart from
scratch each time, which would lead to a) it never finishing and b) many
copies of the messages in the trash.
full transactions as we do for "proper" syncing would be over the top,
as it's not *that* bad if some messages get duplicated in the trash. so
we record only the messages for which trashing completed, thus allowing
some overlap between the attempts.
turns out i misread the spec in a subtle way: while all other folders
are physically nested under INBOX, the IMAP view puts them at the same
(root) level. to get them shown as subfolders of INBOX, they need to
have _two_ leading dots.
this also implies that the Maildir++ mode has no use for a Path, so
reject attempts to specify one.
the mbsync manual says explicitly that the system's default certificate
store should *not* be specified.
however, the isync manual talked about CA certificates, which is (and
always was) exactly wrong.
also adjust both .sample rc files.
flock() may be implemented via fcntl(), which may cause the process to
deadlock itself when trying to apply both types of locks. this is the
case even on linux when the file lives on NFS.
it's unlikely that anything except mbsync would try to access the
.uidvalidity files anyway, so there is no point in trying to be
compatible with anything else ...
REFMAIL: uddy4g589ym.fsf@eismej-u14.spgear.lab.emc.com
it is legal for an email system to simply change the case of rfc2822
headers, and at least one imap server apparently does just that.
this would lead to us not finding our own header, which is obviously not
helpful.
REFMAIL: CA+fD2U3hJEszmvwBsXEpTsaWgJ2Dh373mCESM3M0kg3ZwAYjaw@mail.gmail.com
that pattern may very well expand to INBOXNOT, which would naturally
live under Path, so we need to look into the Path. of course, this
actually makes sense only if there *is* a Path, and complaining about
it being absent is backwards.
the idea that this is even possible was based on an incomplete reading
of the imap spec.
however, the infrastructure for supporting multi-char delimiters as such
is retained, as the Flatten option can be used with them.
recycling server connections skips everything up to setting up the
prefix (Path/NAMESPACE). "everything" should obviously include enabling
compression, as that must be done at most once per connection.
any structures may be invalid after callback invocation.
this has the side effect that the socket write callback now returns
void, like all other callbacks do.
the synchronous writing to the socket would have typically invoked the
write callback, which would flush further commands, thus recursing.
we take the easy way out and make it fully asynchronous, i.e., no data
is sent before (re-)entering the event loop.
this also has the effect that socket_write() cannot fail any more, and
any errors will be reported asynchronously. this is consistent with
socket_read(), and produces cleaner code.
this introduces a marginal performance regression: the maildir driver is
synchronous, so all messages (which fit into memory) will be read before
any data is sent. this is not considered relevant.
the legacy style is a poorly executed attempt at Maildir++, so introduce
the latter for the sake of completeness. but most users will probably
just want to use subfolders without any additional dots.
move the unconditional addition of INBOX out ouf the function.
this makes it possible to move the folder check and addition to the
listing before the recursion, which seems clearer.
in the case of imap stores, the failure is bound to the server config,
not just the store config.
that means that the storage of the failure state needs to be private to
the driver, accessible only through a function.
simply make the code symmetrical to the inverse case.
note that the result will be sort of awkward, as the folders under Path
(and thus the subfolders of Inbox) don't start with a dot, while the
subfolders of these folders do. this needs to be addressed separately.
when we run into Inbox while listing Path, check whether Inbox is being
listed anyway, and just skip it if so, instead of listing it right away
and resetting LIST_INBOX (and thus having a calling order dependency).
USER (the authorization identity) specifies whom to act for.
AUTHNAME (the authentication identity) specifies who is acting (and
thus whose PASS is being used).
USER is derived from AUTHNAME if omitted, but apparently the
GSS-API module automatically adds the REALM, which is not helpful.
it appears to be common to set both USER and AUTHNAME to the same value,
so let's just do it as well.
REFMAIL: 20150407194807.GA1714@leeloo.kyriasis.com
the PassCmd will be typically non-interactive (or it will use a gui
password agent), so starting a new line just makes the progress counter
uglier. so make it configurable and default to no line break.
- the old meaning of -V[V] was moved to -D{n|N}, as these are really
debugging options.
- don't print the info messages by default; this can be re-enabled with
the -V switch, and is implied by most debug options (it was really
kind of stupid that verbose/debug operation disabled these).
- the sync algo/state debugging can be separately enabled with -Ds now.
... instead of determining them on the fly, because
- it enables early display of totals (to be used soon)
- it enables re-use of the data (to be used at some point)
- the code is less cryptic
note that we leak the data created in main(), consistently with other
configuration-related data.
instead of creating three lists of mailboxes (common, master, slave)
and deriving the mailbox presence from the list being processed, create
a single joined list which contains the presence information.
additionally, the list is sorted alphabetically (with INBOX first),
which is neater.
it helps if the code actually does what the comment above it claims.
clarify it a bit, so i don't get stupid ideas again.
This reverts commit cf6a7b4d18.
propagating many messages from a fast store (typically maildir or a
local IMAP server) to a slow asynchronous store could cause gigabytes of
data being buffered. avoid this by throttling fetches if the target
context reports memory usage above a configurable limit.
REFMAIL: 9737edb14457c71af4ed156c1be0ae59@mpcjanssen.nl
some servers actually bother to close down the SSL connection before
closing the socket.
this fixes the spurious "unhandled SSL error 6" messages.
REFMAIL: 20150120114805.GA17586@leeloo.kyriasis.com
the server can actually close the zlib stream before closing the socket,
so we need to accept it.
we don't do anything beyond that - the actual EOF will be signaled by
the socket, and if the server (erroneously) sends more data, zlib will
tell us about it.
REFMAIL: 1423048708-975-1-git-send-email-alex.bennee@linaro.org
zlib reports Z_BUF_ERROR when a flush is attempted without any activity
since the previous flush (if any). while this is harmless as such,
discerning the condition from genuine errors would be much harder than
avoiding the pointless flush in the first place.
REFMAIL: eb5681612f17be777bc8d138d31dd6d6@mpcjanssen.nl
don't retry dead Stores for every Channel.
this also introduces a state for transient errors (specifically, connect
failures), but this is currently unused.
a directory is no mailbox unless it contains a cur/ subdir.
but if that one is present, create new/ and tmp/ if they are missing.
this makes it possible to resume interrupted maildir creations.
don't try to lock it until we actually read or write it.
the idea is to not fail with SyncState * if we tried to load the state
before selecting a non-existing mailbox. this is ok, because if the
mailbox is missing, we obviously have no sync state pertaining to it,
either.
as a side effect, this allows simplifying an error path.
when LITERAL+ is used, the server has no chance for early rejection of
messages. this means that the client can upload megabytes for nothing.
so simply don't use LITERAL+ for big messages. of course this adds
server roundtrips, but that's tough luck.
the limit could be arguably higher than 100k (or even configurable).
i set it to ~2 sec with my fairly average DSL line.
the primary objective is reducing the number of small SSL packets (which
are always padded), but fewer syscalls in the non-SSL case should be
good as well.
instead of keeping the structures in an opaque array (which was a shadow
of the struct pollfd array if poll() was supported), make them directly
addressable.
this has the advantage that notifier-altering operations (mostly
en-/disabling) don't need to look up the structure by file handle each
time.
on the downside, data locality in the main loop is worse.
neither of these have any real effect on performance.
note that the structures are not allocated separately, but embedded into
the the parent structure (like sockets already were).
the seznam.cz IMAP server seems very eager to send UIDNEXT responses
despite not supporting UIDPLUS. this doesn't appear to be a particularly
sensible combination, but it's valid nonetheless.
however, that means that we need to save the UIDNEXT value before we
start storing messages, lest imap_find_new_msgs() will simply overlook
them. we do that outside the driver, in an already present field - this
actually makes the main path more consistent with the journal recovery
path.
analysis by Tomas Tintera <trosos@seznam.cz>.
REFMAIL: 20141220215032.GA10115@kyvadlo.trosos.seznam.cz
we can't leave the store FRESH, as otherwise the error handling code
will assume it is still being opened and will return to the main loop.
depending on the config this would cause an immediate termination or an
indefinite wait.
this re-introduces 6741bc94 (just a bit differently), thus effectively
reverting fbfcfea5. i suppose this extra CRLF is needed by a broken
CRAM-MD5 implementation of some server, which is why it was there in the
original implementation as well. however, it breaks more pedantic
non-broken servers. if somebody complains, we'll need to add a much
more sophisticated hack.
we should make no assumptions about the layout of OpenSSL's certificate
store, in particular when they are wrong. so copy the interesting part
instead of "deep-linking" into it.
this code is uglier than it should be, but OpenSSL's extensive use of
macros to manage data types would force us to include the ssl headers
into our headers otherwise, which would be even uglier.
REFMAIL: <545442CC.9020400@nodivisions.com>
... for windows fs compatibility.
the maildir-specific InfoDelimiter inherits the global FieldDelimiter
(which affects SyncState), based on the assumption that if the sync
state is on a windows FS, the mailboxes certainly will be as well, while
the inverse is not necessarily true (when running on unix, anyway).
REFMAIL: <CA+m_8J1ynqAjHRJagvKt9sb31yz047Q7NH-ODRmHOKyfru8vtA@mail.gmail.com>
patch initially by Jack Stone <jwjstone@fastmail.fm>,
cleaned up by Jan Synacek <jsynacek@redhat.com>,
... and then almost completely rewritten by me. ^^
RequireCRAM (another fairly stupid "use if available" option) is now
deprecated. instead, the AuthMech option can be used to give a precise
list of acceptable authentication mechanisms (which is currently "a bit"
short). in particular, this allows *not* using CRAM-MD5 even if it's
available.
