1 <?xml version="1.0" encoding="ISO-8859-1"?>
2 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
3 "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
4 <html xmlns="http://www.w3.org/1999/xhtml">
6 <title>Design notes on fetchmail</title>
7 <link rev="made" href="mailto:esr@snark.thyrsus.com" />
8 <meta name="description" content="Design notes on fetchmail." />
9 <meta name="keywords" content="fetchmail, POP, POP2, POP3, IMAP, remote mail" />
10 <style type="text/css">
12 h1.c1 {text-align: center}
17 <table width="100%" cellpadding="0" summary="Canned page header">
19 <td width="30%">Back to <a href="/~esr/index.html">Fetchmail Home Page</a></td>
20 <td width="30%" align="center">To <a href="/~esr/sitemap.html">Site Map</a></td>
21 <td width="30%" align="right">$Date: 2002/07/30 13:31:18 $</td>
26 <h1 class="c1">Design Notes On Fetchmail</h1>
28 <p>These notes are for the benefit of future hackers and
29 maintainers. The following sections are both functional and
30 narrative, read from beginning to end.</p>
34 <p>A direct ancestor of the fetchmail program was originally
35 authored (under the name popclient) by Carl Harris
36 <ceharris@mal.com>. I took over development in June 1996 and
37 subsequently renamed the program `fetchmail' to reflect the
38 addition of IMAP support and SMTP delivery. In early November 1996
39 Carl officially ended support for the last popclient versions.</p>
41 <p>Before accepting responsibility for the popclient sources from
42 Carl, I had investigated and used and tinkered with every other
43 UNIX remote-mail forwarder I could find, including fetchpop1.9,
44 PopTart-0.9.3, get-mail, gwpop, pimp-1.0, pop-perl5-1.2, popc,
45 popmail-1.6 and upop. My major goal was to get a header-rewrite
46 feature like fetchmail's working so I wouldn't have reply problems
49 <p>Despite having done a good bit of work on fetchpop1.9, when I
50 found popclient I quickly concluded that it offered the solidest
51 base for future development. I was convinced of this primarily by
52 the presence of multiple-protocol support. The competition didn't
53 do POP2/RPOP/APOP, and I was already having vague thoughts of maybe
54 adding IMAP. (This would advance two other goals: learn IMAP and
55 get comfortable writing TCP/IP client software.)</p>
57 <p>Until popclient 3.05 I was simply following out the implications
58 of Carl's basic design. He already had daemon.c in the
59 distribution, and I wanted daemon mode almost as badly as I wanted
60 the header rewrite feature. The other things I added were bug fixes
61 or minor extensions.</p>
63 <p>After 3.1, when I put in SMTP-forwarding support (more about
64 this below) the nature of the project changed -- it became a
65 carefully-thought-out attempt to render obsolete every other
66 program in its class. The name change quickly followed.</p>
68 <h1>The rewrite option</h1>
70 <p>MTAs ought to canonicalize the addresses of outgoing non-local
71 mail so that From:, To:, Cc:, Bcc: and other address headers
72 contain only fully qualified domain names. Failure to do so can
73 break the reply function on many mailers. (Sendmail has an option
76 <p>This problem only becomes obvious when a reply is generated on a
77 machine different from where the message was delivered. The two
78 machines will have different local username spaces, potentially
79 leading to misrouted mail.</p>
81 <p>Most MTAs (and sendmail in particular) do not canonicalize
82 address headers in this way (violating RFC 1123). Fetchmail
83 therefore has to do it. This is the first feature I added to the
84 ancestral popclient.</p>
86 <h1>Reorganization</h1>
88 <p>The second thing I did reorganize and simplify popclient a lot.
89 Carl Harris's implementation was very sound, but exhibited a kind
90 of unnecessary complexity common to many C programmers. He treated
91 the code as central and the data structures as support for the
92 code. As a result, the code was beautiful but the data structure
93 design ad-hoc and rather ugly (at least to this old LISP
96 <p>I was able to improve matters significantly by reorganizing most
97 of the program around the `query' data structure and eliminating a
98 bunch of global context. This especially simplified the main
99 sequence in fetchmail.c and was critical in enabling the daemon
102 <h1>IMAP support and the method table</h1>
104 <p>The next step was IMAP support. I initially wrote the IMAP code
105 as a generic query driver and a method table. The idea was to have
106 all the protocol-independent setup logic and flow of control in the
107 driver, and the protocol-specific stuff in the method table.</p>
109 <p>Once this worked, I rewrote the POP3 code to use the same
110 organization. The POP2 code kept its own driver for a couple more
111 releases, until I found sources of a POP2 server to test against
112 (the breed seems to be nearly extinct).</p>
114 <p>The purpose of this reorganization, of course, is to trivialize
115 the development of support for future protocols as much as
116 possible. All mail-retrieval protocols have to have pretty similar
117 logical design by the nature of the task. By abstracting out that
118 common logic and its interface to the rest of the program, both the
119 common and protocol-specific parts become easier to understand.</p>
121 <p>Furthermore, many kinds of new features can instantly be
122 supported across all protocols by modifying the one driver
125 <h1>Implications of smtp forwarding</h1>
127 <p>The direction of the project changed radically when Harry
128 Hochheiser sent me his scratch code for forwarding fetched mail to
129 the SMTP port. I realized almost immediately that a reliable
130 implementation of this feature would make all the other delivery
133 <p>Why mess with all the complexity of configuring an MDA or
134 setting up lock-and-append on a mailbox when port 25 is guaranteed
135 to be there on any platform with TCP/IP support in the first place?
136 Especially when this means retrieved mail is guaranteed to look
137 like normal sender- initiated SMTP mail, which is really what we
140 <p>Clearly, the right thing to do was (1) hack SMTP forwarding
141 support into the generic driver, (2) make it the default mode, and
142 (3) eventually throw out all the other delivery modes.</p>
144 <p>I hesitated over step 3 for some time, fearing to upset
145 long-time popclient users dependent on the alternate delivery
146 mechanisms. In theory, they could immediately switch to .forward
147 files or their non-sendmail equivalents to get the same effects. In
148 practice the transition might have been messy.</p>
150 <p>But when I did it (see the NEWS note on the great options
151 massacre) the benefits proved huge. The cruftiest parts of the
152 driver code vanished. Configuration got radically simpler -- no
153 more grovelling around for the system MDA and user's mailbox, no
154 more worries about whether the underlying OS supports file
157 <p>Also, the only way to lose mail vanished. If you specified
158 localfolder and the disk got full, your mail got lost. This can't
159 happen with SMTP forwarding because your SMTP listener won't return
160 OK unless the message can be spooled or processed.</p>
162 <p>Also, performance improved (though not so you'd notice it in a
163 single run). Another not insignificant benefit of this change was
164 that the manual page got a lot simpler.</p>
166 <p>Later, I had to bring --mda back in order to allow handling of
167 some obscure situations involving dynamic SLIP. But I found a much
168 simpler way to do it.</p>
170 <p>The moral? Don't hesitate to throw away superannuated features
171 when you can do it without loss of effectiveness. I tanked a couple
172 I'd added myself and have no regrets at all. As Saint-Exupery said,
173 "Perfection [in design] is achieved not when there is nothing more
174 to add, but rather when there is nothing more to take away." This
175 program isn't perfect, but it's trying.</p>
177 <h1>The most-requested features that I will never add, and why
180 <h2>Password encryption in .fetchmailrc</h2>
182 <p>The reason there's no facility to store passwords encrypted in
183 the .fetchmailrc file is because this doesn't actually add
186 <p>Anyone who's acquired the 0600 permissions needed to read your
187 .fetchmailrc file will be able to run fetchmail as you anyway --
188 and if it's your password they're after, they'd be able to rip the
189 necessary decoder out of the fetchmail code itself to get it.</p>
191 <p>All .fetchmailrc encryption would do is give a false sense of
192 security to people who don't think very hard.</p>
194 <h2>Truly concurrent queries to multiple hosts</h2>
196 <p>Occasionally I get a request for this on "efficiency" grounds.
197 These people aren't thinking either. True concurrency would do
198 nothing to lessen fetchmail's total IP volume. The best it could
199 possibly do is change the usage profile to shorten the duration of
200 the active part of a poll cycle at the cost of increasing its
201 demand on IP volume per unit time.</p>
203 <p>If one could thread the protocol code so that fetchmail didn't
204 block on waiting for a protocol response, but rather switched to
205 trying to process another host query, one might get an efficiency
206 gain (close to constant loading at the single-host level).</p>
208 <p>Fortunately, I've only seldom seen a server that incurred
209 significant wait time on an individual response. I judge the gain
210 from this not worth the hideous complexity increase it would
211 require in the code.</p>
213 <h2>Multiple concurrent instances of fetchmail</h2>
215 <p>Fetchmail locking is on a per-invoking-user because
216 finer-grained locks would be really hard to implement in a portable
217 way. The problem is that you don't want two fetchmails querying the
218 same site for the same remote user at the same time.</p>
220 <p>To handle this optimally, multiple fetchmails would have to
221 associate a system-wide semaphore with each active pair of a remote
222 user and host canonical address. A fetchmail would have to block
223 until getting this semaphore at the start of a query, and release
224 it at the end of a query.</p>
226 <p>This would be way too complicated to do just for an "it might be
227 nice" feature. Instead, you can run a single root fetchmail polling
228 for multiple users in either single-drop or multidrop mode.</p>
230 <p>The fundamental problem here is how an instance of fetchmail
231 polling host foo can assert that it's doing so in a way visible to
232 all other fetchmails. System V semaphores would be ideal for this
233 purpose, but they're not portable.</p>
235 <p>I've thought about this a lot and roughed up several designs.
236 All are complicated and fragile, with a bunch of the standard
237 problems (what happens if a fetchmail aborts before clearing its
238 semaphore, and how do we recover reliably?).</p>
240 <p>I'm just not satisfied that there's enough functional gain here
241 to pay for the large increase in complexity that adding these
242 semaphores would entail.</p>
244 <h1>Multidrop and alias handling</h1>
246 <p>I decided to add the multidrop support partly because some users
247 were clamoring for it, but mostly because I thought it would shake
248 bugs out of the single-drop code by forcing me to deal with
249 addressing in full generality. And so it proved.</p>
251 <p>There are two important aspects of the features for handling
252 multiple-drop aliases and mailing lists which future hackers should
253 be careful to preserve.</p>
257 <p>The logic path for single-recipient mailboxes doesn't involve
258 header parsing or DNS lookups at all. This is important -- it means
259 the code for the most common case can be much simpler and more
264 <p>The multidrop handing does <em>not</em> rely on doing the
265 equivalent of passing the message to sendmail -oem -t. Instead, it
266 explicitly mines members of a specified set of local usernames out
271 <p>We do <em>not</em> attempt delivery to multidrop mailboxes in
272 the presence of DNS errors. Before each multidrop poll we probe DNS
273 to see if we have a nameserver handy. If not, the poll is skipped.
274 If DNS crashes during a poll, the error return from the next
275 nameserver lookup aborts message delivery and ends the poll. The
276 daemon mode will then quietly spin until DNS comes up again, at
277 which point it will resume delivering mail.</p>
281 <p>When I designed this support, I was terrified of doing anything
282 that could conceivably cause a mail loop (you should be too).
283 That's why the code as written can only append <em>local</em> names
284 (never @-addresses) to the recipients list.</p>
286 <p>The code in mxget.c is nasty, no two ways about it. But it's
287 utterly necessary, there are a lot of MX pointers out there. It
288 really ought to be a (documented!) entry point in the bind
291 <h1>DNS error handling</h1>
293 <p>Fetchmail's behavior on DNS errors is to suppress forwarding and
294 deletion of the individual message that each occurs in, leaving it
295 queued on the server for retrieval on a subsequent poll. The
296 assumption is that DNS errors are transient, due to temporary
299 <p>Unfortunately this means that if a DNS error is permanent a
300 message can be perpetually stuck in the server mailbox. We've had a
301 couple bug reports of this kind due to subtle RFC822 parsing errors
302 in the fetchmail code that resulted in impossible things getting
303 passed to the DNS lookup routines.</p>
305 <p>Alternative ways to handle the problem: ignore DNS errors
306 (treating them as a non-match on the mailserver domain), or forward
307 messages with errors to fetchmail's invoking user in addition to
308 any other recipients. These would fit an assumption that DNS lookup
309 errors are likely to be permanent problems associated with an
312 <h1>IPv6 and IPSEC</h1>
314 <p>The IPv6 support patches are really more protocol-family
315 independence patches. Because of this, in most places, "ports"
316 (numbers) have been replaced with "services" (strings, that may be
317 digits). This allows us to run with certain protocols that use
318 strings as "service names" where we in the IP world think of port
319 numbers. Someday we'll plumb strings all over and then, if inet6 is
320 not enabled, do a getservbyname() down in SocketOpen. The IPv6
321 support patches use getaddrinfo(), which is a POSIX p1003.1g
322 mandated function. So, in the not too distant future, we'll zap the
323 ifdefs and just let autoconf check for getaddrinfo. IPv6 support
324 comes pretty much automatically once you have protocol family
327 <h1>Internationalization</h1>
329 <p>Internationalization is handled using GNU gettext (see the file
330 ABOUT_NLS in the source distribution). This places some minor
331 constraints on the code.</p>
333 <p>Strings that must be subject to translation should be wrapped
334 with GT_() or N_() -- the former in function arguments, the latter
335 in static initializers and other non-function-argument
338 <h1>Checklist for Adding Options</h1>
340 <p>Adding a control option is not complicated in principle, but
341 there are a lot of fiddly details in the process. You'll need to do
342 the following minimum steps.</p>
345 <li>Add a field to represent the control in <code>struct
346 run</code>, <code>struct query</code>, or <code>struct
347 hostdata</code>.</li>
349 <li>Go to <code>rcfile_y.y</code>. Add the token to the grammar.
350 Don't forget the <code>%token</code> declaration.</li>
352 <li>Pick an actual string to declare the option in the .fetchmailrc
353 file. Add the token to <code>rcfile_l</code>.</li>
355 <li>Pick a long-form option name, and a one-letter short option if
356 any are left. Go to <code>options.c</code>. Pick a new
357 <code>LA_</code> value. Hack the <code>longoptions</code> table to
358 set up the association. Hack the big switch statement to set the
359 option. Hack the `?' message to describe it.</li>
361 <li>If the default is nonzero, set it in <code>def_opts</code> near
362 the top of <code>load_params</code> in
363 <code>fetchmail.c</code>.</li>
365 <li>Add code to dump the option value in
366 <code>fetchmail.c:dump_params</code>.</li>
368 <li>For a per-site or per-user option, add proper
369 <code>FLAG_MERGE</code> actions in fetchmail.c's optmerge()
370 function. For a global option, add an override at the end of
371 load_params; this will involve copying a "cmd_run." field to a
372 corresponding "run." field, see the existing code for models.</li>
374 <li>Document the option in fetchmail.man. This will require at
375 least two changes; one to the collected table of options, and one
376 full text description of the option.</li>
378 <li>Hack fetchmailconf to configure it. Bump the fetchmailconf
381 <li>Hack conf.c to dump the option so we won't have a version-skew
384 <li>Add an entry to NEWS.</li>
386 <li>If the option implements a new feature, add a note to the
390 <p>There may be other things you have to do in the way of logic, of
393 <p>Before you implement an option, though, think hard. Is there any
394 way to make fetchmail automatically detect the circumstances under
395 which it should change its behavior? If so, don't write an option.
396 Just do the check!</p>
398 <h1>Lessons learned</h1>
400 <h3>1. Server-side state is essential</h3>
402 <p>The person(s) responsible for removing LAST from POP3 deserve to
403 suffer. Without it, a client has no way to know which messages in a
404 box have been read by other means, such as an MUA running on the
407 <p>The POP3 UID feature described in RFC1725 to replace LAST is
408 insufficient. The only problem it solves is tracking which messages
409 have been read <em>by this client</em> -- and even that requires
410 tricky, fragile implementation.</p>
412 <p>The underlying lesson is that maintaining accessible server-side
413 `seen' state bits associated with Status headers is indispensible
414 in a Unix/RFC822 mail server protocol. IMAP gets this right.</p>
416 <h3>2. Readable text protocol transactions are a Good Thing</h3>
418 <p>A nice thing about the general class of text-based protocols
419 that SMTP, POP2, POP3, and IMAP belongs to is that client/server
420 transactions are easy to watch and transaction code correspondingly
421 easy to debug. Given a decent layer of socket utility functions
422 (which Carl provided) it's easy to write protocol engines and not
423 hard to show that they're working correctly.</p>
425 <p>This is an advantage not to be despised! Because of it, this
426 project has been interesting and fun -- no serious or persistent
427 bugs, no long hours spent looking for subtle pathologies.</p>
429 <h3>3. IMAP is a Good Thing.</h3>
431 <p>Now that there is a standard IMAP equivalent of the POP3 APOP
432 validation in CRAM-MD5, POP3 is completely obsolete.</p>
434 <h3>4. SMTP is the Right Thing</h3>
436 <p>In retrospect it seems clear that this program (and others like
437 it) should have been designed to forward via SMTP from the
438 beginning. This lesson may be applicable to other Unix programs
439 that now call the local MDA/MTA as a program.</p>
441 <h3>5. Syntactic noise can be your friend</h3>
443 <p>The optional `noise' keywords in the rc file syntax started out
444 as a late-night experiment. The English-like syntax they allow is
445 considerably more readable than the traditional terse keyword-value
446 pairs you get when you strip them all out. I think there may be a
447 wider lesson here.</p>
449 <h1>Motivation and validation</h1>
451 <p>It is truly written: the best hacks start out as personal
452 solutions to the author's everyday problems, and spread because the
453 problem turns out to be typical for a large class of users. So it
454 was with Carl Harris and the ancestral popclient, and so with me
457 <p>It's gratifying that fetchmail has become so popular. Until just
458 before 1.9 I was designing strictly to my own taste. The multi-drop
459 mailbox support and the new --limit option were the first features
460 to go in that I didn't need myself.</p>
462 <p>By 1.9, four months after I started hacking on popclient and a
463 month after the first fetchmail release, there were literally a
464 hundred people on the fetchmail-friends contact list. That's pretty
465 powerful motivation. And they were a good crowd, too, sending fixes
466 and intelligent bug reports in volume. A user population like that
467 is a gift from the gods, and this is my expression of
470 <p>The beta testers didn't know it at the time, but they were also
471 the subjects of a sociological experiment. The results are
472 described in my paper, <a
473 href="//www.tuxedo.org/~esr/writings/cathedral-bazaar/">The
474 Cathedral And The Bazaar</a>.</p>
478 <p>Special thanks go to Carl Harris, who built a good solid code
479 base and then tolerated me hacking it out of recognition. And to
480 Harry Hochheiser, who gave me the idea of the SMTP-forwarding
483 <p>Other significant contributors to the code have included Dave
484 Bodenstab (error.c code and --syslog), George Sipe (--monitor and
485 --interface), Gordon Matzigkeit (netrc.c), Al Longyear (UIDL
486 support), Chris Hanson (Kerberos V4 support), and Craig Metz (OPIE,
491 <p>At this point, the fetchmail code appears to be pretty stable.
492 It will probably undergo substantial change only if and when
493 support for a new retrieval protocol or authentication method is
496 <h1>Relevant RFCS</h1>
498 <p>Not all of these describe standards explicitly used in
499 fetchmail, but they all shaped the design in one way or
503 <dt><a href="ftp://ftp.isi.edu/in-notes/rfc821.txt">RFC821</a></dt>
505 <dd>SMTP protocol</dd>
507 <dt><a href="ftp://ftp.isi.edu/in-notes/rfc822.txt">RFC822</a></dt>
509 <dd>Mail header format</dd>
511 <dt><a href="ftp://ftp.isi.edu/in-notes/rfc937.txt">RFC937</a></dt>
513 <dd>Post Office Protocol - Version 2</dd>
515 <dt><a href="ftp://ftp.isi.edu/in-notes/rfc974.txt">RFC974</a></dt>
519 <dt><a href="ftp://ftp.isi.edu/in-notes/rfc976.txt">RFC976</a></dt>
521 <dd>UUCP mail format</dd>
524 href="ftp://ftp.isi.edu/in-notes/rfc1081.txt">RFC1081</a></dt>
526 <dd>Post Office Protocol - Version 3</dd>
529 href="ftp://ftp.isi.edu/in-notes/rfc1123.txt">RFC1123</a></dt>
531 <dd>Host requirements (modifies 821, 822, and 974)</dd>
534 href="ftp://ftp.isi.edu/in-notes/rfc1176.txt">RFC1176</a></dt>
536 <dd>Interactive Mail Access Protocol - Version 2</dd>
539 href="ftp://ftp.isi.edu/in-notes/rfc1203.txt">RFC1203</a></dt>
541 <dd>Interactive Mail Access Protocol - Version 3</dd>
544 href="ftp://ftp.isi.edu/in-notes/rfc1225.txt">RFC1225</a></dt>
546 <dd>Post Office Protocol - Version 3</dd>
549 href="ftp://ftp.isi.edu/in-notes/rfc1344.txt">RFC1344</a></dt>
551 <dd>Implications of MIME for Internet Mail Gateways</dd>
554 href="ftp://ftp.isi.edu/in-notes/rfc1413.txt">RFC1413</a></dt>
556 <dd>Identification server</dd>
559 href="ftp://ftp.isi.edu/in-notes/rfc1428.txt">RFC1428</a></dt>
561 <dd>Transition of Internet Mail from Just-Send-8 to 8-bit
565 href="ftp://ftp.isi.edu/in-notes/rfc1460.txt">RFC1460</a></dt>
567 <dd>Post Office Protocol - Version 3</dd>
570 href="ftp://ftp.isi.edu/in-notes/rfc1508.txt">RFC1508</a></dt>
572 <dd>Generic Security Service Application Program Interface</dd>
575 href="ftp://ftp.isi.edu/in-notes/rfc1521.txt">RFC1521</a></dt>
577 <dd>MIME: Multipurpose Internet Mail Extensions</dd>
580 href="ftp://ftp.isi.edu/in-notes/rfc1869.txt">RFC1869</a></dt>
582 <dd>SMTP Service Extensions (ESMTP spec)</dd>
585 href="ftp://ftp.isi.edu/in-notes/rfc1652.txt">RFC1652</a></dt>
587 <dd>SMTP Service Extension for 8bit-MIMEtransport</dd>
590 href="ftp://ftp.isi.edu/in-notes/rfc1725.txt">RFC1725</a></dt>
592 <dd>Post Office Protocol - Version 3</dd>
595 href="ftp://ftp.isi.edu/in-notes/rfc1730.txt">RFC1730</a></dt>
597 <dd>Interactive Mail Access Protocol - Version 4</dd>
600 href="ftp://ftp.isi.edu/in-notes/rfc1731.txt">RFC1731</a></dt>
602 <dd>IMAP4 Authentication Mechanisms</dd>
605 href="ftp://ftp.isi.edu/in-notes/rfc1732.txt">RFC1732</a></dt>
607 <dd>IMAP4 Compatibility With IMAP2 And IMAP2bis</dd>
610 href="ftp://ftp.isi.edu/in-notes/rfc1734.txt">RFC1734</a></dt>
612 <dd>POP3 AUTHentication command</dd>
615 href="ftp://ftp.isi.edu/in-notes/rfc1870.txt">RFC1870</a></dt>
617 <dd>SMTP Service Extension for Message Size Declaration</dd>
620 href="ftp://ftp.isi.edu/in-notes/rfc1891.txt">RFC1891</a></dt>
622 <dd>SMTP Service Extension for Delivery Status Notifications</dd>
625 href="ftp://ftp.isi.edu/in-notes/rfc1892.txt">RFC1892</a></dt>
627 <dd>The Multipart/Report Content Type for the Reporting of Mail
628 System Administrative Messages</dd>
631 href="ftp://ftp.isi.edu/in-notes/rfc1894.txt">RFC1894</a></dt>
633 <dd>An Extensible Message Format for Delivery Status
637 href="ftp://ftp.isi.edu/in-notes/rfc1893.txt">RFC1893</a></dt>
639 <dd>Enhanced Mail System Status Codes</dd>
642 href="ftp://ftp.isi.edu/in-notes/rfc1894.txt">RFC1894</a></dt>
644 <dd>An Extensible Message Format for Delivery Status
648 href="ftp://ftp.isi.edu/in-notes/rfc1938.txt">RFC1938</a></dt>
650 <dd>A One-Time Password System</dd>
653 href="ftp://ftp.isi.edu/in-notes/rfc1939.txt">RFC1939</a></dt>
655 <dd>Post Office Protocol - Version 3</dd>
658 href="ftp://ftp.isi.edu/in-notes/rfc1957.txt">RFC1957</a></dt>
660 <dd>Some Observations on Implementations of the Post Office
664 href="ftp://ftp.isi.edu/in-notes/rfc1985.txt">RFC1985</a></dt>
666 <dd>SMTP Service Extension for Remote Message Queue Starting</dd>
669 href="ftp://ftp.isi.edu/in-notes/rfc2033.txt">RFC2033</a></dt>
671 <dd>Local Mail Transfer Protocol</dd>
674 href="ftp://ftp.isi.edu/in-notes/rfc2060.txt">RFC2060</a></dt>
676 <dd>Internet Message Access Protocol - Version 4rev1</dd>
679 href="ftp://ftp.isi.edu/in-notes/rfc2061.txt">RFC2061</a></dt>
681 <dd>IMAP4 Compatibility With IMAP2bis</dd>
684 href="ftp://ftp.isi.edu/in-notes/rfc2062.txt">RFC2062</a></dt>
686 <dd>Internet Message Access Protocol - Obsolete Syntax</dd>
689 href="ftp://ftp.isi.edu/in-notes/rfc2195.txt">RFC2195</a></dt>
691 <dd>IMAP/POP AUTHorize Extension for Simple Challenge/Response</dd>
694 href="ftp://ftp.isi.edu/in-notes/rfc2177.txt">RFC2177</a></dt>
696 <dd>IMAP IDLE command</dd>
699 href="ftp://ftp.isi.edu/in-notes/rfc2449.txt">RFC2449</a></dt>
701 <dd>POP3 Extension Mechanism</dd>
704 href="ftp://ftp.isi.edu/in-notes/rfc2554.txt">RFC2554</a></dt>
706 <dd>SMTP Service Extension for Authentication</dd>
709 href="ftp://ftp.isi.edu/in-notes/rfc2595.txt">RFC2595</a></dt>
711 <dd>Using TLS with IMAP, POP3 and ACAP</dd>
714 href="ftp://ftp.isi.edu/in-notes/rfc2645.txt">RFC2645</a></dt>
716 <dd>On-Demand Mail Relay: SMTP with Dynamic IP Addresses</dd>
719 href="ftp://ftp.isi.edu/in-notes/rfc2683.txt">RFC2683</a></dt>
721 <dd>IMAP4 Implementation Recommendations</dd>
724 href="ftp://ftp.isi.edu/in-notes/rfc2821.txt">RFC2821</a></dt>
726 <dd>Simple Mail Transfer Protocol</dd>
729 href="ftp://ftp.isi.edu/in-notes/rfc2822.txt">RFC2822</a></dt>
731 <dd>Internet Message Format</dd>
735 RFC2192 IMAP URL Scheme
736 RFC2193 IMAP4 Mailbox Referrals
737 RFC2221 IMAP4 Login Referrals
740 <table width="100%" cellpadding="0" summary="Canned page footer">
742 <td width="30%">Back to <a href="index.html">Fetchmail Home Page</a></td>
743 <td width="30%" align="center">To <a href="/~esr/sitemap.html">Site Map</a></td>
744 <td width="30%" align="right">$Date: 2002/07/30 13:31:18 $</td>
749 <address>Eric S. Raymond <a
750 href="mailto:esr@thyrsus.com"><esr@snark.thyrsus.com></a></address>