docs/master
view ch03.roff @ 93:093ccf39a45e
Write text about removal of switches; Included the grap figure.
| author | markus schnalke <meillo@marmaro.de> |
|---|---|
| date | Fri, 15 Jun 2012 23:37:58 +0200 |
| parents | 83bfb4dbf59f |
| children | edac7e46a9f2 |
line source
1 .H0 "Discussion
2 .P
3 This main chapter discusses the practical work done in the mmh project.
4 It is structured along the goals to achieve. The concrete work done
5 is described in the examples of how the general goals were achieved.
6 The discussion compares the current version of mmh with the state of
7 nmh just before the mmh project started, i.e. Fall 2011.
8 Current changes of nmh will be mentioned only as side notes.
9 .\" XXX where do I discuss the parallel development of nmh?
13 .H1 "Stream-lining
15 .P
16 MH had been considered an all-in-one system for mail handling.
17 The community around nmh has a similar understanding.
18 In fundamental difference, mmh shall be a MUA only.
19 I believe that the development of all-in-one mail systems is obsolete.
20 Today, email is too complex to be fully covered by single projects.
21 Such a project won't be able to excel in all aspects.
22 Instead, the aspects of email should be covered my multiple projects,
23 which then can be combined to form a complete system.
24 Excellent implementations for the various aspects of email exist already.
25 Just to name three examples: Postfix is a specialized MTA,
26 Procmail is a specialized MDA, and Fetchmail is a specialized MRA.
27 I believe that it is best to use such specialized tools instead of
28 providing the same function again as a side-component in the project.
29 .P
30 Doing something well, requires to focus on a small set of specific aspects.
31 Under the assumption that focused development produces better results
32 in the particular area, specialized projects will likely be superior
33 in their field of focus.
34 Hence, all-in-one mail system projects \(en no matter if monolithic
35 or modular \(en will never be the best choice in any of the fields.
36 Even in providing the best consistent all-in-one system they are likely
37 to be beaten by projects that focus only on integrating existing mail
38 components to a homogeneous system.
39 .P
40 The limiting resource in Free Software community development
41 is usually man power.
42 If the development power is spread over a large development area,
43 it becomes even more difficult to compete with the specialists in the
44 various fields.
45 The concrete situation for MH-based mail systems is even tougher,
46 given the small and aged community, including both developers and users,
47 it has.
48 .P
49 In consequence, I believe that the available development resources
50 should be focused on the point where MH is most unique.
51 This is clearly the user interface \(en the MUA.
52 Peripheral parts should be removed to stream-line mmh for the MUA task.
55 .H2 "Removal of the Mail Transfer Facilities
56 .P
57 In contrast to nmh, which also provides mail submission and mail retrieval
58 agents, mmh is a MUA only.
59 This general difference in the view on the character of nmh
60 initiated the development of mmh.
61 Removing the mail transfer facilities had been the first work task
62 in the mmh project.
63 .P
64 The MSA is called \fIMessage Transfer Service\fP (MTS) in nmh.
65 The facility established network connections and spoke SMTP to submit
66 messages for relay to the outside world.
67 This part was implemented by the
68 .Pn post
69 command.
70 The changes in emailing in the last years
71 demanded changes in this part of nmh too.
72 Encryption and authentication for network connections
73 needed to be supported, hence TLS and SASL were introduced into nmh.
74 This added complexity to nmh without improving it in its core functions.
75 Also, keeping up with recent developments in the field of
76 mail transfer requires development power and specialists.
77 In mmh this whole facility was simply cut off.
78 .Ci f6aa95b724fd8c791164abe7ee5468bf5c34f226
79 .Ci fecd5d34f65597a4dfa16aeabea7d74b191532c3
80 .Ci 156d35f6425bea4c1ed3c4c79783dc613379c65b
81 Instead, mmh depends on an external MSA.
82 The only outgoing interface available to mmh is the
83 .Pn sendmail
84 command, which almost any MSA provides.
85 If not, a wrapper program can be written.
86 It must read the message from the standard input, extract the
87 recipient addresses from the message header, and hand the message
88 over to the MSA.
89 For example, a wrapper script for qmail would be:
90 .VS
91 #!/bin/sh
92 # ignore command line arguments
93 exec qmail-inject
94 VE
95 The requirement to parse the recipient addresses out of the message header
96 is likely to be removed in the future.
97 Then mmh would give the recipient addresses as command line arguments.
98 This is clearly the better interface, but mmh does not provide it yet.
99 .\" XXX implement it
100 .P
101 To retrieve mail, the
102 .Pn inc
103 command established network connections
104 and spoke POP3 to retrieve mail from remote servers.
105 As with mail submission, the network connections required encryption and
106 authentication, thus TLS and SASL were added.
107 Support for message retrieval through IMAP will become necessary
108 to be added soon, too, and so on for any changes in mail transfer.
109 Mmh has dropped the support for retrieving mail from remote locations.
110 .Ci ab7b48411962d26439f92f35ed084d3d6275459c
111 Instead, it depends on an external tool to cover this task.
112 In mmh there exist two paths for messages to enter mmh's mail storage:
113 (1) Mail can be incorporate with
114 .Pn inc
115 from the system maildrop, or (2) with
116 .Pn rcvstore
117 by reading them, one at a time, from the standard input.
118 .P
119 With the removal of the MSA and MRA, mmh converted from an all-in-one
120 mail system to being a MUA only.
121 Following the Unix philosophy, it now focuses on one job and
122 tries to do that one well.
123 Not only the programs follow that tenet but also the project itself does so.
124 Now, of course, mmh depends on third-party software.
125 An external MSA is required to transfer mail to the outside world;
126 an external MRA is required to retrieve mail from remote machines.
127 There exist excellent implementations of such software,
128 which do this specific task likely better than the internal
129 versions had done it.
130 Also, the best suiting programs can be freely chosen.
131 .P
132 As it had already been possible to use an external MSA or MRA,
133 why not keep the internal version for convenience?
134 The question whether there is sense in having a fall-back pager in all
135 the command line tools, for the cases when
136 .Pn more
137 or
138 .Pn less
139 aren't available, appears to be ridiculous.
140 Now, an MSA or MRA is more complex than a text pager
141 and not necessarily available but still the concept of orthogonal
142 design holds: ``Write programs that do one thing and do it well.''
143 .[
144 mcilroy unix phil
145 p. 53
146 .]
147 .[
148 mcilroy bstj foreword
149 .]
150 Here, this part of the Unix philosophy was applied not only
151 to the programs but to the project itself.
152 In other words:
153 ``Develop projects that focus on one thing and do it well.''
154 Projects grown complex should be split for the same reasons programs grown
155 complex should be split.
156 If it is conceptionally more elegant to have the MSA and MRA
157 separate projects then they should be separated.
158 This is the case here, in my opinion.
159 The RFCs propose this separation by clearly distinguishing the different
160 mail handling tasks.
161 .[
162 rfc 821
163 .]
164 The small interfaces between the mail agents support the separation.
165 .P
166 In the beginning, email had been small and simple.
167 (\c
168 .Pn /bin/mail
169 had once covered anything there was to email and still had been small
170 and simple.)
171 Then the essential complexity of email increased.
172 (Essential complexity is the complexity defined by the problem itself.\0
173 .[[
174 brooks no silver bullet
175 .]])
176 Email systems reacted to this change: They grew.
177 RFCs started to introduce mail agents and separated the various tasks
178 because the existing tasks became more extensive and new tasks appeared.
179 Again, email systems grew, or they split parts off.
180 In nmh, for instance, the POP server, which the original MH had included,
181 was removed.
182 Now is the time to go one step further and remove the MSA and MRA, too.
183 Not only does this decrease the code size of the project,
184 but, more important, it unburdens mmh of the whole field of
185 message transfer with all its implications for the project.
186 There's no more need to concern with changes in network transfer.
187 This independence is received by depending on an external program
188 that covers the field.
189 Today, this is a reasonable exchange.
190 .P
191 Function can be added in three different ways:
192 .BU
193 Implementing the function originally in the project.
194 .BU
195 Depending on a library that provides the function.
196 .BU
197 Depending on a program that provides the function.
198 .P
199 Whereas adding the function originally to the project increases the
200 code size most and requires most maintenance and development work,
201 it makes the project most independent of other software.
202 Using libraries or external programs require less maintenance work
203 but introduces dependencies on external software.
204 Programs have the smallest interfaces and provide the best separation
205 but possibly limit the information exchange.
206 External libraries are stronger connected than external programs,
207 thus information can be exchanged more flexible.
208 Adding code to a project increases maintenance work.
209 .\" XXX ref
210 Implementing complex functions originally in the project will add
211 a lot of code.
212 This should be avoided if possible.
213 Hence, the dependencies only change in kind, not in their existence.
214 In mmh, library dependencies on
215 .Pn libsasl2
216 and
217 .Pn libcrypto /\c
218 .Pn libssl
219 were treated against program dependencies on an MSA and an MRA.
220 This also meant treating build-time dependencies against run-time
221 dependencies.
222 Besides program dependencies providing the stronger separation
223 and being more flexible, they also allowed
224 over 6\|000 lines of code to be removed from mmh.
225 This made mmh's code base about 12\|% smaller.
226 Reducing the project's code size by such an amount without actually
227 losing functionality is a convincing argument.
228 Actually, as external MSAs and MRAs are likely superior to the
229 project's internal versions, the common user even gains functionality.
230 .P
231 Users of MH should not have problems to set up an external MSA and MRA.
232 Also, the popular MSAs and MRAs have large communities and a lot
233 of documentation available.
234 Choices for MSAs range from full-featured MTAs like
235 .I Postfix
236 over mid-size MTAs like
237 .I masqmail
238 and
239 .I dma
240 to small forwarders like
241 .I ssmtp
242 and
243 .I nullmailer .
244 Choices for MRAs include
245 .I fetchmail ,
246 .I getmail ,
247 .I mpop
248 and
249 .I fdm .
252 .H2 "Removal of non-MUA Tools
253 .P
254 One goal of mmh is to remove the tools that are not part of the MUA's task.
255 Further more, any tools that don't improve the MUA's job significantly
256 should be removed.
257 Loosely related and rarely used tools distract from the lean appearance.
258 They require maintenance work without adding much to the core task.
259 On removing these tools, the project shall become more stream-lined
260 and focused.
261 In mmh the following tools are not available anymore:
262 .BU
263 .Pn conflict
264 was removed
265 .Ci 8b235097cbd11d728c07b966cf131aa7133ce5a9
266 because it is a mail system maintenance tool that is not MUA-related.
267 It even checked
268 .Fn /etc/passwd
269 and
270 .Fn /etc/group
271 for consistency, which is completely unrelated to email.
272 A tool like
273 .Pn conflict
274 is surely useful, but it should not be shipped with mmh.
275 .\" XXX historic reasons?
276 .BU
277 .Pn rcvtty
278 was removed
279 .Ci 14767c94b3827be7c867196467ed7aea5f6f49b0
280 because its use case of writing to the user's terminal
281 on receiving of mail is obsolete.
282 If users like to be informed of new mail, the shell's
283 .Ev MAILPATH
284 variable or graphical notifications are technically more appealing.
285 Writing directly to a terminals is hardly ever wanted today.
286 If though one wants to have it this way, the standard tool
287 .Pn write
288 can be used in a way similar to:
289 .VS
290 scan -file - | write `id -un`
291 VE
292 .BU
293 .Pn viamail
294 was removed
295 .Ci eda72d6a7a7c20ff123043fb7f19c509ea01f932
296 when the new attachment system was activated, because
297 .Pn forw
298 could then cover the task itself.
299 The program
300 .Pn sendfiles
301 was rewritten as a shell script wrapper around
302 .Pn forw .
303 .Ci 0e82199cf3c991a173e0ac8aa776efdb3ded61e6
304 .BU
305 .Pn msgchk
306 was removed
307 .Ci bb9360ead7eb7a3fedcce2eeedfc660014e41dbe ,
308 because it lost its use case when POP support was removed.
309 A call to
310 .Pn msgchk
311 provided hardly more information than:
312 .VS
313 ls -l /var/mail/meillo
314 VE
315 It did distinguished between old and new mail, but
316 this detail information and can be retrieved with
317 .Pn stat (1),
318 too.
319 A very small shell script could be written to output the information
320 in a similar way, if truly necessary.
321 As mmh's
322 .Pn inc
323 only incorporates mail from the user's local maildrop,
324 and thus no data transfers over slow networks are involved,
325 there's hardly any need to check for new mail before incorporating it.
326 .BU
327 .Pn msh
328 was removed
329 .Ci 916690191222433a6923a4be54b0d8f6ac01bd02
330 because the tool was in conflict with the philosophy of MH.
331 It provided an interactive shell to access the features of MH,
332 but it wasn't just a shell, tailored to the needs of mail handling.
333 Instead it was one large program that had several MH tools built in.
334 This conflicts with the major feature of MH of being a tool chest.
335 .Pn msh 's
336 main use case had been accessing Bulletin Boards, which have seized to
337 be popular.
338 .P
339 Removing
340 .Pn msh ,
341 together with the truly archaic code relicts
342 .Pn vmh
343 and
344 .Pn wmh ,
345 saved more than 7\|000 lines of C code \(en
346 about 15\|% of the project's original source code amount.
347 .P
348 Having less code (with equal readability, of course)
349 for the same functionality is an advantage.
350 Less code means less bugs and less maintenance work.
351 As
352 .Pn rcvtty
353 and
354 .Pn msgchk
355 are assumed to be rarely used and can be implemented in different ways,
356 why should one keep them?
357 Removing them stream-lines mmh.
358 .Pn viamail 's
359 use case is now partly obsolete and partly covered by
360 .Pn forw ,
361 hence there's no reason to still maintain it.
362 .Pn conflict
363 is not related to the mail client, and
364 .Pn msh
365 conflicts with the basic concept of MH.
366 Theses two tools might still be useful, but they should not be part of mmh.
367 .P
368 Finally, there's
369 .Pn slocal .
370 .Pn slocal
371 is an MDA and thus not directly MUA-related.
372 It should be removed, because including it is a violation
373 of the idea that mmh is a MUA only.
374 It should become a separate project.
375 However,
376 .Pn slocal
377 provides rule-based processing of messages, like filing them into
378 different folders, which is otherwise not available in mmh.
379 Although
380 .Pn slocal
381 does neither pull in dependencies nor does it include a separate
382 technical area (cf. Sec. XXX),
383 still it accounts for about 1\|000 lines of code that need to be maintained.
384 As
385 .Pn slocal
386 is almost self-standing, it should be split off into a separate project.
387 This would cut the strong connection between the MUA mmh and the MDA
388 .Pn slocal .
389 For anyone not using MH,
390 .Pn slocal
391 would become yet another independent MDA, like
392 .I procmail .
393 The need to install a complete MH system to have
394 .Pn slocal
395 would be gone.
396 Likewise, mmh users could decide to use
397 .I procmail
398 without having a second, unused MDA,
399 .Pn slocal ,
400 installed.
401 That's conceptionally the best solution.
402 Yet,
403 .Pn slocal
404 is not split off.
405 I feel unsure with removing it from mmh.
406 Hence, I defer the decision over
407 .Pn slocal .
408 In the meanwhile
409 .Pn slocal
410 does not hurt because it is unrelated to the rest of mmh.
413 .H2 "\fLshow\fP and \fPmhshow\fP
414 .P
415 Since the very beginning \(en already in the first concept paper \(en
416 .Pn show
417 had been MH's message display program.
418 .Pn show
419 mapped message numbers and sequences to files and invoked
420 .Pn mhl
421 to have the files formatted.
422 With MIME, this approach wasn't sufficient anymore.
423 MIME messages can consist of multiple parts, some of which aren't
424 directly displayable, further more text content might be encoded in
425 foreign charsets.
426 .Pn show 's
427 understanding of messages and
428 .Pn mhl 's
429 display capabilities couldn't cope with the task any longer.
430 .P
431 Instead of extending these tools, additional tools were written from
432 scratch and added to the MH tool chest.
433 Doing so is encouraged by the tool chest approach.
434 Modular design is a great advantage for extending a system,
435 as new tools can be added without interfering with existing ones.
436 First, the new MIME features were added in form of the single program
437 .Pn mhn .
438 The command
439 .Cl "mhn -show 42
440 would show the MIME message numbered 42.
441 With the 1.0 release of nmh in February 1999, Richard Coleman finished
442 the split of
443 .Pn mhn
444 into a set of specialized tools, which together covered the
445 multiple aspects of MIME.
446 One of them was
447 .Pn mhshow ,
448 which replaced
449 .Cl "mhn -show" .
450 It was capable of displaying MIME messages appropriately.
451 .P
452 From then on, two message display tools were part of nmh,
453 .Pn show
454 and
455 .Pn mhshow .
456 To ease the life of users,
457 .Pn show
458 was extended to automatically hand the job over to
459 .Pn mhshow
460 if displaying the message would be beyond
461 .Pn show 's
462 abilities.
463 In consequence, the user would simply invoke
464 .Pn show
465 (possibly through
466 .Pn next
467 or
468 .Pn prev )
469 and get the message printed with either
470 .Pn show
471 or
472 .Pn mhshow ,
473 whatever was more appropriate.
474 .P
475 Having two similar tools for essentially the same task is redundant.
476 Usually,
477 users wouldn't distinguish between
478 .Pn show
479 and
480 .Pn mhshow
481 in their daily mail reading.
482 Having two separate display programs was therefore mainly unnecessary
483 from a user's point of view.
484 Besides, the development of both programs needed to be in sync,
485 to ensure that the programs behaved in a similar way,
486 because they were used like a single tool.
487 Different behavior would have surprised the user.
488 .P
489 Today, non-MIME messages are rather seen to be a special case of
490 MIME messages, although it's the other way round.
491 As
492 .Pn mhshow
493 had already be able to display non-MIME messages, it appeared natural
494 to drop
495 .Pn show
496 in favor of using
497 .Pn mhshow
498 exclusively.
499 .Ci 4c1efddfd499300c7e74263e57d8aa137e84c853
500 Removing
501 .Pn show
502 is no loss in function, because functionally
503 .Pn mhshow
504 covers it completely.
505 The old behavior of
506 .Pn show
507 can still be emulated with the simple command line:
508 .VS
509 mhl `mhpath c`
510 VE
511 .P
512 For convenience,
513 .Pn mhshow
514 was renamed to
515 .Pn show
516 after
517 .Pn show
518 was gone.
519 It is clear that such a rename may confuse future developers when
520 trying to understand the history.
521 Nevertheless, I consider the convenience on the user's side,
522 to call
523 .Pn show
524 when they want a message to be displayed, to outweigh the inconvenience
525 on the developer's side when understanding the project history.
526 .P
527 To prepare for the transition,
528 .Pn mhshow
529 was reworked to behave more like
530 .Pn show
531 first.
532 (cf. Sec. XXX)
533 Once the tools behaved more alike, the replacing appeared to be
534 even more natural.
535 Today, mmh's new
536 .Pn show
537 became the one single message display program again, with the difference
538 that today it handles MIME messages as well as non-MIME messages.
539 The outcome of the transition is one program less to maintain,
540 no second display program for users to deal with,
541 and less system complexity.
542 .P
543 Still, removing the old
544 .Pn show
545 hurts in one regard: It had been such a simple program.
546 Its lean elegance is missing to the new
547 .Pn show .
548 But there is no chance;
549 supporting MIME demands for higher essential complexity.
552 .H2 "Removal of Configure Options
553 .P
554 Customization is a double-edged sword.
555 It allows better suiting setups, but not for free.
556 There is the cost of code complexity to be able to customize.
557 There is the cost of less tested setups, because there are
558 more possible setups and especially corner-cases.
559 And, there is the cost of choice itself.
560 The code complexity directly affects the developers.
561 Less tested code affects both, users and developers.
562 The problem of choice affects the users, for once by having to
563 choose, but also by complexer interfaces that require more documentation.
564 Whenever options add little advantages, they should be considered for
565 removal.
566 I have reduced the number of project-specific configure options from
567 fifteen to three.
569 .U3 "Mail Transfer Facilities
570 .P
571 With the removal of the mail transfer facilities five configure
572 options vanished:
573 .P
574 The switches
575 .Sw --with-tls
576 and
577 .Sw --with-cyrus-sasl
578 had activated the support for transfer encryption and authentication.
579 This is not needed anymore.
580 .Ci fecd5d34f65597a4dfa16aeabea7d74b191532c3
581 .Ci 156d35f6425bea4c1ed3c4c79783dc613379c65b
582 .P
583 The configure switch
584 .Sw --enable-pop
585 activated the message retrieval facility.
586 The code area that would be conditionally compiled in for TLS and SASL
587 support had been small.
588 The conditionally compiled code area for POP support had been much larger.
589 Whereas the code base changes would only slightly change on toggling
590 TLS or SASL support, it changed much on toggling POP support.
591 The changes in the code base could hardly be overviewed.
592 By having POP support togglable a second code base had been created,
593 one that needed to be tested.
594 This situation is basically similar for the conditional TLS and SASL
595 code, but there the changes are minor and can yet be overviewed.
596 Still, conditional compilation of a code base creates variations
597 of the original program.
598 More variations require more testing and maintenance work.
599 .P
600 Two other options only specified default configuration values:
601 .Sw --with-mts=[smtp|sendmail]
602 defined the default transport service.
603 In mmh this fixed to
604 .Ar sendmail .
605 .Ci f6aa95b724fd8c791164abe7ee5468bf5c34f226
606 With
607 .Sw --with-smtpservers=[server1...]
608 default SMTP servers for the
609 .Ar smtp
610 transport service could be specified.
611 .Ci 128545e06224233b7e91fc4c83f8830252fe16c9
612 Both of them became irrelevant.
614 .U3 "Backup Prefix
615 .P
616 The backup prefix is the string that was prepended to message
617 filenames to tag them as deleted.
618 By default it had been the comma character `\f(CW,\fP'.
619 In July 2000, Kimmo Suominen introduced
620 the configure option
621 .Sw --with-hash-backup
622 to change the default to the hash symbol `\f(CW#\fP'.
623 The choice was probably personal preference, because first, the
624 option was named
625 .Sw --with-backup-prefix.
626 and had the prefix symbol as argument.
627 Because giving the hash symbol as argument caused to many problems
628 for configure,
629 the option was limited to use the hash symbol as the default prefix.
630 This makes me believe, that the choice for the hash was personal preference.
631 Being it related or not, words that start with the hash symbol
632 introduce a comment in the Unix shell.
633 Thus, the command line
634 .Cl "rm #13 #15
635 calls
636 .Pn rm
637 without arguments because the first hash symbol starts the comment
638 that reaches until the end of the line.
639 To delete the backup files,
640 .Cl "rm ./#13 ./#15"
641 needs to be used.
642 Using the hash as backup prefix can be seen as a precaution agains
643 data loss.
644 .P
645 I removed the configure option but added the profile entry
646 .Pe backup-prefix ,
647 which allows to specify an arbitrary string as backup prefix.
648 .Ci 6c40d481d661d532dd527eaf34cebb6d3f8ed086
649 Profile entries are the common method to change mmh's behavior.
650 This change did not remove the choice but moved it to a location where
651 it suited better.
652 .P
653 Eventually, however, the new trash folder concept
654 .Cf "Sec. XXX
655 obsoleted the concept of the backup prefix completely.
656 .Ci 8edc5aaf86f9f77124664f6801bc6c6cdf258173
657 (Well, there still are corner-cases to remove until the backup
658 prefix can be laid to rest, eventually.)
659 .\" FIXME: Do this work in the code!
661 .U3 "Editor and Pager
662 .P
663 The two configure options
664 .CW --with-editor=EDITOR
665 .CW --with-pager=PAGER
666 were used to specify the default editor and pager at configure time.
667 Doing so at configure time made sense in the Eighties,
668 when the set of available editors and pagers varied much across
669 different systems.
670 Today, the situation is more homogeneous.
671 The programs
672 .Pn vi
673 and
674 .Pn more
675 can be expected to be available on every Unix system,
676 as they are specified by POSIX since two decades.
677 (The specifications for
678 .Pn vi
679 and
680 .Pn more
681 appeared in
682 .[
683 posix 1987
684 .]
685 and,
686 .[
687 posix 1992
688 .]
689 respectively.)
690 As a first step, these two tools were hard-coded as defaults.
691 .Ci 5d43a99db70c12a673028c7758c20cbe3e13ef5f
692 Not changed were the
693 .Pe editor
694 and
695 .Pe moreproc
696 profile entries, which allowed the user to override the system defaults.
697 Later, the concept was reworked to respect the standard environment
698 variables
699 .Ev VISUAL
700 and
701 .Ev PAGER
702 if they are set.
703 Today, mmh determines the editor to use in the following order,
704 taking the first available and non-empty item:
705 .IP (1)
706 Environment variable
707 .Ev MMHEDITOR
708 .IP (2)
709 Profile entry
710 .Pe Editor
711 .IP (3)
712 Environment variable
713 .Ev VISUAL
714 .IP (4)
715 Environment variable
716 .Ev EDITOR
717 .IP (5)
718 Command
719 .Pn vi .
720 .P
721 .Ci f85f4b7ae62e3d05a945dcd46ead51f0a2a89a9b
722 .P
723 The pager to use is determined in a similar order,
724 also taking the first available and non-empty item:
725 .IP (1)
726 Environment variable
727 .Ev MMHPAGER
728 .IP (2)
729 Profile entry
730 .Pe Pager
731 (replaces
732 .Pe moreproc )
733 .IP (3)
734 Environment variable
735 .Ev PAGER
736 .IP (4)
737 Command
738 .Pn more .
739 .P
740 .Ci 0c4214ea2aec6497d0d67b436bbee9bc1d225f1e
741 .P
742 By respecting the
743 .Ev VISUAL /\c
744 .Ev EDITOR
745 and
746 .Ev PAGER
747 environment variables,
748 the new behavior confirms better to the common style on Unix systems.
749 Additionally, the new approach is more uniform and clearer to users.
751 .U3 "Locale
752 .P
753 The configure option
754 .Sw --disable-locale
755 was removed because POSIX provides locale support and there's
756 hardly any need to disable locale support.
757 .Ci ccf4f175ef4c4e7522f9510a4a1149c15d810dd9
759 .U3 "ndbm
760 .P
761 .Pn slocal
762 used to depend on
763 .I ndbm ,
764 a database library.
765 The database is used to store the `\fLMessage-ID\fP's of all
766 messages delivered.
767 This enables
768 .Pn slocal
769 to suppress delivering the same message to the same user twice.
770 (This features was enabled by the
771 .Sw -suppressdup
772 switch.)
773 .P
774 A variety of version of the database library exist.
775 .[
776 wolter unix incompat notes dbm
777 .]
778 Complicated autoconf code was needed to detect them correctly.
779 Further more, the configure switches
780 .Sw --with-ndbm=ARG
781 and
782 .Sw --with-ndbmheader=ARG
783 were added to help with difficult setups that would
784 not be detected automatically or correctly.
785 .P
786 By removing the suppress duplicates feature of
787 .Pn slocal ,
788 the dependency on
789 .I ndbm
790 vanished and 120 lines of complex autoconf code could be saved.
791 .Ci ecd6d6a20cb7a1507e3a20d6c4cb3a1cf14c6bbf
792 The change removed functionality too, but that is minor to the
793 improvement by dropping the dependency and the complex autoconf code.
795 .U3 "mh-e Support
796 .P
797 The configure option
798 .Sw --disable-mhe
799 was removed when the mh-e support was reworked.
800 Mh-e is the Emacs front-end to MH.
801 It requires MH to provide minor additional functions.
802 The
803 .Sw --disable-mhe
804 configure option could switch these extensions off.
805 After removing the support for old versions of mh-e,
806 only the
807 .Sw -build
808 switches of
809 .Pn forw
810 and
811 .Pn repl
812 are left to be mh-e extensions.
813 They are now always built in because they add little code and complexity.
814 In consequence, the
815 .Sw --disable-mhe
816 configure option was removed
817 .Ci a7ce7b4a580d77b6c2c4d980812beb589aa4c643
818 Removing the option removed a second code setup that would have
819 needed to be tested.
820 This change was first done in nmh and thereafter merged into mmh.
821 .P
822 The interface changes in mmh require mh-e to be adjusted in order
823 to be able to use mmh as back-end.
824 This will require minor changes to mh-e, but removing the
825 .Sw -build
826 switches would require more rework.
828 .U3 "Masquerading
829 .P
830 The configure option
831 .Sw --enable-masquerade
832 could take up to three arguments:
833 `draft_from', `mmailid', and `username_extension'.
834 They activated different types of address masquerading.
835 All of them were implemented in the SMTP-speaking
836 .Pn post
837 command, which provided an MSA.
838 Address masquerading is an MTA's task and mmh does not cover
839 this field anymore.
840 Hence, true masquerading needs to be implemented in the external MTA.
841 .P
842 The
843 .I mmailid
844 masquerading type is the oldest one of the three and the only one
845 available in the original MH.
846 It provided a
847 .I username
848 to
849 .I fakeusername
850 mapping, based on the password file's GECOS field.
851 The man page
852 .Mp mh-tailor(5)
853 described the use case as being the following:
854 .QP
855 This is useful if you want the messages you send to always
856 appear to come from the name of an MTA alias rather than your
857 actual account name. For instance, many organizations set up
858 `First.Last' sendmail aliases for all users. If this is
859 the case, the GECOS field for each user should look like:
860 ``First [Middle] Last <First.Last>''
861 .P
862 As mmh sends outgoing mail via the local MTA only,
863 the best location to do such global rewrites is there.
864 Besides, the MTA is conceptionally the right location because it
865 does the reverse mapping for incoming mail (aliasing), too.
866 Further more, masquerading set up there is readily available for all
867 mail software on the system.
868 Hence, mmailid masquerading was removed.
869 .Ci 0836c8000ccb34b59410ef1c15b1b7feac70ce5f
870 .P
871 The
872 .I username_extension
873 masquerading type did not replace the username but would append a suffix,
874 specified by the
875 .Ev USERNAME_EXTENSION
876 environment variable, to it.
877 This provided support for the
878 .I user-extension
879 feature of qmail and the similar
880 .I "plussed user
881 processing of sendmail.
882 The decision to remove this username_extension masquerading was
883 motivated by the fact that
884 .Pn spost
885 hadn't supported it already.
886 .Ci 2abae0bfd0ad5bf898461e50aa4b466d641f23d9
887 Username extensions are possible in mmh, but less convenient to use.
888 .\" XXX format file %(getenv USERNAME_EXTENSION)
889 .P
890 The
891 .I draft_from
892 masquerading type instructed
893 .Pn post
894 to use the value of the
895 .Hd From
896 header field as SMTP envelope sender.
897 Sender addresses could be replaced completely.
898 .Ci b14ea6073f77b4359aaf3fddd0e105989db9
899 Mmh offers a kind of masquerading similar in effect, but
900 with technical differences.
901 As mmh does not transfer messages itself, the local MTA has final control
902 over the sender's address. Any masquerading mmh introduces may be reverted
903 by the MTA.
904 In times of pedantic spam checking, an MTA will take care to use
905 sensible envelope sender addresses to keep its own reputation up.
906 Nonetheless, the MUA can set the
907 .Hd From
908 header field and thereby propose
909 a sender address to the MTA.
910 The MTA may then decide to take that one or generate the canonical sender
911 address for use as envelope sender address.
912 .P
913 In mmh, the MTA will always extract the recipient and sender from the
914 message header (\c
915 .Pn sendmail 's
916 .Sw -t
917 switch).
918 The
919 .Hd From
920 header field of the draft may be set arbitrary by the user.
921 If it is missing, the canonical sender address will be generated by the MTA.
923 .U3 "Remaining Options
924 .P
925 Two configure options remain in mmh.
926 One is the locking method to use:
927 .Sw --with-locking=[dot|fcntl|flock|lockf] .
928 The idea of removing all methods except the portable dot locking
929 and having that one as the default is appealing, but this change
930 requires deeper technical investigation into the topic.
931 The other option,
932 .Sw --enable-debug ,
933 compiles the programs with debugging symbols and does not strip them.
934 This option is likely to stay.
939 .H2 "Removal of Switches
940 .P
941 The command line switches of MH tools follow the X Window style.
942 They are words, introduced by a single dash.
943 For example:
944 .Cl "-truncate" .
945 Every program in mmh has two generic switches:
946 .Sw -help ,
947 to print a short message on how to use the program, and
948 .Sw -Version ,
949 to tell what version of mmh the program belongs to.
950 .P
951 Switches change the behavior of programs.
952 Programs that do one thing in one way require no switches.
953 In most cases, doing something in exactly one way is too limiting.
954 If it is basically the same task to accomplish, but it should be done
955 in various ways, switches are a good approach to alter the behavior
956 of a program.
957 Changing the behavior of programs provides flexibility and customization
958 to users, but in the same way it complicates the code, documentation and
959 usage of the program.
960 Therefore, the number of switches should be kept small.
961 A small set of well-chosen switches does no harm.
962 But usually, the number of switches increases over time.
963 Already in 1985, Rose and Romine have identified this as a major
964 problem of MH:
965 .[ [
966 rose romine real work
967 .], p. 12]
968 .sp
969 .QP
970 A complaint often heard about systems which undergo substantial development
971 by many people over a number of years, is that more and more options are
972 introduced which add little to the functionality but greatly increase the
973 amount of information a user needs to know in order to get useful work done.
974 This is usually referred to as creeping featurism.
975 .QP
976 Unfortunately MH, having undergone six years of off-and-on development by
977 ten or so well-meaning programmers (the present authors included),
978 suffers mightily from this.
979 .sp
980 .P
981 Adding new switches only reluctantly is one part of the counter-action,
982 the other is removing hardly used switches.
983 Now that there are lots of switches already implemented,
984 removing some of them is more important.
985 Removing existing functionality is always difficult because it
986 breaks programs that use these functions.
987 Also, for every obsolete feature, there'll always be someone who still
988 uses it and thus opposes its removal.
989 This puts the developer into the position,
990 where sensible improvements to style are regarded as destructive acts.
991 Yet, living with the featurism is far worse, in my eyes.
992 Future needs will demand adding new features,
993 worsening the situation more and more.
994 Rose and Romine added in a footnote,
995 ``[...]
996 .Pn send
997 will no doubt acquire an endless number of switches in the years to come.''
998 Although clearly humorous, the comment displays the nature of
999 the problem.
1000 Though refusing to add any new switches would encounter the problem
1001 at its root, it is not practical.
1002 But removing obsolete switches is an effective approach to deal with the
1003 problem.
1004 Working on an experimental branch,
1005 eased this work because I had not to offend users.
1006 .P
1007 Rose and Romine counted 24 visible and 9 more hidden switches for
1008 .Pn send .
1009 At the beginning of mmh, it were 32 visible and 12 hidden ones.
1010 At the time of writing, mmh's
1011 .Pn send
1012 has 7 visible switches and 1 hidden switch.
1013 (In each of the examples, the two generic help and version switches
1014 are included.)
1015 .P
1016 Figure XXX
1017 .\" XXX Ref
1018 displays the number of switches for each of the tools that was not
1019 removed from or newly added to mmh.
1020 Both, visible and hidden switches, were counted, but
1021 not the generic help and version switches.
1022 Whereas in the beginning of the project, the average tool had 11 switches,
1023 now it has no more than 5 \(en only half as many.
1024 If the `no' switches and similar inverse variant are folded onto
1025 their counter-parts, the numbers are 8 in pre-mmh to 4 now.
1026 The total number of functional switches in mmh dropped from 465
1027 to 234.
1029 .KS
1030 .in 1c
1031 .so input/switches.grap
1032 .KE
1034 .P
1035 A part of the switches vanished after functions were removed.
1036 This was the case for network mail transfer, for instance.
1037 Sometimes the work flow was the other way:
1038 The trying to reduce the number of switches suggested the removal of
1039 functions.
1041 .U3 "Draft Folder Facility
1042 .P
1043 A change early in the project was the completely transition from
1044 the single draft message to the draft folder facility.
1045 The draft folder facility was introduced in the mid-Eighties.
1046 (Rose and Romine called it a ``relatively new feature''
1047 .[
1048 rose romine real work
1049 .]
1050 in 1985.)
1051 Since then, the facility had existed but had remained deactive by default.
1052 The default activation and the related rework of the tools made it
1053 possible to remove the
1054 .Sw -[no]draftfolder ,
1055 and
1056 .Sw -draftmessage
1057 switches from
1058 .Pn comp ,
1059 .Pn repl ,
1060 .Pn forw ,
1061 .Pn dist ,
1062 .Pn whatnow ,
1063 and
1064 .Pn send .
1065 The only flexibility removed is having multiple draft folders
1066 within one profile.
1067 I consider this only a theoretical setup.
1068 In the same go, the
1069 .Sw -draft
1070 switch of
1071 .Pn anno ,
1072 .Pn refile ,
1073 and
1074 .Pn send
1075 was removed.
1076 The special-casing of `the' draft message became irrelevant after
1077 the rework of the draft system.
1078 (See Sec. XXX.)
1080 .U3 "Inplace
1081 .P
1082 .Pn anno
1083 had the switches
1084 .Sw -[no]inplace
1085 to either annotate the message inplace and thus preserve hard links,
1086 or annotate a copy to replace the original message, breaking hard links.
1087 Following the assumption that linked messages are the same message,
1088 and annotating it should not break the link, the
1089 .Sw -[no]inplace
1090 switches were removed and the previous default
1091 .Sw -inplace
1092 was made the only behavior.
1093 The
1094 .Sw -[no]inplace
1095 switches of
1096 .Pn repl ,
1097 .Pn forw ,
1098 and
1099 .Pn dist
1100 could be removed, too, as they were simply passed through to
1101 .Pn anno .
1102 .P
1103 .Pn burst
1104 also had
1105 .Sw -[no]inplace
1106 switches, but they had different meaning, as written in nmh's
1107 .Mp burst(1)
1108 man page:
1109 .sp
1110 .QP
1111 If -noinplace is given, each digest is preserved, no table
1112 of contents is produced, and the messages contained within
1113 the digest are placed at the end of the folder. Other messages
1114 are not tampered with in any way.
1115 .sp
1116 .P
1117 With
1118 .Sw -inplace ,
1119 the digest had been replaced by the table of contents (i.e. the
1120 introduction text) and the bursted messages were placed right
1121 after this message, renumbering all following messages.
1122 Also, any trailing text of the digest will be lost, though,
1123 in practice, it usually consists of an end-of-digest marker only.
1124 The decision to drop the
1125 .Sw -inplace
1126 behavior was supported by the complexity and possible data loss
1127 it introduced.
1128 .Sw -noinplace
1129 was the default behavior already and is the chosen behavior now.
1131 .U3 "mbox and MMDF
1132 .P
1133 packf: file mbox mmdf
1134 rcvpack: mbox mmdf
1137 .ig
1139 ap: width
1140 dp: width
1141 burst: [no]quiet
1142 flist: [no]total
1143 folder: [no]header
1145 comp: file
1146 dist: file(msh)
1147 forw: filter, [no]mime, [no]dashstuffing(mhl)
1148 repl: [no]format/filter width
1150 mhmail: resent queued
1151 inc: snoop, (pop)
1152 mhbuild: [no]check, [no]ebcdicsafe, [no]headers, [no]list, [no]realsize
1153 [no]rfc934mode, [no]contentid (caching)
1154 mhlist: [no]check [no]headers [no]realsize (caching)
1155 mhstore: [no]check [no]verbose (caching)
1157 scan: [no]clear [no]header [no]reverse
1159 mhl: [no]bell [no]clear [no]faceproc folder [no]moreproc length sleep
1160 [no]dashstuffing(forw) digest list volume number issue number
1161 mhshow: [no]check [no]pause [no]serialonly (caching) [no]moreproc
1162 length width
1164 prompter: erase kill [no]doteof
1166 refile: [no]preserve [no]unlink [no]rmmproc
1168 send: filter [no]format [no]forward [no]mime [no]msgid
1169 [no]push split [no]unique (sasl) width snoop [no]dashstuffing
1170 attach attachformat
1171 whatnow: (noedit) attach
1173 slocal: [no]suppressdups
1175 sortm: subject
1177 spost: [no]filter [no]format [no]remove [no]backup width [no]push idanno
1178 [no]check(whom) whom(whom)
1180 whom: ???
1183 (pop) host, port, user, [no]pack, proxy
1184 (smtp) mail saml send soml client server user port
1185 (sasl) sasl, saslmech
1186 (tls)
1187 (caching) rcache wcache
1189 noedit
1190 nowhatnowproc
1193 format -> form
1195 version -> Version
1197 ..
1199 .ig
1201 .P
1202 To ease typing, the switches can be abbreviated as much as the remaining
1203 prefix remains unambiguous.
1204 If in our example no other switch would start with the letter `t', then
1205 .Cl "-truncate" ,
1206 .Cl "-trunc" ,
1207 .Cl "-tr" ,
1208 and
1209 .Cl "-t
1210 would all be the same.
1211 As a result, switches can neither be grouped (as in
1212 .Cl "ls -ltr" )
1213 nor can switch arguments be appended directly to the switch (as in
1214 .Cl "sendmail -q30m" ).
1215 .P
1216 Many switches have negating counter-parts, which start with `no'.
1217 For example
1218 .Cl "-notruncate
1219 inverts the
1220 .Cl "-truncate
1221 switch.
1222 They exist to undo the effect of default switches in the profile.
1223 If the user has chosen to change the default behavior of some tool
1224 by adding a default switch to the profile,
1225 he can still undo this change in behavior by specifiying the inverse
1226 switch on the command line.
1227 .P
1228 In the best case, all switches are unambiguous on the first character,
1229 or on the three-letter prefix for the `no' variants.
1230 Reducing switch prefix collisions, shortens the necessary prefix lenght
1231 the user must type.
1232 Having less switches helps best.
1234 ..
1236 .H1 "Modernizing
1239 .H2 "Removal of Code Relicts
1240 .P
1241 The code base of mmh originates from the late Seventies,
1242 had been extensively
1243 worked on in the mid Eighties, and had been partly reorganized and extended
1244 in the Nineties. Relicts of all those times had gathered in the code base.
1245 My goal was to remove any ancient code parts. One part of the task was
1246 converting obsolete code constructs to standard constructs, the other part
1247 was dropping obsolete functions.
1248 .P
1249 As I'm not even thirty years old and have no more than seven years of
1250 Unix experience, I needed to learn about the history in retrospective.
1251 Older people likely have used those ancient constructs themselves
1252 and have suffered from their incompatibilities and have longed for
1253 standardization. Unfortunately, I have only read that others had done so.
1254 This put me in a much more difficult positions when working on the old
1255 code. I needed to recherche what other would have known by heart from
1256 experience. All my programming experience comes from a time past ANSI C
1257 and past POSIX. Although I knew about the times before, I took the
1258 current state implicitly for granted most of the time.
1259 .P
1260 Being aware of
1261 these facts, I rather let people with more historic experience solve the
1262 task of converting the ancient code constructs to standardized ones.
1263 Luckily, Lyndon Nerenberg focused on this task at the nmh project.
1264 He converted large parts of the code to POSIX constructs, removing
1265 the conditionals compilation for now standardized features.
1266 I'm thankful for this task being solved. I only pulled the changes into
1267 mmh.
1268 .P
1269 The other task \(en dropping ancient functionality to remove old code \(en
1270 I did myself, though. My position to strip mmh to the bare minimum of
1271 frequently used features is much more revolutional than the nmh community
1272 likes it. Without the need to justify my decisions, I was able to quickly
1273 remove functionality I considered ancient.
1274 The need to discuss my decisions with
1275 peers likely would have slowed this process down. Of course, I researched
1276 if a particular feature really should be dropped. Having not had any
1277 contact to this feature within my computer life was a first indicator to
1278 drop it, but I also asked others and searched the literature for modern
1279 usage of the feature. If it appeared to be truly ancient, I dropped it.
1280 The reason for dropping is always part of the commit message in the
1281 version control system. Thus, it is easy for others to check their
1282 view on the topic with mine and possibly to argue for reinclusion.
1284 .U2 "MMDF maildrop support
1285 .P
1286 I did drop any support for the MMDF maildrop format. This type of format
1287 is conceptionally similar to the mbox format, but uses four bytes with
1288 value 1 (\fL^A^A^A^A\fP) as message delimiter,
1289 instead of the string ``\fLFrom\ \fP''.
1290 Due to the similarity and mbox being the de-facto standard maildrop
1291 format on Unix, but also due to the larger influence of Sendmail than MMDF,
1292 the MMDF maildrop format had vanished.
1293 .P
1294 The simplifications within the code were only moderate. Switches could
1295 be removed from tools like
1296 .L packf ,
1297 which generate packed mailboxes. Only one packed mailbox format remained:
1298 mbox.
1299 The most important changes affect the equally named mail parsing routine in
1300 .L sbr/m_getfld.c .
1301 The direct MMDF code had been removed, but as now only one packed mailbox
1302 format is left, code structure simplifications are likely possible.
1303 The reason why they are still outstanding is the heavily optimized code
1304 of
1305 .Fu m_getfld() .
1306 Changes beyond a small local scope \(en
1307 which restructuring in its core is \(en cause a high risk of damaging
1308 the intricate workings of the optimized code. This problem is know
1309 to the developers of nmh, too. They also avoid touching this minefield
1310 if possible.
1312 .U2 "UUCP Bang Paths
1313 .P
1314 More questionably than the former topic is the removal of support for the
1315 UUCP bang path address style. However, the user may translate the bang
1316 paths on retrieval to Internet addresses and the other way on posting
1317 messages. The former can be done my an MDA like procmail; the latter
1318 by a sendmail wrapper. This would ensure that any address handling would
1319 work as expected. However, it might just work well without any
1320 such modifications, as mmh does not touch addresses much, in general.
1321 But I can't ensure as I have never used an environment with bang paths.
1322 Also, the behavior might break at any point in further development.
1324 .U2 "Hardcopy terminal support
1325 .P
1326 More of a funny anecdote is the remaining of a check for printing to a
1327 hardcopy terminal until Spring 2012, when I finally removed it.
1328 I surely would be very happy to see such a terminal in action, maybe
1329 actually being able to work on it, but I fear my chances are null.
1330 .P
1331 The check only prevented a pager to be placed between the outputting
1332 program (\c
1333 .Pn mhl )
1334 and the terminal. This could have been ensured with
1335 the
1336 .Sw -nomoreproc
1337 at the command line statically, too.
1339 .U2 "Removed support for header fields
1340 .P
1341 The
1342 .Hd Encrypted
1343 header field had been introduced by RFC\^822, but already
1344 marked legacy in RFC 2822. It was superseded by FIXME.
1345 Mmh does no more support this header field.
1346 .P
1347 Native support for
1348 .Hd Face
1349 header fields had been removed, as well.
1350 The feature is similar to the
1351 .Hd X-Face
1352 header field in its intent,
1353 but takes a different approach to store the image.
1354 Instead of encoding the image data directly into the header field,
1355 the it contains the hostname and UDP port where the image
1356 date could be retrieved.
1357 Neither
1358 .Hd X-Face
1359 nor the here described
1360 .Hd Face
1361 system
1362 \**
1363 .FS
1364 There is also a newer but different system, invented 2005,
1365 using
1366 .Hd Face
1367 headers.
1368 It is the successor of
1369 .Hd X-Face
1370 providing colored PNG images.
1371 .FE
1372 became well used in the large scale.
1373 It's still possible to use a Face systems,
1374 although mmh does not provide support for any of the different systems
1375 anymore. It's fairly easy to write a small shell script to
1376 extract the embedded or fetch the external Face data and display the image.
1377 Own
1378 .Hd Face
1379 header field can be added into the draft template files.
1380 .P
1381 .Hd Content-MD5
1382 header fields were introduced by RFC\^1864. They provide only
1383 a verification of data corruption during the transfer. By no means can
1384 they ensure verbatim end-to-end delivery of the contents. This is clearly
1385 stated in the RFC. The proper approach to provide verificationability
1386 of content in an end-to-end relationship is the use of digital cryptography
1387 (RFCs FIXME). On the other hand, transfer protocols should ensure the
1388 integrity of the transmission. In combinations these two approaches
1389 make the
1390 .Hd Content-MD5
1391 header field useless. In consequence, I removed
1392 the support for it. By this removal, MD5 computation is not needed
1393 anywhere in mmh. Hence, over 500 lines of code were removed by this one
1394 change. Even if the
1395 .Hd Content-MD5
1396 header field is useful sometimes,
1397 I value its usefulness less than the improvement in maintainability, caused
1398 by the removal.
1400 .U2 "Prompter's Control Keys
1401 .P
1402 The program
1403 .Pn prompter
1404 queries the user to fill in a message form. When used by
1405 .Pn comp
1406 as:
1407 .VS
1408 comp -editor prompter
1409 VE
1410 the resulting behavior is similar to
1411 .Pn mailx .
1412 Apparently,
1413 .Pn prompter
1414 hadn't been touched lately. Otherwise it's hardly explainable why it
1415 still offered the switches
1416 .Sw -erase
1417 .Ar chr
1418 and
1419 .Sw -kill
1420 .Ar chr
1421 to name the characters for command line editing.
1422 The times when this had been necessary are long time gone.
1423 Today these things work out-of-the-box, and if not, are configured
1424 with the standard tool
1425 .Pn stty .
1427 .U2 "Vfork and Retry Loops
1428 .P
1429 MH creates many processes, which is a consequence of the tool chest approach.
1430 In earlier times
1431 .Fu fork()
1432 had been an expensive system call, as the process's whole image needed
1433 to be duplicated. One common case is replacing the image with
1434 .Fu exec()
1435 right after having forked the child process.
1436 To speed up this case, the
1437 .Fu vfork()
1438 system call was invented at Berkeley. It completely omits copying the
1439 image. If the image gets replaced right afterwards then unnecessary
1440 work is omited. On old systems this results in large speed ups.
1441 MH uses
1442 .Fu vfork()
1443 whenever possible.
1444 .P
1445 Memory management units that support copy-on-write semantics make
1446 .Fu fork()
1447 almost as fast as
1448 .Fu vfork()
1449 in the cases when they can be exchanged.
1450 With
1451 .Fu vfork()
1452 being more error-prone and hardly faster, it's preferable to simply
1453 use
1454 .Fu fork()
1455 instead.
1456 .P
1457 Related to the costs of
1458 .Fu fork()
1459 is the probability of its success.
1460 Today on modern systems, the system call will succeed almost always.
1461 In the Eighties on heavy loaded systems, as they were common at
1462 universities, this had been different. Thus, many of the
1463 .Fu fork()
1464 calls were wrapped into loops to retry to fork several times in
1465 short intervals, in case of previous failure.
1466 In mmh, the program aborts at once if the fork failed.
1467 The user can reexecute the command then. This is expected to be a
1468 very rare case on modern systems, especially personal ones, which are
1469 common today.
1472 .H2 "Attachments
1473 .P
1474 MIME
1477 .H2 "Digital Cryptography
1478 .P
1479 Signing and encryption.
1482 .H2 "Good Defaults
1483 .P
1484 foo
1489 .H1 "Code style
1490 .P
1491 foo
1494 .H2 "Standard Code
1495 .P
1496 POSIX
1499 .H2 "Separation
1501 .U2 "MH Directory Split
1502 .P
1503 In MH and nmh, a personal setup had consisted of two parts:
1504 The MH profile, named
1505 .Fn \&.mh_profile
1506 and being located directly in the user's home directory.
1507 And the MH directory, where all his mail messages and also his personal
1508 forms, scan formats, other configuration files are stored. The location
1509 of this directory could be user-chosen. The default was to name it
1510 .Fn Mail
1511 and have it directly in the home directory.
1512 .P
1513 I've never liked the data storage and the configuration to be intermixed.
1514 They are different kinds of data. One part, are the messages,
1515 which are the data to operate on. The other part, are the personal
1516 configuration files, which are able to change the behavior of the operations.
1517 The actual operations are defined in the profile, however.
1518 .P
1519 When storing data, one should try to group data by its type.
1520 There's sense in the Unix file system hierarchy, where configuration
1521 file are stored separate (\c
1522 .Fn /etc )
1523 to the programs (\c
1524 .Fn /bin
1525 and
1526 .Fn /usr/bin )
1527 to their sources (\c
1528 .Fn /usr/src ).
1529 Such separation eases the backup management, for instance.
1530 .P
1531 In mmh, I've reorganized the file locations.
1532 Still there are two places:
1533 There's the mail storage directory, which, like in MH, contains all the
1534 messages, but, unlike in MH, nothing else.
1535 Its location still is user-chosen, with the default name
1536 .Fn Mail ,
1537 in the user's home directory. This is much similar to the case in nmh.
1538 The configuration files, however, are grouped together in the new directory
1539 .Fn \&.mmh
1540 in the user's home directory.
1541 The user's profile now is a file, named
1542 .Fn profile ,
1543 in this mmh directory.
1544 Consistently, the context file and all the personal forms, scan formats,
1545 and the like, are also there.
1546 .P
1547 The naming changed with the relocation.
1548 The directory where everything, except the profile, had been stored (\c
1549 .Fn $HOME/Mail ),
1550 used to be called \fIMH directory\fP. Now, this directory is called the
1551 user's \fImail storage\fP. The name \fImmh directory\fP is now given to
1552 the new directory
1553 (\c
1554 .Fn $HOME/.mmh ),
1555 containing all the personal configuration files.
1556 .P
1557 The separation of the files by type of content is logical and convenient.
1558 There are no functional differences as any possible setup known to me
1559 can be implemented with both approaches, although likely a bit easier
1560 with the new approach. The main goal of the change had been to provide
1561 sensible storage locations for any type of personal mmh file.
1562 .P
1563 In order for one user to have multiple MH setups, he can use the
1564 environment variable
1565 .Ev MH
1566 the point to a different profile file.
1567 The MH directory (mail storage plus personal configuration files) is
1568 defined by the
1569 .Pe Path
1570 profile entry.
1571 The context file could be defined by the
1572 .Pe context
1573 profile entry or by the
1574 .Ev MHCONTEXT
1575 environment variable.
1576 The latter is useful to have a distinct context (e.g. current folders)
1577 in each terminal window, for instance.
1578 In mmh, there are three environment variables now.
1579 .Ev MMH
1580 may be used to change the location of the mmh directory.
1581 .Ev MMHP
1582 and
1583 .Ev MMHC
1584 change the profile and context files, respectively.
1585 Besides providing a more consistent feel (which simply is the result
1586 of being designed anew), the set of personal configuration files can
1587 be chosen independently from the profile (including mail storage location)
1588 and context, now. Being it relevant for practical use or not, it
1589 de-facto is an improvement. However, the main achievement is the
1590 split between mail storage and personal configuration files.
1593 .H2 "Modularization
1594 .P
1595 whatnowproc
1596 .P
1597 The \fIMH library\fP
1598 .Fn libmh.a
1599 collects a bunch of standard functions that many of the MH tools need,
1600 like reading the profile or context files.
1601 This doesn't hurt the separation.
1604 .H2 "Style
1605 .P
1606 Code layout, goto, ...
1611 .H1 "Concept Exploitation/Homogeneity
1614 .H2 "Draft Folder
1615 .P
1616 Historically, MH provided exactly one draft message, named
1617 .Fn draft
1618 and
1619 being located in the MH directory. When starting to compose another message
1620 before the former one was sent, the user had been questioned whether to use,
1621 refile or replace the old draft. Working on multiple drafts at the same time
1622 was impossible. One could only work on them in alteration by refiling the
1623 previous one to some directory and fetching some other one for reediting.
1624 This manual draft management needed to be done each time the user wanted
1625 to switch between editing one draft to editing another.
1626 .P
1627 To allow true parallel editing of drafts, in a straight forward way, the
1628 draft folder facility exists. It had been introduced already in July 1984
1629 by Marshall T. Rose. The facility was deactivated by default.
1630 Even in nmh, the draft folder facility remained deactivated by default.
1631 At least, Richard Coleman added the man page
1632 .Mp mh-draft(5)
1633 to document
1634 the feature well.
1635 .P
1636 The only advantage of not using the draft folder facility is the static
1637 name of the draft file. This could be an issue for MH front-ends like mh-e.
1638 But as they likely want to provide working on multiple drafts in parallel,
1639 the issue is only concerning compatibility. The aim of nmh to stay compatible
1640 prevented the default activation of the draft folder facility.
1641 .P
1642 On the other hand, a draft folder is the much more natural concept than
1643 a draft message. MH's mail storage consists of folders and messages,
1644 the messages named with ascending numbers. A draft message breaks with this
1645 concept by introducing a message in a file named
1646 .Fn draft .
1647 This draft
1648 message is special. It can not be simply listed with the available tools,
1649 but instead requires special switches. I.e. corner-cases were
1650 introduced. A draft folder, in contrast, does not introduce such
1651 corner-cases. The available tools can operate on the messages within that
1652 folder like on any messages within any mail folders. The only difference
1653 is the fact that the default folder for
1654 .Pn send
1655 is the draft folder,
1656 instead of the current folder, like for all other tools.
1657 .P
1658 The trivial part of the change was activating the draft folder facility
1659 by default and setting a default name for this folder. Obviously, I chose
1660 the name
1661 .Fn +drafts .
1662 This made the
1663 .Sw -draftfolder
1664 and
1665 .Sw -draftmessage
1666 switches useless, and I could remove them.
1667 The more difficult but also the part that showed the real improvement,
1668 was updating the tools to the new concept.
1669 .Sw -draft
1670 switches could
1671 be dropped, as operating on a draft message became indistinguishable to
1672 operating on any other message for the tools.
1673 .Pn comp
1674 still has its
1675 .Sw -use
1676 switch for switching between its two modes: (1) Compose a new
1677 draft, possibly by taking some existing message as a form. (2) Modify
1678 an existing draft. In either case, the behavior of
1679 .Pn comp is
1680 deterministic. There is no more need to query the user. I consider this
1681 a major improvement. By making
1682 .Pn send
1683 simply operate on the current
1684 message in the draft folder by default, with message and folder both
1685 overridable by specifying them on the command line, it is now possible
1686 to send a draft anywhere within the storage by simply specifying its folder
1687 and name.
1688 .P
1689 All theses changes converted special cases to regular cases, thus
1690 simplifying the tools and increasing the flexibility.
1693 .H2 "Trash Folder
1694 .P
1695 Similar to the situation for drafts is the situation for removed messages.
1696 Historically, a message was deleted by renaming. A specific
1697 \fIbackup prefix\fP, often comma (\c
1698 .Fn , )
1699 or hash (\c
1700 .Fn # ),
1701 being prepended to the file name. Thus, MH wouldn't recognize the file
1702 as a message anymore, as only files whose name consists of digits only
1703 are treated as messages. The removed messages remained as files in the
1704 same directory and needed some maintenance job to truly delete them after
1705 some grace time. Usually, by running a command similar to
1706 .VS
1707 find /home/user/Mail -ctime +7 -name ',*' | xargs rm
1708 VE
1709 in a cron job. Within the grace time interval
1710 the original message could be restored by stripping the
1711 the backup prefix from the file name. If however, the last message of
1712 a folder is been removed \(en say message
1713 .Fn 6
1714 becomes file
1715 .Fn ,6
1716 \(en and a new message enters the same folder, thus the same
1717 numbered being given again \(en in our case
1718 .Fn 6
1719 \(en, if that one
1720 is removed too, then the backup of the former message gets overwritten.
1721 Thus, the ability to restore removed messages does not only depend on
1722 the ``sweeping cron job'' but also on the removing of further messages.
1723 This is undesirable, because the real mechanism is hidden from the user
1724 and the consequences of further removals are not always obvious.
1725 Further more, the backup files are scattered within the whole mail
1726 storage, instead of being collected at one place.
1727 .P
1728 To improve the situation, the profile entry
1729 .Pe rmmproc
1730 (previously named
1731 .Pe Delete-Prog )
1732 was introduced, very early.
1733 It could be set to any command, which would care for the mail removal
1734 instead of taking the default action, described above.
1735 Refiling the to-be-removed files to some garbage folder was a common
1736 example. Nmh's man page
1737 .Mp rmm(1)
1738 proposes
1739 .Cl "refile +d
1740 to move messages to the garbage folder and
1741 .Cl "rm `mhpath +d all`
1742 the empty the garbage folder.
1743 Managing the message removal this way is a sane approach. It keeps
1744 the removed messages in one place, makes it easy to remove the backup
1745 files, and, most important, enables the user to use the tools of MH
1746 itself to operate on the removed messages. One can
1747 .Pn scan
1748 them,
1749 .Pn show
1750 them, and restore them with
1751 .Pn refile .
1752 There's no more
1753 need to use
1754 .Pn mhpath
1755 to switch over from MH tools to Unix tools \(en MH can do it all itself.
1756 .P
1757 This approach matches perfect with the concepts of MH, thus making
1758 it powerful. Hence, I made it the default. And even more, I also
1759 removed the old backup prefix approach, as it is clearly less powerful.
1760 Keeping unused alternative in the code is a bad choice as they likely
1761 gather bugs, by not being constantly tested. Also, the increased code
1762 size and more conditions crease the maintenance costs. By strictly
1763 converting to the trash folder approach, I simplified the code base.
1764 .Pn rmm
1765 calls
1766 .Pn refile
1767 internally to move the to-be-removed
1768 message to the trash folder (\c
1769 .Fn +trash
1770 by default). Messages
1771 there can be operated on like on any other message in the storage.
1772 The sweep clean, one can use
1773 .Cl "rmm -unlink +trash a" ,
1774 where the
1775 .Sw -unlink
1776 switch causes the files to be truly unliked instead
1777 of moved to the trash folder.
1780 .H2 "Path Notations
1781 .P
1782 foo
1785 .H2 "MIME Integration
1786 .P
1787 user-visible access to whole messages and MIME parts are inherently
1788 different
1791 .H2 "Of One Cast
1792 .P