docs/master
diff ch03.roff @ 58:814c33b96d89
Restructured the content in ch03.
author | markus schnalke <meillo@marmaro.de> |
---|---|
date | Fri, 01 Jun 2012 12:33:49 +0200 |
parents | 49cf68506b5d |
children | 6a92e0208de0 |
line diff
1.1 --- a/ch03.roff Sun May 20 22:10:22 2012 +0200 1.2 +++ b/ch03.roff Fri Jun 01 12:33:49 2012 +0200 1.3 @@ -1,10 +1,159 @@ 1.4 -.H0 "Work Report 1.5 +.H0 "Discussion 1.6 .P 1.7 -foo 1.8 +This main chapter discusses the practical work done in the mmh project. 1.9 +It is structured along the goals to achieve. The concrete work done 1.10 +is described in the examples of how the general goals were achieved. 1.11 + 1.12 + 1.13 + 1.14 + 1.15 +.H1 "Stream-lining 1.16 + 1.17 + 1.18 +.H2 "Removal of non-MUA Tools 1.19 .P 1.20 -bar 1.21 +MH had been considered an all-in-one system for mail handling. 1.22 +The community around nmh has a similar understanding. 1.23 +In fundamental difference, I believe that mmh should be a MUA but 1.24 +nothing more. I believe that all-in-one mail systems are not the way 1.25 +to go. There are excellent specialized MTAs, like Postfix; 1.26 +there are specialized MDAs, like Procmail; there are specialized 1.27 +MRAs, like Fetchmail. I believe it's best to use them instead of 1.28 +providing the same function ourselves. Doing something well requires to 1.29 +focus on this particular aspect or a small set of aspects. The more 1.30 +it is possible to focus, the better the result in this particular 1.31 +area will be. The limiting resource in Free Software community development 1.32 +usually is human power. If the low development power is even parted 1.33 +into multiple development areas, it will hardly be possible to 1.34 +compete with the specialists in the various fields. This is even 1.35 +increased, given the small community \(en developers and users \(en 1.36 +that MH-based mail systems have. In consequence, I believe that the 1.37 +available resources should be concentrated at the point where MH is 1.38 +most unique. This is clearly the MUA part. 1.39 +.P 1.40 +Several of nmh's tools were removed from mmh because they didn't 1.41 +match the main focus of adding to the MUA's task. 1.42 +.P 1.43 +.Pn conflict 1.44 +was removed because it is a mail system maintenance tool. 1.45 +Besides, it also checks the 1.46 +.Fn /etc/passwd 1.47 +and 1.48 +.Fn /etc/group 1.49 +files. 1.50 +The tool might be useful, but it should not be shipped with mmh. 1.51 +.P 1.52 +.Pn rcvtty 1.53 +was removed because its usecase of writing to the user's terminal 1.54 +on receiving of mail is hardly wanted today. If users like to be 1.55 +informed of new mail, then using the shell's 1.56 +.Ev MAILPATH 1.57 +variable or different (graphical) notifications are likely more 1.58 +appealing. Writing directly to other terminals is hardly ever wanted 1.59 +today. If though one wants to have it this way, the standard tool 1.60 +.Pn write 1.61 +can be used in a way similar to: 1.62 +.DS 1.63 +scan -file - | write `id -un` 1.64 +.DE 1.65 +.P 1.66 +When the new attachment system was introduced, 1.67 +.Pn viamail 1.68 +was removed because then 1.69 +.Pn forw 1.70 +could cover the task itself. 1.71 +The wrapper program 1.72 +.Pn sendfiles 1.73 +was rewritten as a shell script to use 1.74 +.Pn forw . 1.75 +.P 1.76 +.Pn msgchk 1.77 +was removed as it became hardly useful when POP support was removed. 1.78 +It is questionable if 1.79 +.Pn msgchk 1.80 +provides more information than: 1.81 +.DS 1.82 +ls -l /var/mail/meillo 1.83 +.DE 1.84 +It does separate between old and new mail, but that's not very 1.85 +useful and can be found out with 1.86 +.Pn stat (1) 1.87 +too. A very small shell script could care for the form of output. 1.88 +As mmh's inc only incorporates mail from the user's local maildrop 1.89 +and thus no long data transfers are involved, 1.90 +there's no need to check for new mail before incorporating it. 1.91 +.P 1.92 +.Pn msh 1.93 +was removed because the tool was in conflict with the original 1.94 +philosophy of MH. It provided an interactive shell to access the 1.95 +features of MH. One major feature of MH is being a tool chest. 1.96 +.Pn msh 1.97 +wouldn't be just another shell, tailored to the needs of mail 1.98 +handling, but one large program to have the MH tools built in. 1.99 +It's main use was for accessing Bulletin Boards, which have seized to 1.100 +be popular. Removing 1.101 +.Pn msh , 1.102 +together with the truly obsolete programs 1.103 +.Pn vmh 1.104 +and 1.105 +.Pn wmh , 1.106 +saved more than 7\|000 lines of C code \(en a major achievement. 1.107 1.108 -.H1 "Removal of Code Relicts 1.109 +.U2 "Removal of the MTS 1.110 +.P 1.111 + 1.112 + 1.113 +.H2 "mhshow show Merge 1.114 +.P 1.115 +Since the very beginning, already in the first concept paper, 1.116 +.Pn show 1.117 +had been MH's mail display program. 1.118 +.Pn show 1.119 +found out which pathnames the relevant messages had and invoked 1.120 +.Pn mhl 1.121 +then to let it render the content. 1.122 +With the advent of MIME, this approach wasn't sufficient anymore. 1.123 +MIME messages can consist of multiple parts, some of which aren't 1.124 +directly displayable, and text content can be encoded in 1.125 +foreign charsets. 1.126 +.Pn show 's 1.127 +simple approach and 1.128 +.Pn mhl 's 1.129 +limited display facilities couldn't cope with the task any longer. 1.130 +Instead of extending these tools, new ones were written from scratch 1.131 +and then added to the MH tool chest. Doing so is encouraged by the 1.132 +tool chest approach. The new tools could be added without interfering 1.133 +with the existing ones. This is great. It allowed MH to be the 1.134 +first MUA to implement MIME. 1.135 +.P 1.136 +The new MIME features were added in form of the single program 1.137 +.Pn mhn . 1.138 +The command 1.139 +.DS 1.140 +mhn \-show 42 1.141 +.DE 1.142 +would show the MIME message numbered 42. 1.143 +With the 1.0 release of nmh in February 1999, Richard Coleman finished 1.144 +the split of 1.145 +.Pn mhn 1.146 +into a set of specialized programs, which together covered the 1.147 +aspects of MIME. One of these resulting tools was 1.148 +.Pn mhshow . 1.149 + 1.150 + 1.151 +.H2 "Removal of Configure Options 1.152 +.P 1.153 + 1.154 +.H2 "Removal of switches 1.155 +.P 1.156 + 1.157 + 1.158 + 1.159 + 1.160 +.H1 "Moderizing 1.161 + 1.162 + 1.163 +.H2 "Removal of Code Relicts 1.164 .P 1.165 The code base of mmh originates from the late Seventies, 1.166 had been extensively 1.167 @@ -215,263 +364,36 @@ 1.168 common today. 1.169 1.170 1.171 -.H1 "Removal of Tools 1.172 +.H2 "Attachments 1.173 .P 1.174 -MH had been considered an all-in-one system for mail handling. 1.175 -The community around nmh has a similar understanding. 1.176 -In fundamental difference, I believe that mmh should be a MUA but 1.177 -nothing more. I believe that all-in-one mail systems are not the way 1.178 -to go. There are excellent specialized MTAs, like Postfix; 1.179 -there are specialized MDAs, like Procmail; there are specialized 1.180 -MRAs, like Fetchmail. I believe it's best to use them instead of 1.181 -providing the same function ourselves. Doing something well requires to 1.182 -focus on this particular aspect or a small set of aspects. The more 1.183 -it is possible to focus, the better the result in this particular 1.184 -area will be. The limiting resource in Free Software community development 1.185 -usually is human power. If the low development power is even parted 1.186 -into multiple development areas, it will hardly be possible to 1.187 -compete with the specialists in the various fields. This is even 1.188 -increased, given the small community \(en developers and users \(en 1.189 -that MH-based mail systems have. In consequence, I believe that the 1.190 -available resources should be concentrated at the point where MH is 1.191 -most unique. This is clearly the MUA part. 1.192 +MIME 1.193 + 1.194 + 1.195 +.H2 "Digital Cryptography 1.196 .P 1.197 -Several of nmh's tools were removed from mmh because they didn't 1.198 -match the main focus of adding to the MUA's task. 1.199 +Signing and encryption. 1.200 + 1.201 + 1.202 +.H2 "Good Defaults 1.203 .P 1.204 -.Pn conflict 1.205 -was removed because it is a mail system maintenance tool. 1.206 -Besides, it also checks the 1.207 -.Fn /etc/passwd 1.208 -and 1.209 -.Fn /etc/group 1.210 -files. 1.211 -The tool might be useful, but it should not be shipped with mmh. 1.212 +foo 1.213 + 1.214 + 1.215 + 1.216 + 1.217 +.H1 "Code style 1.218 .P 1.219 -.Pn rcvtty 1.220 -was removed because its usecase of writing to the user's terminal 1.221 -on receiving of mail is hardly wanted today. If users like to be 1.222 -informed of new mail, then using the shell's 1.223 -.Ev MAILPATH 1.224 -variable or different (graphical) notifications are likely more 1.225 -appealing. Writing directly to other terminals is hardly ever wanted 1.226 -today. If though one wants to have it this way, the standard tool 1.227 -.Pn write 1.228 -can be used in a way similar to: 1.229 -.DS 1.230 -scan -file - | write `id -un` 1.231 -.DE 1.232 +foo 1.233 + 1.234 + 1.235 +.H2 "Standard Code 1.236 .P 1.237 -When the new attachment system was introduced, 1.238 -.Pn viamail 1.239 -was removed because then 1.240 -.Pn forw 1.241 -could cover the task itself. 1.242 -The wrapper program 1.243 -.Pn sendfiles 1.244 -was rewritten as a shell script to use 1.245 -.Pn forw . 1.246 -.P 1.247 -.Pn msgchk 1.248 -was removed as it became hardly useful when POP support was removed. 1.249 -It is questionable if 1.250 -.Pn msgchk 1.251 -provides more information than: 1.252 -.DS 1.253 -ls -l /var/mail/meillo 1.254 -.DE 1.255 -It does separate between old and new mail, but that's not very 1.256 -useful and can be found out with 1.257 -.Pn stat (1) 1.258 -too. A very small shell script could care for the form of output. 1.259 -As mmh's inc only incorporates mail from the user's local maildrop 1.260 -and thus no long data transfers are involved, 1.261 -there's no need to check for new mail before incorporating it. 1.262 -.P 1.263 -.Pn msh 1.264 -was removed because the tool was in conflict with the original 1.265 -philosophy of MH. It provided an interactive shell to access the 1.266 -features of MH. One major feature of MH is being a tool chest. 1.267 -.Pn msh 1.268 -wouldn't be just another shell, tailored to the needs of mail 1.269 -handling, but one large program to have the MH tools built in. 1.270 -It's main use was for accessing Bulletin Boards, which have seized to 1.271 -be popular. Removing 1.272 -.Pn msh , 1.273 -together with the truly obsolete programs 1.274 -.Pn vmh 1.275 -and 1.276 -.Pn wmh , 1.277 -saved more than 7\|000 lines of C code \(en a major achievement. 1.278 +POSIX 1.279 1.280 1.281 -.H1 "Draft and Trash Folders 1.282 -.U2 "Draft Folder 1.283 -.P 1.284 -Historically, MH provided exactly one draft message, named 1.285 -.Fn draft 1.286 -and 1.287 -being located in the MH directory. When starting to compose another message 1.288 -before the former one was sent, the user had been questioned whether to use, 1.289 -refile or replace the old draft. Working on multiple drafts at the same time 1.290 -was impossible. One could only work on them in alteration by refiling the 1.291 -previous one to some directory and fetching some other one for reediting. 1.292 -This manual draft management needed to be done each time the user wanted 1.293 -to switch between editing one draft to editing another. 1.294 -.P 1.295 -To allow true parallel editing of drafts, in a straight forward way, the 1.296 -draft folder facility exists. It had been introduced already in July 1984 1.297 -by Marshall T. Rose. The facility was deactivated by default. 1.298 -Even in nmh, the draft folder facility remained deactivated by default. 1.299 -At least, Richard Coleman added the man page 1.300 -.Mp mh-draft(5) 1.301 -to document 1.302 -the feature well. 1.303 -.P 1.304 -The only advantage of not using the draft folder facility is the static 1.305 -name of the draft file. This could be an issue for MH frontends like mh-e. 1.306 -But as they likely want to provide working on multiple drafts in parallel, 1.307 -the issue is only concerning compatibility. The aim of nmh to stay compatible 1.308 -prevented the default activation of the draft folder facility. 1.309 -.P 1.310 -On the other hand, a draft folder is the much more natural concept than 1.311 -a draft message. MH's mail storage consists of folders and messages, 1.312 -the messages named with ascending numbers. A draft message breaks with this 1.313 -concept by introducing a message in a file named 1.314 -.Fn draft . 1.315 -This draft 1.316 -message is special. It can not be simply listed with the available tools, 1.317 -but instead requires special switches. I.e. corner-cases were 1.318 -introduced. A draft folder, in contrast, does not introduce such 1.319 -corner-cases. The available tools can operate on the messages within that 1.320 -folder like on any messages within any mail folders. The only difference 1.321 -is the fact that the default folder for 1.322 -.Pn send 1.323 -is the draft folder, 1.324 -instead of the current folder, like for all other tools. 1.325 -.P 1.326 -The trivial part of the change was activating the draft folder facility 1.327 -by default and setting a default name for this folder. Obviously, I chose 1.328 -the name 1.329 -.Fn +drafts . 1.330 -This made the 1.331 -.Sw \-draftfolder 1.332 -and 1.333 -.Sw \-draftmessage 1.334 -switches useless, and I could remove them. 1.335 -The more difficult but also the part that showed the real improvement, 1.336 -was updating the tools to the new concept. 1.337 -.Sw \-draft 1.338 -switches could 1.339 -be dropped, as operating on a draft message became indistinguishable to 1.340 -operating on any other message for the tools. 1.341 -.Pn comp 1.342 -still has its 1.343 -.Sw \-use 1.344 -switch for switching between its two modes: (1) Compose a new 1.345 -draft, possibly by taking some existing message as a form. (2) Modify 1.346 -an existing draft. In either case, the behavior of 1.347 -.Pn comp is 1.348 -deterministic. There is no more need to query the user. I consider this 1.349 -a major improvement. By making 1.350 -.Pn send 1.351 -simply operate on the current 1.352 -message in the draft folder by default, with message and folder both 1.353 -overridable by specifying them on the command line, it is now possible 1.354 -to send a draft anywhere within the storage by simply specifying its folder 1.355 -and name. 1.356 -.P 1.357 -All theses changes converted special cases to regular cases, thus 1.358 -simplifying the tools and increasing the flexibility. 1.359 +.H2 "Separation 1.360 1.361 -.U2 "Trash Folder 1.362 -.P 1.363 -Similar to the situation for drafts is the situation for removed messages. 1.364 -Historically, a message was deleted by renaming. A specific 1.365 -\fIbackup prefix\fP, often comma (\c 1.366 -.Fn , ) 1.367 -or hash (\c 1.368 -.Fn # ), 1.369 -being prepended to the file name. Thus, MH wouldn't recognize the file 1.370 -as a message anymore, as only files whose name consists of digits only 1.371 -are treated as messages. The removed messages remained as files in the 1.372 -same directory and needed some maintenance job to truly delete them after 1.373 -some grace time. Usually, by running a command similar to 1.374 -.DS 1.375 -find /home/user/Mail \-ctime +7 \-name ',*' | xargs rm 1.376 -.DE 1.377 -in a cron job. Within the grace time interval 1.378 -the original message could be restored by stripping the 1.379 -the backup prefix from the file name. If however, the last message of 1.380 -a folder is been removed \(en say message 1.381 -.Fn 6 1.382 -becomes file 1.383 -.Fn ,6 1.384 -\(en and a new message enters the same folder, thus the same 1.385 -numbered being given again \(en in our case 1.386 -.Fn 6 1.387 -\(en, if that one 1.388 -is removed too, then the backup of the former message gets overwritten. 1.389 -Thus, the ability to restore removed messages does not only depend on 1.390 -the ``sweeping cron job'' but also on the removing of further messages. 1.391 -This is undesirable, because the real mechanism is hidden from the user 1.392 -and the consequences of further removals are not always obvious. 1.393 -Further more, the backup files are scattered within the whole mail 1.394 -storage, instead of being collected at one place. 1.395 -.P 1.396 -To improve the situation, the profile entry 1.397 -.Pe rmmproc 1.398 -(previously named 1.399 -.Pe Delete-Prog ) 1.400 -was introduced, very early. 1.401 -It could be set to any command, which would care for the mail removal 1.402 -instead of taking the default action, described above. 1.403 -Refiling the to-be-removed files to some garbage folder was a common 1.404 -example. Nmh's man page 1.405 -.Mp rmm(1) 1.406 -proposes 1.407 -.Cl "refile +d 1.408 -to move messages to the garbage folder and 1.409 -.Cl "rm `mhpath +d all` 1.410 -the empty the garbage folder. 1.411 -Managing the message removal this way is a sane approach. It keeps 1.412 -the removed messages in one place, makes it easy to remove the backup 1.413 -files, and, most important, enables the user to use the tools of MH 1.414 -itself to operate on the removed messages. One can 1.415 -.Pn scan 1.416 -them, 1.417 -.Pn show 1.418 -them, and restore them with 1.419 -.Pn refile . 1.420 -There's no more 1.421 -need to use 1.422 -.Pn mhpath 1.423 -to switch over from MH tools to Unix tools \(en MH can do it all itself. 1.424 -.P 1.425 -This approach matches perfect with the concepts of MH, thus making 1.426 -it powerful. Hence, I made it the default. And even more, I also 1.427 -removed the old backup prefix approach, as it is clearly less powerful. 1.428 -Keeping unused alternative in the code is a bad choice as they likely 1.429 -gather bugs, by not being constantly tested. Also, the increased code 1.430 -size and more conditions crease the maintenance costs. By strictly 1.431 -converting to the trash folder approach, I simplified the code base. 1.432 -.Pn rmm 1.433 -calls 1.434 -.Pn refile 1.435 -internally to move the to-be-removed 1.436 -message to the trash folder (\c 1.437 -.Fn +trash 1.438 -by default). Messages 1.439 -there can be operated on like on any other message in the storage. 1.440 -The sweep clean, one can use 1.441 -.Cl "rmm \-unlink +trash a" , 1.442 -where the 1.443 -.Sw \-unlink 1.444 -switch causes the files to be truly unliked instead 1.445 -of moved to the trash folder. 1.446 - 1.447 - 1.448 -.H1 "MH Directory Split 1.449 +.U2 "MH Directory Split 1.450 .P 1.451 In MH and nmh, a personal setup had consisted of two parts: 1.452 The MH profile, named 1.453 @@ -563,77 +485,203 @@ 1.454 split between mail storage and personal configuration files. 1.455 1.456 1.457 -.H1 "Path Notations 1.458 +.H2 "Modularization 1.459 .P 1.460 -foo 1.461 - 1.462 -.H1 "Attachments 1.463 +whatnowproc 1.464 .P 1.465 -foo 1.466 - 1.467 -.H1 "mhshow to show Transition 1.468 -.P 1.469 -Since the very beginning, already in the first concept paper, 1.470 -.Pn show 1.471 -had been MH's mail display program. 1.472 -.Pn show 1.473 -found out which pathnames the relevant messages had and invoked 1.474 -.Pn mhl 1.475 -then to let it render the content. 1.476 -With the advent of MIME, this approach wasn't sufficient anymore. 1.477 -MIME messages can consist of multiple parts, some of which aren't 1.478 -directly displayable, and text content can be encoded in 1.479 -foreign charsets. 1.480 -.Pn show 's 1.481 -simple approach and 1.482 -.Pn mhl 's 1.483 -limited display facilities couldn't cope with the task any longer. 1.484 -Instead of extending these tools, new ones were written from scratch 1.485 -and then added to the MH tool chest. Doing so is encouraged by the 1.486 -tool chest approach. The new tools could be added without interfering 1.487 -with the existing ones. This is great. It allowed MH to be the 1.488 -first MUA to implement MIME. 1.489 -.P 1.490 -The new MIME features were added in form of the single program 1.491 -.Pn mhn . 1.492 -The command 1.493 -.DS 1.494 -mhn \-show 42 1.495 -.DE 1.496 -would show the MIME message numbered 42. 1.497 -With the 1.0 release of nmh in February 1999, Richard Coleman finished 1.498 -the split of 1.499 -.Pn mhn 1.500 -into a set of specialized programs, which together covered the 1.501 -aspects of MIME. One of these resulting tools was 1.502 -.Pn mhshow . 1.503 - 1.504 - 1.505 -.H1 "Blind Carbon Copies 1.506 -.P 1.507 -foo 1.508 - 1.509 -.H1 "Good Defaults 1.510 -.P 1.511 -foo 1.512 - 1.513 -.H1 "Modularization 1.514 -.P 1.515 -foo 1.516 - 1.517 -.H1 "Code style 1.518 -.P 1.519 -foo 1.520 - 1.521 - 1.522 - 1.523 -(e.g. the 1.524 -user-visible access to whole messages and MIME parts are inherently 1.525 -different). 1.526 - 1.527 The \fIMH library\fP 1.528 .Fn libmh.a 1.529 collects a bunch of standard functions that many of the MH tools need, 1.530 like reading the profile or context files. 1.531 This doesn't hurt the separation. 1.532 1.533 + 1.534 +.H2 "Style 1.535 +.P 1.536 +Code layout, goto, ... 1.537 + 1.538 + 1.539 + 1.540 + 1.541 +.H1 "Concept Exploitation/Homogeniety 1.542 + 1.543 + 1.544 +.H2 "Draft Folder 1.545 +.P 1.546 +Historically, MH provided exactly one draft message, named 1.547 +.Fn draft 1.548 +and 1.549 +being located in the MH directory. When starting to compose another message 1.550 +before the former one was sent, the user had been questioned whether to use, 1.551 +refile or replace the old draft. Working on multiple drafts at the same time 1.552 +was impossible. One could only work on them in alteration by refiling the 1.553 +previous one to some directory and fetching some other one for reediting. 1.554 +This manual draft management needed to be done each time the user wanted 1.555 +to switch between editing one draft to editing another. 1.556 +.P 1.557 +To allow true parallel editing of drafts, in a straight forward way, the 1.558 +draft folder facility exists. It had been introduced already in July 1984 1.559 +by Marshall T. Rose. The facility was deactivated by default. 1.560 +Even in nmh, the draft folder facility remained deactivated by default. 1.561 +At least, Richard Coleman added the man page 1.562 +.Mp mh-draft(5) 1.563 +to document 1.564 +the feature well. 1.565 +.P 1.566 +The only advantage of not using the draft folder facility is the static 1.567 +name of the draft file. This could be an issue for MH frontends like mh-e. 1.568 +But as they likely want to provide working on multiple drafts in parallel, 1.569 +the issue is only concerning compatibility. The aim of nmh to stay compatible 1.570 +prevented the default activation of the draft folder facility. 1.571 +.P 1.572 +On the other hand, a draft folder is the much more natural concept than 1.573 +a draft message. MH's mail storage consists of folders and messages, 1.574 +the messages named with ascending numbers. A draft message breaks with this 1.575 +concept by introducing a message in a file named 1.576 +.Fn draft . 1.577 +This draft 1.578 +message is special. It can not be simply listed with the available tools, 1.579 +but instead requires special switches. I.e. corner-cases were 1.580 +introduced. A draft folder, in contrast, does not introduce such 1.581 +corner-cases. The available tools can operate on the messages within that 1.582 +folder like on any messages within any mail folders. The only difference 1.583 +is the fact that the default folder for 1.584 +.Pn send 1.585 +is the draft folder, 1.586 +instead of the current folder, like for all other tools. 1.587 +.P 1.588 +The trivial part of the change was activating the draft folder facility 1.589 +by default and setting a default name for this folder. Obviously, I chose 1.590 +the name 1.591 +.Fn +drafts . 1.592 +This made the 1.593 +.Sw \-draftfolder 1.594 +and 1.595 +.Sw \-draftmessage 1.596 +switches useless, and I could remove them. 1.597 +The more difficult but also the part that showed the real improvement, 1.598 +was updating the tools to the new concept. 1.599 +.Sw \-draft 1.600 +switches could 1.601 +be dropped, as operating on a draft message became indistinguishable to 1.602 +operating on any other message for the tools. 1.603 +.Pn comp 1.604 +still has its 1.605 +.Sw \-use 1.606 +switch for switching between its two modes: (1) Compose a new 1.607 +draft, possibly by taking some existing message as a form. (2) Modify 1.608 +an existing draft. In either case, the behavior of 1.609 +.Pn comp is 1.610 +deterministic. There is no more need to query the user. I consider this 1.611 +a major improvement. By making 1.612 +.Pn send 1.613 +simply operate on the current 1.614 +message in the draft folder by default, with message and folder both 1.615 +overridable by specifying them on the command line, it is now possible 1.616 +to send a draft anywhere within the storage by simply specifying its folder 1.617 +and name. 1.618 +.P 1.619 +All theses changes converted special cases to regular cases, thus 1.620 +simplifying the tools and increasing the flexibility. 1.621 + 1.622 + 1.623 +.H2 "Trash Folder 1.624 +.P 1.625 +Similar to the situation for drafts is the situation for removed messages. 1.626 +Historically, a message was deleted by renaming. A specific 1.627 +\fIbackup prefix\fP, often comma (\c 1.628 +.Fn , ) 1.629 +or hash (\c 1.630 +.Fn # ), 1.631 +being prepended to the file name. Thus, MH wouldn't recognize the file 1.632 +as a message anymore, as only files whose name consists of digits only 1.633 +are treated as messages. The removed messages remained as files in the 1.634 +same directory and needed some maintenance job to truly delete them after 1.635 +some grace time. Usually, by running a command similar to 1.636 +.DS 1.637 +find /home/user/Mail \-ctime +7 \-name ',*' | xargs rm 1.638 +.DE 1.639 +in a cron job. Within the grace time interval 1.640 +the original message could be restored by stripping the 1.641 +the backup prefix from the file name. If however, the last message of 1.642 +a folder is been removed \(en say message 1.643 +.Fn 6 1.644 +becomes file 1.645 +.Fn ,6 1.646 +\(en and a new message enters the same folder, thus the same 1.647 +numbered being given again \(en in our case 1.648 +.Fn 6 1.649 +\(en, if that one 1.650 +is removed too, then the backup of the former message gets overwritten. 1.651 +Thus, the ability to restore removed messages does not only depend on 1.652 +the ``sweeping cron job'' but also on the removing of further messages. 1.653 +This is undesirable, because the real mechanism is hidden from the user 1.654 +and the consequences of further removals are not always obvious. 1.655 +Further more, the backup files are scattered within the whole mail 1.656 +storage, instead of being collected at one place. 1.657 +.P 1.658 +To improve the situation, the profile entry 1.659 +.Pe rmmproc 1.660 +(previously named 1.661 +.Pe Delete-Prog ) 1.662 +was introduced, very early. 1.663 +It could be set to any command, which would care for the mail removal 1.664 +instead of taking the default action, described above. 1.665 +Refiling the to-be-removed files to some garbage folder was a common 1.666 +example. Nmh's man page 1.667 +.Mp rmm(1) 1.668 +proposes 1.669 +.Cl "refile +d 1.670 +to move messages to the garbage folder and 1.671 +.Cl "rm `mhpath +d all` 1.672 +the empty the garbage folder. 1.673 +Managing the message removal this way is a sane approach. It keeps 1.674 +the removed messages in one place, makes it easy to remove the backup 1.675 +files, and, most important, enables the user to use the tools of MH 1.676 +itself to operate on the removed messages. One can 1.677 +.Pn scan 1.678 +them, 1.679 +.Pn show 1.680 +them, and restore them with 1.681 +.Pn refile . 1.682 +There's no more 1.683 +need to use 1.684 +.Pn mhpath 1.685 +to switch over from MH tools to Unix tools \(en MH can do it all itself. 1.686 +.P 1.687 +This approach matches perfect with the concepts of MH, thus making 1.688 +it powerful. Hence, I made it the default. And even more, I also 1.689 +removed the old backup prefix approach, as it is clearly less powerful. 1.690 +Keeping unused alternative in the code is a bad choice as they likely 1.691 +gather bugs, by not being constantly tested. Also, the increased code 1.692 +size and more conditions crease the maintenance costs. By strictly 1.693 +converting to the trash folder approach, I simplified the code base. 1.694 +.Pn rmm 1.695 +calls 1.696 +.Pn refile 1.697 +internally to move the to-be-removed 1.698 +message to the trash folder (\c 1.699 +.Fn +trash 1.700 +by default). Messages 1.701 +there can be operated on like on any other message in the storage. 1.702 +The sweep clean, one can use 1.703 +.Cl "rmm \-unlink +trash a" , 1.704 +where the 1.705 +.Sw \-unlink 1.706 +switch causes the files to be truly unliked instead 1.707 +of moved to the trash folder. 1.708 + 1.709 + 1.710 +.H2 "Path Notations 1.711 +.P 1.712 +foo 1.713 + 1.714 + 1.715 +.H2 "MIME Integration 1.716 +.P 1.717 +user-visible access to whole messages and MIME parts are inherently 1.718 +different 1.719 + 1.720 + 1.721 +.H2 "Of One Cast 1.722 +.P