docs/unix-phil

annotate unix-phil-slides.ms @ 25:e6ac51dc18bf

find changesets by author, revision, files, or words in the commit message
work on the slides
author meillo@marmaro.de
date Sun, 07 Mar 2010 18:13:30 +0100
parents 2335f5658fca
children b368ff5cdb5f
rev   line source
meillo@24 1 .TL
meillo@24 2 \fR\s-6Why\s0\fP
meillo@24 3 .br
meillo@24 4 the Unix Philosophy
meillo@24 5 .br
meillo@24 6 .vs -4
meillo@24 7 \fR\s-6(still) matters\s0\fP
meillo@24 8 .AU
meillo@24 9 .ft R
meillo@24 10 .sp 1.2i
meillo@24 11 .ps 11
meillo@24 12 markus schnalke <meillo@marmaro.de>
meillo@24 13
meillo@24 14
meillo@24 15 .S "goals of this talk
meillo@24 16 .I
meillo@24 17 introduce the Unix Phil
meillo@24 18 .I
meillo@24 19 explain why most modern software is crap
meillo@24 20 .I
meillo@24 21 explain why the Unix Phil leads to better software
meillo@24 22 .I
meillo@24 23 convince you that good software is of matter
meillo@24 24 .sp 2
meillo@24 25 .I
meillo@24 26 make you think
meillo@24 27
meillo@24 28
meillo@24 29 .S "roadmap
meillo@24 30 .I
meillo@24 31 Historical background
meillo@24 32 .I
meillo@24 33 What is the Unix Philosophy?
meillo@24 34 .I
meillo@24 35 The Unix Phil after Gancarz
meillo@24 36 .I
meillo@24 37 Discussion on real world examples
meillo@24 38 .I
meillo@24 39 The Unix Phil is more than software dev guidelines
meillo@24 40
meillo@24 41
meillo@24 42
meillo@24 43 .S "me and the UP
meillo@24 44 .LP
meillo@24 45 first contact: through the suckless project
meillo@24 46 .LP
meillo@24 47 ``cat -v Considered Harmful''
meillo@24 48 .LP
meillo@24 49 ``The Unix and the Echo''
meillo@24 50 .LP
meillo@24 51 ``The Unix Programming Environment''
meillo@24 52 .sp 1
meillo@24 53 .LP
meillo@24 54 better understanding through digging in the past
meillo@24 55
meillo@24 56
meillo@24 57
meillo@24 58
meillo@24 59
meillo@24 60 .P "historical background
meillo@24 61
meillo@24 62
meillo@24 63
meillo@24 64
meillo@24 65 .S "historical background
meillo@24 66 .I
meillo@24 67 the late 60s and early 70s
meillo@24 68 .I
meillo@24 69 operating systems are complex
meillo@24 70 .I
meillo@24 71 MULTICS just failed
meillo@24 72 .I
meillo@24 73 Brooks' ``The Mythical Man-Month''
meillo@24 74 .I
meillo@24 75 a lot of different hardware
meillo@24 76 .I
meillo@24 77 limited computing power
meillo@24 78 .I
meillo@24 79 textual input and output (line printers)
meillo@24 80
meillo@24 81
meillo@24 82 .S "everything is a file
meillo@24 83 .LP
meillo@24 84 is the(?) basic concept in Unix (and even more in Plan9)
meillo@24 85 .LP
meillo@24 86 made simple operating systems possible
meillo@24 87 .LP
meillo@24 88 it is not covered by the Unix Phil; the Unix Phil is on a different level
meillo@24 89 .sp 2
meillo@24 90 .LP
meillo@24 91 Unix is mainly two things:
meillo@24 92 .I
meillo@24 93 an operating system (system calls)
meillo@24 94 .I
meillo@24 95 a toolchest (coreutils)
meillo@24 96
meillo@24 97
meillo@24 98
meillo@24 99 .P "What is the Unix Phil
meillo@24 100
meillo@24 101
meillo@24 102
meillo@24 103 .S "What is the Unix Phil *itself*?
meillo@24 104 .LP
meillo@24 105 ``The Unix philosophy is a set of cultural norms and philosophical
meillo@24 106 approaches to developing software based on the experience of
meillo@24 107 leading developers of the Unix operating system.''
meillo@24 108 (wikipedia)
meillo@24 109 .sp 2
meillo@24 110 .LP
meillo@24 111 How the inventors of Unix write software.
meillo@24 112 .LP
meillo@24 113 Common things of classic Unix tools.
meillo@24 114 .sp 2
meillo@24 115 .LP
meillo@24 116 difficult to define
meillo@24 117
meillo@24 118
meillo@24 119 .S "Unix Phil vs. SW dev processes
meillo@24 120 .LP
meillo@24 121 the Unix Phil
meillo@24 122 .I
meillo@24 123 much: *what* to program
meillo@24 124 .I
meillo@24 125 few: *how* to program
meillo@24 126
meillo@24 127 .LP
meillo@24 128 Software developments processes:
meillo@24 129 .I
meillo@24 130 few: *what* to program
meillo@24 131 .I
meillo@24 132 much: *how* to program
meillo@24 133
meillo@24 134 .sp 2
meillo@24 135 .LP
meillo@24 136 Extreme Programming is like the Unix Phil
meillo@24 137 but with more *how* than *what*,
meillo@24 138 and with formalisms
meillo@24 139
meillo@24 140
meillo@24 141
meillo@24 142
meillo@24 143
meillo@24 144 .S "What is the Unix Phil?
meillo@24 145 .I
meillo@24 146 Doug McIlroy (1978)
meillo@24 147 .I
meillo@24 148 Mike Gancarz: ``The Unix Philosophy'' (1994)
meillo@24 149 .I
meillo@24 150 Eric S. Raymond: ``The Art of Unix Programming'' (2003)
meillo@24 151 .sp 2
meillo@24 152 .I
meillo@24 153 Richard Gabriel: ``Worse is Better'' (1989)
meillo@24 154
meillo@24 155
meillo@24 156 .S "Doug McIlroy
meillo@24 157 .LP
meillo@24 158 This is the Unix philosophy:
meillo@24 159 .I
meillo@24 160 Write programs that do one thing and do it well.
meillo@24 161 .I
meillo@24 162 Write programs to work together.
meillo@24 163 .I
meillo@24 164 Write programs to handle text streams, because that is a universal interface.
meillo@24 165
meillo@24 166
meillo@24 167 .S "Mike Gancarz: ``The Unix Philosophy''
meillo@24 168 .I
meillo@24 169 Small is beautiful.
meillo@24 170 .I
meillo@24 171 Make each program do one thing well.
meillo@24 172 .I
meillo@24 173 Build a prototype as soon as possible.
meillo@24 174 .I
meillo@24 175 Choose portability over efficiency.
meillo@24 176 .I
meillo@24 177 Store data in flat text files.
meillo@24 178 .I
meillo@24 179 Use software leverage to your advantage.
meillo@24 180 .I
meillo@24 181 Use shell scripts to increase leverage and portability.
meillo@24 182 .I
meillo@24 183 Avoid captive user interfaces.
meillo@24 184 .I
meillo@24 185 Make every program a filter.
meillo@24 186 .sp 1
meillo@24 187 .LP
meillo@24 188 plus ten lesser tenets
meillo@24 189
meillo@24 190
meillo@24 191
meillo@24 192
meillo@24 193 .P "The Unix Phil after Gancarz
meillo@24 194
meillo@25 195
meillo@25 196
meillo@25 197 .S "Small is beautiful
meillo@25 198 .I
meillo@25 199 small software is easier to understand, write, maintain
meillo@25 200 .I
meillo@25 201 less lines of code contain less bugs
meillo@25 202 .I
meillo@25 203 monsters are large
meillo@25 204
meillo@25 205
meillo@25 206 .S "Make each program do one thing well
meillo@25 207 .I
meillo@25 208 programs with many functions are large
meillo@25 209 .I
meillo@25 210 one thing is easier to understand
meillo@25 211 .I
meillo@25 212 often straight forward to implement
meillo@25 213 .I
meillo@25 214 reusable
meillo@25 215
meillo@25 216
meillo@25 217 .S "Build a prototype as soon as possible
meillo@25 218 .I
meillo@25 219 shows the quality of the design
meillo@25 220 .I
meillo@25 221 shows the problems of the software
meillo@25 222 .I
meillo@25 223 the best way to shape a software
meillo@25 224 .I
meillo@25 225 users find bugs
meillo@25 226
meillo@25 227
meillo@25 228 .S "Choose portability over efficiency
meillo@25 229 .I
meillo@25 230 (comes from incompatible hardware in history)
meillo@25 231 .I
meillo@25 232 use is most important
meillo@25 233 .I
meillo@25 234 availability
meillo@25 235 .I
meillo@25 236 only needs to be fast enough
meillo@25 237
meillo@25 238
meillo@25 239 .S "Store data in flat text files
meillo@24 240 .LP
meillo@25 241 binary data is machine dependent
meillo@25 242 .LP
meillo@25 243 human readable data is:
meillo@25 244 .I
meillo@25 245 is very likely supported
meillo@25 246 .I
meillo@25 247 good to port
meillo@25 248 .I
meillo@25 249 many tools work on it (Unix toolchest)
meillo@25 250 .I
meillo@25 251 as generic as possible
meillo@25 252 .LP
meillo@25 253 processing needs only to be fast enough
meillo@24 254
meillo@24 255
meillo@25 256 .S "Use software leverage to your advantage
meillo@24 257 .LP
meillo@25 258 what do we have computers for?
meillo@25 259 .LP
meillo@25 260 make best use of computing power
meillo@25 261 .LP
meillo@25 262 reduce development effort
meillo@24 263
meillo@24 264
meillo@25 265 .S "Use shell scripts to increase leverage and portability
meillo@24 266 .LP
meillo@25 267 was very important in history
meillo@25 268 .LP
meillo@25 269 high level languages
meillo@25 270 .LP
meillo@25 271 prototyping
meillo@25 272 .LP
meillo@25 273 quick hacks
meillo@24 274
meillo@24 275
meillo@25 276 .S "Avoid captive user interfaces
meillo@24 277 .LP
meillo@25 278 don't assume the user to be human
meillo@25 279 .LP
meillo@25 280 exclude the user whenever possible
meillo@25 281 .LP
meillo@25 282 automate
meillo@24 283
meillo@24 284
meillo@25 285 .S "Make every program a filter
meillo@24 286 .LP
meillo@25 287 programs transform data
meillo@25 288 .LP
meillo@25 289 combine programs
meillo@25 290 .LP
meillo@25 291 have one common interface
meillo@24 292
meillo@24 293
meillo@24 294
meillo@25 295 .S "a different POV
meillo@25 296 .I
meillo@25 297 pipes
meillo@25 298 .I
meillo@25 299 interface design
meillo@25 300 .I
meillo@25 301 the toolchest approach
meillo@25 302 .I
meillo@25 303 a powerful shell
meillo@25 304 .I
meillo@25 305 worse is better
meillo@25 306 .I
meillo@25 307 upgrowth and survival
meillo@24 308
meillo@24 309
meillo@24 310
meillo@24 311 .P "real world examples
meillo@24 312
meillo@24 313
meillo@24 314 .S "various
meillo@24 315 .LP
meillo@24 316 who uses
meillo@24 317 .CW "grep -R
meillo@24 318 ?
meillo@24 319
meillo@24 320 .sp 1
meillo@24 321 .LP
meillo@24 322 .CW "cat -v
meillo@24 323
meillo@24 324 .sp 1
meillo@24 325 .LP
meillo@24 326 pagers are taken for granted
meillo@24 327
meillo@24 328 .sp 1
meillo@24 329 .LP
meillo@24 330 what about the readline?
meillo@24 331
meillo@24 332
meillo@24 333
meillo@24 334 .S "find -printf
meillo@24 335 .LP
meillo@24 336 How to reformat the output of find(1) to have ``FILENAME PATH''
meillo@24 337 instead of ``PATH/FILENAME''?
meillo@24 338
meillo@24 339 .LP
meillo@24 340 the ``easy'' way:
meillo@24 341 \f(CWfind /dir -printf "%P %h\en"\fP
meillo@24 342
meillo@24 343 .LP
meillo@24 344 the ``good'' way:
meillo@24 345 .br
meillo@24 346 .CW "\s-1find /dir | sed 's,\e(.*\e)/\e(.*\e),\e2 \e1,'
meillo@24 347
meillo@24 348 .LP
meillo@24 349 The difference shows off when one wants, for instance,
meillo@24 350 the path to be manipulated further.
meillo@24 351
meillo@24 352
meillo@24 353 .S "MH / nmh
meillo@24 354 .LP
meillo@24 355 a Mail User Agent (MUA)
meillo@25 356 .LP
meillo@25 357 is a toolchest
meillo@25 358 .LP
meillo@25 359 work with mails like with generic files
meillo@25 360 .LP
meillo@25 361 the only(?) MUA that follows the Unix Phil
meillo@25 362 .LP
meillo@25 363 has a very special feeling
meillo@24 364
meillo@24 365
meillo@24 366 .S "uzbl
meillo@24 367 .LP
meillo@24 368 a web browser that adheres to the Unix Phil
meillo@25 369 .LP
meillo@25 370 a young project (about 1 year)
meillo@25 371 .LP
meillo@25 372 central question: what is the one task a web browser covers?
meillo@25 373 .LP
meillo@25 374 makes very visible use of software leverage
meillo@25 375 .LP
meillo@25 376 suffers hard from our broken web
meillo@24 377
meillo@24 378
meillo@24 379
meillo@24 380
meillo@24 381
meillo@24 382 .P "more than software dev guidelines
meillo@24 383
meillo@24 384
meillo@25 385 .S "say no
meillo@25 386 .LP
meillo@25 387 in today's computer world, the Unix Phil is much of asceticism
meillo@25 388 .LP
meillo@25 389 one needs to abjure a lot of ``nice'' features
meillo@25 390 .LP
meillo@25 391 IMO that leads to a valuable attitude
meillo@25 392
meillo@25 393
meillo@25 394 .S "avoid complexity
meillo@25 395 .LP
meillo@25 396 avoid complexity first and foremost
meillo@25 397 .LP
meillo@25 398 complexity is the ``boss enemy'' software developers fight against
meillo@25 399 .LP
meillo@25 400 strive for simplicity, clarity, generality
meillo@25 401
meillo@25 402
meillo@25 403 .S "good solutions
meillo@25 404 .LP
meillo@25 405 we don't need just solutions, we need good ones
meillo@25 406 .LP
meillo@25 407 today, we can make things possible,
meillo@25 408 but we still cannot make them good
meillo@25 409
meillo@25 410
meillo@25 411 .S "live it
meillo@25 412 .LP
meillo@25 413 the Unix Phil is not just a few guidelines
meillo@25 414 .LP
meillo@25 415 you cannot follow just some of the tenets
meillo@25 416 .LP
meillo@25 417 you will not understand the Unix Phil when you don't engage with it
meillo@25 418 .sp 3
meillo@25 419 .LP
meillo@25 420 .ce
meillo@25 421 it's a philosophy \(en live it!
meillo@24 422
meillo@24 423
meillo@24 424
meillo@24 425 .P "literature
meillo@24 426
meillo@24 427
meillo@24 428 .S literature
meillo@24 429 .I
meillo@24 430 ``\fBText Processing and Typesetting with Unix\fP''
meillo@24 431 by Barron and Rees
meillo@24 432 explains everything pretty good (focus on
meillo@24 433 .CW nroff )
meillo@24 434 .I
meillo@24 435 ``\fBHeirloom Documentation Tools Nroff/Troff User's Manual\fP''
meillo@24 436 by Ossanna, Kernighan, and Ritter
meillo@24 437 is more a technical reference
meillo@24 438 .I
meillo@24 439 Various documents collected on
meillo@24 440 .CW \s-2http://troff.org\s+2
meillo@24 441
meillo@24 442
meillo@24 443 .S
meillo@24 444 .LP
meillo@24 445 this talk was prepared using tools of the Heirloom project:
meillo@24 446 .CW \s-2http://heirloom.sf.net\s+2
meillo@24 447 .LP
meillo@24 448 the slides macros are based on
meillo@24 449 .br
meillo@24 450 .CW \s-2http://repo.cat-v.org/troff-slider/\s+2
meillo@24 451 .sp
meillo@24 452
meillo@24 453 .LP
meillo@24 454 the slides are available on my website
meillo@24 455 .CW \s-2http://marmaro.de/docs\s+2
meillo@24 456 and on
meillo@24 457 .CW \s-2http://ulm.ccc.de/ChaosSeminar/\s+2
meillo@24 458 .sp
meillo@24 459
meillo@24 460 2010-03-08