Commit | Line | Data |
---|---|---|
0f4ca9d1 JM |
1 | # The Make-A-Lisp Process |
2 | ||
3 | So you want to write a Lisp interpreter? Welcome! | |
4 | ||
5 | The goal of the Make-A-Lisp project is to make it easy to write your | |
6 | own Lisp interpreter without sacrificing those many "Aha!" moments | |
7 | that come from ascending the McCarthy mountain. When you reach the peak | |
8 | of this particular mountain, you will have an interpreter for the mal | |
9 | Lisp language that is powerful enough to be self-hosting, meaning it | |
10 | will be able to run a mal interpreter written in mal itself. | |
11 | ||
12 | So jump right in (er ... start the climb)! | |
13 | ||
14 | ## Pick a language | |
15 | ||
16 | You might already have a language in mind that you want to use. | |
17 | Technically speaking, mal can be implemented in any sufficiently | |
c10dcb94 | 18 | complete programming language (i.e. Turing complete), however, there are a few |
0f4ca9d1 JM |
19 | language features that can make the task MUCH easier. Here are some of |
20 | them in rough order of importance: | |
21 | ||
22 | * A sequential compound data structure (e.g. arrays, lists, | |
23 | vectors, etc) | |
24 | * An associative compound data structure (e.g. a dictionary, | |
25 | hash-map, associative array, etc) | |
26 | * Function references (first class functions, function pointers, | |
27 | etc) | |
28 | * Real exception handling (try/catch, raise, throw, etc) | |
29 | * Variable argument functions (variadic, var args, splats, apply, etc) | |
30 | * Function closures | |
31 | * PCRE regular expressions | |
32 | ||
33 | In addition, the following will make your task especially easy: | |
34 | ||
bf518367 JM |
35 | * Dynamic typing / boxed types (specifically, the ability to store |
36 | different data types in the sequential and associative structures | |
37 | and the language keeps track of the type for you) | |
0f4ca9d1 JM |
38 | * Compound data types support arbitrary runtime "hidden" data |
39 | (metadata, metatables, dynamic fields attributes) | |
40 | ||
41 | Here are some examples of languages that have all of the above | |
42 | features: JavaScript, Ruby, Python, Lua, R, Clojure. | |
43 | ||
44 | Many of the most popular languages already have Mal implementations. | |
45 | However, this should not discourage you from creating your own | |
46 | implementation in a language that already has one. However, if you go | |
47 | this route, I suggest you avoid referring to the existing | |
48 | implementations (i.e. "cheating") to maximize your learning experience | |
49 | instead of just borrowing mine. On the other hand, if your goal is to | |
50 | add new implementations to mal as efficiently as possible, then you | |
51 | SHOULD find the most similar target language implementation and refer | |
52 | to it frequently. | |
53 | ||
bf518367 JM |
54 | If you want a fairly long list of programming languages with an |
55 | approximate measure of popularity, try the [Programming Language | |
56 | Popularity Chart](http://langpop.corger.nl/) | |
57 | ||
0f4ca9d1 JM |
58 | |
59 | ## Getting started | |
60 | ||
61 | * Install your chosen language interpreter/compiler, language package | |
62 | manager and build tools (if applicable) | |
63 | ||
64 | * Fork the mal repository on github and then clone your forked | |
65 | repository: | |
66 | ``` | |
67 | git clone git@github.com:YOUR_NAME/mal.git | |
68 | cd mal | |
69 | ``` | |
70 | ||
fc03712f EC |
71 | * Make a new directory for your implementation. For example, if your |
72 | language is called "quux": | |
0f4ca9d1 JM |
73 | ``` |
74 | mkdir quux | |
75 | ``` | |
76 | ||
77 | * Modify the top level Makefile to allow the tests to be run against | |
78 | your implementation. For example, if your language is named "quux" | |
79 | and uses "qx" as the file extension, then make the following | |
80 | 3 modifications to Makefile: | |
81 | ``` | |
82 | IMPLS = ... quux ... | |
83 | ... | |
84 | quux_STEP_TO_PROG = mylang/$($(1)).qx | |
85 | ... | |
86 | quux_RUNSTEP = ../$(2) $(3) | |
87 | ``` | |
88 | ||
89 | This allows you to run tests against your implementation like this: | |
90 | ``` | |
91 | make test^quux^stepX | |
92 | ``` | |
93 | ||
94 | ||
95 | ## General hints | |
96 | ||
97 | Stackoverflow and Google are your best friends. Modern polyglot | |
98 | developers do not memorize dozens of programming languages. Instead, | |
99 | they learn the peculiar terminology used with each language and then | |
100 | use this to search for their answers. | |
101 | ||
102 | Here are some other resources where multiple languages are | |
103 | compared/described: | |
104 | * http://learnxinyminutes.com/ | |
105 | * http://hyperpolyglot.org/ | |
106 | * http://rosettacode.org/ | |
334a71b6 | 107 | * http://rigaux.org/language-study/syntax-across-languages/ |
0f4ca9d1 JM |
108 | |
109 | Do not let yourself be bogged down by specific problems. While the | |
110 | make-a-lisp process is structured as a series of steps, the reality is | |
111 | that building a lisp interpreter is more like a branching tree. If you | |
112 | get stuck on tail call optimization, or hash-maps, move on to other | |
113 | things. You will often have a stroke of inspiration for a problem as | |
114 | you work through other functionality. I have tried to structure this | |
45a8b3ca JM |
115 | guide and the tests to make clear which things can be deferred until |
116 | later. | |
117 | ||
118 | An aside on deferrable/optional bits: when you run the tests for | |
119 | a given step, the last tests are often marked with an "optional" | |
120 | header. This indicates that these are tests for functionality that is | |
121 | not critical to finish a basic mal implementation. Many of the steps | |
122 | in this process guide have a "Deferrable" section, however, it is not | |
123 | quite the same meaning. Those sections include the functionality that | |
124 | is marked as optional in the tests, but they also include | |
0f4ca9d1 JM |
125 | functionality that becomes mandatory at a later step. In other words, |
126 | this is a "make your own Lisp adventure". | |
127 | ||
128 | Use test driven development. Each step of the make-a-lisp process has | |
129 | a bunch of tests associated with it and there is an easy script to run | |
130 | all the tests for a specific step in the process. Pick a failing test, | |
131 | fix it, repeat until all the tests for that step pass. | |
132 | ||
133 | The `process` directory contains abbreviated pseudocode and | |
134 | architecture images for each step of the make-a-lisp process. Use | |
135 | a textual diff/comparison tool to compare the previous pseudocode step | |
136 | with the one you are working on. The architecture images have changes | |
137 | from the previous step highlighted in red. | |
138 | ||
daa1cf3f JM |
139 | If you get completely stuck and are feeling like giving up, then you |
140 | should "cheat" by referring to the same step or functionality in | |
141 | a existing implementation language. You are here to learn, not to take | |
142 | a test, so do not feel bad about it. Okay, you should feel a little | |
143 | bit bad about it. | |
0f4ca9d1 JM |
144 | |
145 | ||
146 | ## The Make-A-Lisp Process | |
147 | ||
148 | In the steps that follow the name of the target language is "quux" and | |
149 | the file extension for that language is "qx". | |
150 | ||
151 | ||
152 | <a name="step0"></a> | |
153 | ||
154 | ### Step 0: The REPL | |
155 | ||
156 | ![step0_repl architecture](step0_repl.png) | |
157 | ||
158 | This step is basically just creating a skeleton of your interpreter. | |
159 | ||
160 | * Create a `step0_repl.qx` file in `quux/`. | |
161 | ||
162 | * Add the 4 trivial functions `READ`, `EVAL`, `PRINT`, and `rep` | |
163 | (read-eval-print). `READ`, `EVAL`, and `PRINT` are basically just | |
164 | stubs that return their first parameter (a string if your target | |
165 | language is a statically typed) and `rep` calls them in order | |
166 | passing the return to the input of the next. | |
167 | ||
58ba5af4 JM |
168 | * Add a main loop that repeatedly prints a prompt (needs to be |
169 | "user> " for later tests to pass), gets a line of input from the | |
170 | user, calls `rep` with that line of input, and then prints out the | |
171 | result from `rep`. It should also exit when you send it an EOF | |
172 | (often Ctrl-D). | |
0f4ca9d1 JM |
173 | |
174 | * If you are using a compiled (ahead-of-time rather than just-in-time) | |
175 | language, then create a Makefile (or appropriate project definition | |
176 | file) in your directory. | |
177 | ||
daa1cf3f JM |
178 | It is time to run your first tests. This will check that your program |
179 | does input and output in a way that can be captured by the test | |
180 | harness. Go to the top level and run the following: | |
181 | ``` | |
182 | make test^quux^step0 | |
183 | ``` | |
0f4ca9d1 JM |
184 | |
185 | Add and then commit your new `step0_repl.qx` and `Makefile` to git. | |
186 | ||
187 | Congratulations! You have just completed the first step of the | |
188 | make-a-lisp process. | |
189 | ||
190 | ||
bd62ff74 | 191 | #### Optional: |
0f4ca9d1 JM |
192 | |
193 | * Add full line editing and command history support to your | |
194 | interpreter REPL. Many languages have a library/module that provide | |
195 | line editing support. Another option if your language supports it is | |
196 | to use an FFI (foreign function interface) to load and call directly | |
197 | into GNU readline, editline, or libnoise library. Add line | |
198 | editing interface code to `readline.qx` | |
199 | ||
200 | ||
201 | <a name="step1"></a> | |
202 | ||
203 | ### Step 1: Read and Print | |
204 | ||
205 | ![step1_read_print architecture](step1_read_print.png) | |
206 | ||
207 | In this step, your interpreter will "read" the string from the user | |
208 | and parse it into an internal tree data structure (an abstract syntax | |
209 | tree) and then take that data structure and "print" it back to | |
210 | a string. | |
211 | ||
212 | In non-lisp languages, this step (called "lexing and parsing") can be | |
213 | one of the most complicated parts of the compiler/interpreter. In | |
214 | Lisp, the data structure that you want in memory is basically | |
215 | represented directly in the code that the programmer writes | |
216 | (homoiconicity). | |
217 | ||
218 | For example, if the string is "(+ 2 (* 3 4))" then the read function | |
219 | will process this into a tree structure that looks like this: | |
220 | ``` | |
221 | List | |
222 | / | \ | |
223 | / | \ | |
224 | / | \ | |
225 | Sym:+ Int:2 List | |
226 | / | \ | |
227 | / | \ | |
228 | / | \ | |
229 | Sym:* Int:3 Int:4 | |
230 | ``` | |
231 | ||
232 | Each left paren and its matching right paren (lisp "sexpr") becomes | |
233 | a node in the tree and everything else becomes a leaf in the tree. | |
234 | ||
235 | If you can find code for an implementation of a JSON encoder/decoder | |
236 | in your target language then you can probably just borrow and modify | |
237 | that and be 75% of the way done with this step. | |
238 | ||
239 | The rest of this section is going to assume that you are not starting | |
240 | from an existing JSON encoder/decoder, but that you do have access to | |
241 | a Perl compatible regular expressions (PCRE) module/library. You can | |
242 | certainly implement the reader using simple string operations, but it | |
243 | is more involved. The `make`, `ps` (postscript) and Haskell | |
244 | implementations have examples of a reader/parser without using regular | |
245 | expression support. | |
246 | ||
247 | * Copy `step0_repl.qx` to `step1_read_print.qx`. | |
248 | ||
249 | * Add a `reader.qx` file to hold functions related to the reader. | |
250 | ||
251 | * If the target language has objects types (OOP), then the next step | |
252 | is to create a simple stateful Reader object in `reader.qx`. This | |
253 | object will store the tokens and a position. The Reader object will | |
254 | have two methods: `next` and `peek`. `next` returns the tokens at | |
255 | the current position and increments the position. `peek` just | |
256 | returns the token at the current position. | |
257 | ||
258 | * Add a function `read_str` in `reader.qx`. This function | |
259 | will call `tokenizer` and then create a new Reader object instance | |
260 | with the tokens. Then it will call `read_form` with the Reader | |
261 | instance. | |
262 | ||
263 | * Add a function `tokenizer` in `reader.qx`. This function will take | |
2315fd53 | 264 | a single string and return an array/list |
0f4ca9d1 JM |
265 | of all the tokens (strings) in it. The following regular expression |
266 | (PCRE) will match all mal tokens. | |
267 | ``` | |
268 | [\s,]*(~@|[\[\]{}()'`~^@]|"(?:\\.|[^\\"])*"|;.*|[^\s\[\]{}('"`,;)]*) | |
269 | ``` | |
270 | ||
271 | * Add the function `read_form` to `reader.qx`. This function | |
272 | will peek at the first token in the Reader object and switch on the | |
273 | first character of that token. If the character is a left paren then | |
274 | `read_list` is called with the Reader object. Otherwise, `read_atom` | |
275 | is called with the Reader Object. The return value from `read_form` | |
276 | is a mal data type. If your target language is statically typed then | |
277 | you will need some way for `read_form` to return a variant or | |
278 | subclass type. For example, if your language is object oriented, | |
2315fd53 | 279 | then you can define a top level MalType (in `types.qx`) that all |
0f4ca9d1 JM |
280 | your mal data types inherit from. The MalList type (which also |
281 | inherits from MalType) will contains a list/array of other MalTypes. | |
282 | If your language is dynamically typed then you can likely just | |
283 | return a plain list/array of other mal types. | |
284 | ||
285 | * Add the function `read_list` to `reader.qx`. This function will | |
286 | repeatedly call `read_form` with the Reader object until it | |
287 | encounters a ')' token (if it reach EOF before reading a ')' then | |
288 | that is an error). It accumulates the results into a List type. If | |
289 | your language does not have a sequential data type that can hold mal | |
290 | type values you may need to implement one (in `types.qx`). Note | |
291 | that `read_list` repeatedly calls `read_form` rather than | |
292 | `read_atom`. This mutually recursive defintion between `read_list` | |
293 | and `read_form` is what allows lists to contain lists. | |
294 | ||
295 | * Add the function `read_atom` to `reader.qx`. This function will | |
296 | look at the contents of the token and return the appropriate scalar | |
297 | (simple/single) data type value. Initially, you can just implement | |
298 | numbers (integers) and symbols . This will allow you to proceed | |
299 | through the next couple of steps before you will need to implement | |
300 | the other fundamental mal types: nil, true, false, and string. The | |
301 | remaining mal types: keyword, vector, hash-map, and atom do not | |
302 | need to be implemented until step 9 (but can be implemented at any | |
303 | point between this step and that). BTW, symbols types are just an | |
304 | object that contains a single string name value (some languages have | |
305 | symbol types already). | |
306 | ||
307 | * Add a file `printer.qx`. This file will contain a single function | |
308 | `pr_str` which does the opposite of `read_str`: take a mal data | |
309 | structure and return a string representation of it. But `pr_str` is | |
310 | much simpler and is basically just a switch statement on the type of | |
311 | the input object: | |
312 | ||
313 | * symbol: return the string name of the symbol | |
314 | * number: return the number as a string | |
315 | * list: iterate through each element of the list calling `pr_str` on | |
316 | it, then join the results with a space separator, and surround the | |
317 | final result with parens | |
318 | ||
319 | * Change the `READ` function in `step1_read_print.qx` to call | |
320 | `reader.read_str` and the `PRINT` function to call `printer.pr_str`. | |
321 | `EVAL` continues to simply return its input but the type is now | |
322 | a mal data type. | |
323 | ||
324 | You now have enough hooked up to begin testing your code. You can | |
325 | manually try some simple inputs: | |
326 | * `123` -> `123` | |
327 | * ` 123 ` -> `123` | |
328 | * `abc` -> `abc` | |
329 | * ` abc ` -> `abc` | |
330 | * `(123 456)` -> `(123 456)` | |
331 | * `( 123 456 789 ) ` -> `(123 456 789)` | |
6a767b0d | 332 | * `( + 2 (* 3 4) ) ` -> `(+ 2 (* 3 4))` |
0f4ca9d1 JM |
333 | |
334 | To verify that your code is doing more than just eliminating extra | |
335 | spaces (and not failing), you can instrument your `reader.qx` functions. | |
336 | ||
337 | Once you have gotten past those simple manual tests, it is time to run | |
338 | the full suite of step 1 tests. Go to the top level and run the | |
339 | following: | |
340 | ``` | |
341 | make test^quux^step1 | |
342 | ``` | |
343 | ||
344 | Fix any test failures related to symbols, numbers and lists. | |
345 | ||
346 | Depending on the functionality of your target language, it is likely | |
347 | that you have now just completed one of the most difficult steps. It | |
348 | is down hill from here. The remaining steps will probably be easier | |
349 | and each step will give progressively more bang for the buck. | |
350 | ||
45a8b3ca | 351 | #### Deferrable: |
0f4ca9d1 JM |
352 | |
353 | ||
354 | * Add error checking to your reader functions to make sure parens | |
355 | are properly matched. Catch and print these errors in your main | |
356 | loop. If your language does not have try/catch style bubble up | |
357 | exception handling, then you will need to add explicit error | |
358 | handling to your code to catch and pass on errors without crashing. | |
359 | ||
360 | * Add support for the other basic data type to your reader and printer | |
361 | functions: string, nil, true, and false. These become mandatory at | |
362 | step 4. When a string is read, a slash followed by a doublequote is | |
943f21c4 | 363 | translated into a plain doublequote character and a slash followed by |
0f4ca9d1 JM |
364 | "n" is translated into a newline. To properly print a string (for |
365 | step 4 string functions), the `pr_str` function needs another | |
366 | parameter called `print_readably`. When `print_readably` is true, | |
367 | doublequotes and newlines are translated into their printed | |
368 | representations (the reverse of the reader). The `PRINT` function in | |
369 | the main program should call `pr_str` with print_readably set to | |
370 | true. | |
371 | ||
372 | * Add support for the other mal types: keyword, vector, hash-map, and | |
2345a3da JM |
373 | atom. |
374 | * keyword: a keyword is a token that begins with a colon. A keyword | |
375 | can just be stored as a string with special unicode prefix like | |
376 | 0x29E (or char 0xff/127 if the target language does not have good | |
377 | unicode support) and the printer translates strings with that | |
378 | prefix back to the keyword representation. This makes it easy to | |
379 | use keywords as hash map keys in most languages. You can also | |
380 | store keywords as a unique data type, but you will need to make | |
381 | sure they can be used as hash map keys (which may involve doing | |
382 | a similar prefixed translation anyways). | |
383 | * vector: a vector can be implemented with same underlying | |
384 | type as a list as long as there is some mechanism to keep track of | |
385 | the difference. You can use the same reader function for both | |
386 | lists and vectors by adding parameters for the starting and ending | |
387 | tokens. | |
388 | * hash-map: a hash-map is an associative data structure that maps | |
389 | strings to other mal values. If you implement keywords as prefixed | |
390 | strings, then you only need a native associative data structure | |
391 | which supports string keys. Clojure allows any value to be a hash | |
392 | map key, but the base functionality in mal is to support strings | |
393 | and keyword keys. Because of the representation of hash-maps as | |
394 | an alternating sequence of keys and values, you can probably use | |
395 | the same reader function for hash-maps as lists and vectors with | |
396 | parameters to indicate the starting and ending tokens. The odd | |
397 | tokens are then used for keys with the corresponding even tokens | |
398 | as the values. | |
399 | ||
400 | * Add support for reader macros which are forms that are | |
401 | transformed into other forms during the read phase. Refer to | |
402 | `tests/step1_read_print.mal` for the form that these macros should | |
403 | take (they are just simple transformations of the token stream). | |
0f4ca9d1 JM |
404 | |
405 | * Add comment support to your reader. The tokenizer should ignore | |
406 | tokens that start with ";". Your `read_str` function will need to | |
407 | properly handle when the tokenizer returns no values. The simplest | |
408 | way to do this is to return `nil` mal value. A cleaner option (that | |
409 | does not print `nil` at the prompt is to throw a special exception | |
410 | that causes the main loop to simply continue at the beginning of the | |
411 | loop without calling `rep`. | |
412 | ||
413 | ||
414 | <a name="step2"></a> | |
415 | ||
416 | ### Step 2: Eval | |
417 | ||
418 | ![step2_eval architecture](step2_eval.png) | |
419 | ||
420 | In step 1 your mal interpreter was basically just a way to validate | |
421 | input and eliminate extraneous white space. In this step you will turn | |
422 | your interpreter into a simple number calculator by adding | |
423 | functionality to the evaluator (`EVAL`). | |
424 | ||
425 | Compare the pseudocode for step 1 and step 2 to get a basic idea of | |
426 | the changes that will be made during this step: | |
427 | ``` | |
428 | diff -urp ../process/step1_read_print.txt ../process/step2_eval.txt | |
429 | ``` | |
430 | ||
431 | * Copy `step1_read_print.qx` to `step2_eval.qx`. | |
432 | ||
433 | * Define a simple initial REPL environment. This environment is an | |
434 | associative structure that maps symbols (or symbol names) to | |
435 | numeric functions. For example, in python this would look something | |
436 | like this: | |
437 | ``` | |
438 | repl_env = {'+': lambda a,b: a+b, | |
439 | '-': lambda a,b: a-b, | |
440 | '*': lambda a,b: a*b, | |
441 | '/': lambda a,b: int(a/b)} | |
442 | ``` | |
443 | ||
444 | * Modify the `rep` function to pass the REPL environment as the second | |
445 | parameter for the `EVAL` call. | |
446 | ||
447 | * Create a new function `eval_ast` which takes `ast` (mal data type) | |
448 | and an associative structure (the environment from above). | |
449 | `eval_ast` switches on the type of `ast` as follows: | |
450 | ||
451 | * symbol: lookup the symbol in the environment structure and return | |
452 | the value or raise an error no value is found | |
453 | * list: return a new list that is the result of calling `EVAL` on | |
454 | each of the members of the list | |
455 | * otherwise just return the original `ast` value | |
456 | ||
457 | * Modify `EVAL` to check if the first parameter `ast` is a list. | |
458 | * `ast` is not a list: then return the result of calling `eval_ast` | |
459 | on it. | |
460 | * `ast` is a list: call `eval_ast` to get a new evaluated list. Take | |
461 | the first item of the evaluated list and call it as function using | |
462 | the rest of the evaluated list as its arguments. | |
463 | ||
464 | If your target language does not have full variable length argument | |
465 | support (e.g. variadic, vararg, splats, apply) then you will need to | |
466 | pass the full list of arguments as a single parameter and split apart | |
467 | the individual values inside of every mal function. This is annoying, | |
468 | but workable. | |
469 | ||
470 | The process of taking a list and invoking or executing it to return | |
471 | something new is known in Lisp as the "apply" phase. | |
472 | ||
473 | Try some simple expressions: | |
474 | ||
475 | * `(+ 2 3)` -> `5` | |
476 | * `(+ 2 (* 3 4))` -> `14` | |
477 | ||
478 | The most likely challenge you will encounter is how to properly call | |
479 | a function references using an arguments list. | |
480 | ||
481 | Now go to the top level, run the step 2 tests and fix the errors. | |
482 | ``` | |
483 | make test^quux^step2 | |
484 | ``` | |
485 | ||
486 | You now have a simple prefix notation calculator! | |
487 | ||
488 | ||
489 | <a name="step3"></a> | |
490 | ||
491 | ### Step 3: Environments | |
492 | ||
493 | ![step3_env architecture](step3_env.png) | |
494 | ||
495 | In step 2 you were already introduced to REPL environment (`repl_env`) | |
496 | where the basic numeric functions were stored and looked up. In this | |
497 | step you will add the ability to create new environments (`let*`) and | |
daa1cf3f | 498 | modify existing environments (`def!`). |
0f4ca9d1 JM |
499 | |
500 | A Lisp environment is an associative data structure that maps symbols (the | |
501 | keys) to values. But Lisp environments have an additional important | |
502 | function: they can refer to another environment (the outer | |
503 | environment). During environment lookups, if the current environment | |
504 | does not have the symbol, the lookup continues in the outer | |
505 | environment, and continues this way until the symbol is either found, | |
506 | or the outer environment is `nil` (the outermost environment in the | |
507 | chain). | |
508 | ||
509 | Compare the pseudocode for step 2 and step 3 to get a basic idea of | |
510 | the changes that will be made during this step: | |
511 | ``` | |
512 | diff -urp ../process/step2_eval.txt ../process/step3_env.txt | |
513 | ``` | |
514 | ||
d9c020b0 | 515 | * Copy `step2_eval.qx` to `step3_env.qx`. |
0f4ca9d1 JM |
516 | |
517 | * Create `env.qx` to hold the environment definition. | |
518 | ||
519 | * Define an `Env` object that is instantiated with a single `outer` | |
520 | parameter and starts with an empty associative data structure | |
521 | property `data`. | |
522 | ||
523 | * Define three methods for the Env object: | |
524 | * set: takes a symbol key and a mal value and adds to the `data` | |
525 | structure | |
526 | * find: takes a symbol key and if the current environment contains | |
527 | that key then return the environment. If no key is found and outer | |
528 | is not `nil` then call find (recurse) on the outer environment. | |
529 | * get: takes a symbol key and uses the `find` method to locate the | |
530 | environment with the key, then returns the matching value. If no | |
531 | key is found up the outer chain, then throws/raises a "not found" | |
532 | error. | |
533 | ||
169ddeb2 | 534 | * Update `step3_env.qx` to use the new `Env` type to create the |
0f4ca9d1 JM |
535 | repl_env (with a `nil` outer value) and use the `set` method to add |
536 | the numeric functions. | |
537 | ||
538 | * Modify `eval_ast` to call the `get` method on the `env` parameter. | |
539 | ||
540 | * Modify the apply section of `EVAL` to switch on the first element of | |
541 | the list: | |
542 | * symbol "def!": call the set method of the current environment | |
543 | (second parameter of `EVAL` called `env`) using the unevaluated | |
544 | first parameter (second list element) as the symbol key and the | |
545 | evaluated second parameter as the value. | |
546 | * symbol "let*": create a new environment using the current | |
547 | environment as the outer value and then use the first parameter as | |
0a63c9f9 | 548 | a list of new bindings in the "let*" environment. Take the second |
0f4ca9d1 JM |
549 | element of the binding list, call `EVAL` using the new "let*" |
550 | environment as the evaluation environment, then call `set` on the | |
0a63c9f9 | 551 | "let*" environment using the first binding list element as the key |
0f4ca9d1 JM |
552 | and the evaluated second element as the value. This is repeated |
553 | for each odd/even pair in the binding list. Note in particular, | |
554 | the bindings earlier in the list can be referred to by later | |
555 | bindings. Finally, the second parameter (third element) of the | |
556 | original `let*` form is evaluated using the new "let*" environment | |
557 | and the result is returned as the result of the `let*` (the new | |
558 | let environment is discarded upon completion). | |
559 | * otherwise: call `eval_ast` on the list and apply the first element | |
560 | to the rest as before. | |
561 | ||
562 | `def!` and `let*` are Lisp "specials" (or "special atoms") which means | |
563 | that they are language level features and more specifically that the | |
564 | rest of the list elements (arguments) may be evaluated differently (or | |
565 | not at all) unlike the default apply case where all elements of the | |
566 | list are evaluated before the first element is invoked. Lists which | |
567 | contain a "special" as the first element are known as "special forms". | |
568 | The are special because the follow special evaluation rules. | |
569 | ||
570 | Try some simple environment tests: | |
571 | ||
572 | * `(def! a 6)` -> `6` | |
573 | * `a` -> `6` | |
574 | * `(def! b (+ a 2))` -> `8` | |
575 | * `(+ a b)` -> `14` | |
576 | * `(let* (c 2) c)` -> `2` | |
577 | ||
578 | Now go to the top level, run the step 3 tests and fix the errors. | |
579 | ``` | |
580 | make test^quux^step3 | |
581 | ``` | |
582 | ||
583 | You mal implementation is still basically just a numeric calculator | |
584 | with save/restore capability. But you have set the foundation for step | |
585 | 4 where it will begin to feel like a real programming language. | |
586 | ||
587 | ||
588 | An aside on mutation and typing: | |
589 | ||
590 | The "!" suffix on symbols is used to indicate that this symbol refers | |
591 | to a function that mutates something else. In this case, the `def!` | |
592 | symbol indicates a special form that will mutate the current | |
593 | environment. Many (maybe even most) of runtime problems that are | |
594 | encountered in software engineering are a result of mutation. By | |
595 | clearly marking code where mutation may occur, you can more easily | |
596 | track down the likely cause of runtime problems when they do occur. | |
597 | ||
598 | Another cause of runtime errors is type errors, where a value of one | |
599 | type is unexpectedly treated by the program as a different and | |
600 | incompatible type. Statically typed languages try to make the | |
601 | programmer solve all type problems before the program is allowed to | |
602 | run. Most Lisp variants tend to be dynamically typed (types of values | |
603 | are checked when they are actually used at runtime). | |
604 | ||
605 | As an aside-aside: The great debate between static and dynamic typing | |
606 | debate can be understood by following the money. Advocates of strict | |
607 | static typing use words like "correctness" and "safety" and thus get | |
608 | government and academic funding. Advocates of dynamic typing use words | |
609 | like "agile" and "time-to-market" and thus get venture capital and | |
610 | commercial funding. | |
611 | ||
612 | ||
613 | <a name="step4"></a> | |
614 | ||
615 | ### Step 4: If Fn Do | |
616 | ||
617 | ![step4_if_fn_do architecture](step4_if_fn_do.png) | |
618 | ||
619 | In step 3 you added environments and the special forms for | |
620 | manipulating environments. In this step you will add 3 new special | |
621 | forms (`if`, `fn*` and `do`) and add several more core functions to | |
622 | the default REPL environment. Our new architecture will look like | |
623 | this: | |
624 | ||
625 | The `fn*` special form is how new user-defined functions are created. | |
626 | In some Lisps, this special form is named "lambda". | |
627 | ||
628 | Compare the pseudocode for step 3 and step 4 to get a basic idea of | |
629 | the changes that will be made during this step: | |
630 | ``` | |
631 | diff -urp ../process/step3_env.txt ../process/step4_if_fn_do.txt | |
632 | ``` | |
633 | ||
634 | * Copy `step3_env.qx` to `step4_if_fn_do.qx`. | |
635 | ||
636 | * If you have not implemented reader and printer support (and data | |
637 | types) for `nil`, `true` and `false`, you will need to do so for | |
638 | this step. | |
639 | ||
640 | * Update the constructor/initializer for environments to take two new | |
641 | arguments: `binds` and `exprs`. Bind (`set`) each element (symbol) | |
642 | of the binds list to the respective element of the `exprs` list. | |
643 | ||
644 | * Add support to `printer.qx` to print functions values. A string | |
645 | literal like "#<function>" is sufficient. | |
646 | ||
1af1f7f1 | 647 | * Add the following special forms to `EVAL`: |
0f4ca9d1 | 648 | |
1af1f7f1 | 649 | * `do`: Evaluate all the elements of the list using `eval_ast` |
f558a0c8 | 650 | and return the final evaluated element. |
0f4ca9d1 JM |
651 | * `if`: Evaluate the first parameter (second element). If the result |
652 | (condition) is anything other than `nil` or `false`, then evaluate | |
1af1f7f1 | 653 | the second parameter (third element of the list) and return the |
0f4ca9d1 JM |
654 | result. Otherwise, evaluate the third parameter (fourth element) |
655 | and return the result. If condition is false and there is no third | |
656 | parameter, then just return `nil`. | |
657 | * `fn*`: Return a new function closure. The body of that closure | |
658 | does the following: | |
659 | * Create a new environment using `env` (closed over from outer | |
660 | scope) as the `outer` parameter, the first parameter (second | |
661 | list element of `ast` from the outer scope) as the `binds` | |
662 | parameter, and the parameters to the closure as the `exprs` | |
663 | parameter. | |
664 | * Call `EVAL` on the second parameter (third list element of `ast` | |
665 | from outer scope), using the new environment. Use the result as | |
666 | the return value of the closure. | |
667 | ||
668 | If your target language does not support closures, then you will need | |
669 | to implement `fn*` using some sort of structure or object that stores | |
670 | the values being closed over: the first and second elements of the | |
671 | `ast` list (function parameter list and function body) and the current | |
672 | environment `env`. In this case, your native functions will need to be | |
673 | wrapped in the same way. You will probably also need a method/function | |
674 | that invokes your function object/structure for the default case of | |
675 | the apply section of `EVAL`. | |
676 | ||
677 | Try out the basic functionality you have implemented: | |
678 | ||
679 | * `(fn* [a] a)` -> `#<function>` | |
680 | * `( (fn* [a] a) 7)` -> `7` | |
681 | * `( (fn* [a] (+ a 1)) 10)` -> `11` | |
682 | * `( (fn* [a b] (+ a b)) 2 3)` -> `5` | |
683 | ||
684 | * Add a new file `core.qx` and define an associative data structure | |
685 | `ns` (namespace) that maps symbols to functions. Move the numeric | |
686 | function definitions into this structure. | |
687 | ||
688 | * Modify `step4_if_fn_do.qx` to iterate through the `core.ns` | |
689 | structure and add (`set`) each symbol/function mapping to the | |
690 | REPL environment (`repl_env`). | |
691 | ||
692 | * Add the following functions to `core.ns`: | |
693 | * `list`: take the parameters and return them as a list. | |
694 | * `list?`: return true if the first parameter is a list, false | |
695 | otherwise. | |
696 | * `empty?`: treat the first parameter as a list and return true if | |
697 | the list is empty and false if it contains any elements. | |
698 | * `count`: treat the first parameter as a list and return the number | |
699 | of elements that it contains. | |
700 | * `=`: compare the first two parameters and return true if they are | |
701 | the same type and contain the same value. In the case of equal | |
702 | length lists, each element of the list should be compared for | |
703 | equality and if they are the same return true, otherwise false. | |
704 | * `<`, `<=`, `>`, and `>=`: treat the first two parameters as | |
705 | numbers and do the corresponding numeric comparison, returning | |
706 | either true or false. | |
707 | ||
708 | Now go to the top level, run the step 4 tests. There are a lot of | |
709 | tests in step 4 but all of the non-optional tests that do not involve | |
710 | strings should be able to pass now. | |
711 | ||
712 | ``` | |
713 | make test^quux^step4 | |
714 | ``` | |
715 | ||
716 | Your mal implementation is already beginning to look like a real | |
717 | language. You have flow control, conditionals, user-defined functions | |
718 | with lexical scope, side-effects (if you implement the string | |
1380d8c1 JM |
719 | functions), etc. However, our little interpreter has not quite reached |
720 | Lisp-ness yet. The next several steps will take your implementation | |
721 | from a neat toy to a full featured language. | |
0f4ca9d1 | 722 | |
45a8b3ca | 723 | #### Deferrable: |
0f4ca9d1 JM |
724 | |
725 | * Implement Clojure-style variadic function parameters. Modify the | |
726 | constructor/initializer for environments, so that if a "&" symbol is | |
727 | encountered in the `binds` list, the next symbol in the `binds` list | |
728 | after the "&" is bound to the rest of the `exprs` list that has not | |
729 | been bound yet. | |
730 | ||
731 | * Defines a `not` function using mal itself. In `step4_if_fn_do.qx` | |
732 | call the `rep` function with this string: | |
733 | "(def! not (fn* (a) (if a false true)))". | |
734 | ||
735 | * Implement the strings functions in `core.qx`. To implement these | |
736 | functions, you will need to implement the string support in the | |
45a8b3ca | 737 | reader and printer (deferrable section of step 1). Each of the string |
0f4ca9d1 JM |
738 | functions takes multiple mal values, prints them (`pr_str`) and |
739 | joins them together into a new string. | |
740 | * `pr-str`: calls `pr_str` on each argument with `print_readably` | |
741 | set to true, joins the results with " " and returns the new | |
742 | string. | |
743 | * `str`: calls `pr_str` on each argument with `print_readably` set | |
744 | to false, concatenates the results together ("" separator), and | |
745 | returns the new string. | |
746 | * `prn`: calls `pr_str` on each argument with `print_readably` set | |
747 | to true, joins the results with " ", prints the string to the | |
748 | screen and then returns `nil`. | |
749 | * `println`: calls `pr_str` on each argument with `print_readably` set | |
750 | to false, joins the results with " ", prints the string to the | |
751 | screen and then returns `nil`. | |
752 | ||
753 | ||
754 | <a name="step5"></a> | |
755 | ||
756 | ### Step 5: Tail call optimization | |
757 | ||
758 | ![step5_tco architecture](step5_tco.png) | |
759 | ||
760 | In step 4 you added special forms `do`, `if` and `fn*` and you defined | |
761 | some core functions. In this step you will add a Lisp feature called | |
762 | tail call optimization (TCO). Also called "tail recursion" or | |
763 | sometimes just "tail calls". | |
764 | ||
765 | Several of the special forms that you have defined in `EVAL` end up | |
766 | calling back into `EVAL`. For those forms that call `EVAL` as the last | |
767 | thing that they do before returning (tail call) you will just loop back | |
768 | to the beginning of eval rather than calling it again. The advantage | |
769 | of this approach is that it avoids adding more frames to the call | |
770 | stack. This is especially important in Lisp languages because they do | |
771 | not tend to have iteration control structures preferring recursion | |
772 | instead. However, with tail call optimization, recursion can be made | |
773 | as stack efficient as iteration. | |
774 | ||
775 | Compare the pseudocode for step 4 and step 5 to get a basic idea of | |
776 | the changes that will be made during this step: | |
777 | ``` | |
778 | diff -urp ../process/step4_if_fn_do.txt ../process/step5_tco.txt | |
779 | ``` | |
780 | ||
781 | * Copy `step4_env.qx` to `step5_tco.qx`. | |
782 | ||
783 | * Add a loop (e.g. while true) around all code in `EVAL`. | |
784 | ||
785 | * Modify each of the following form cases to add tail call recursion | |
786 | support: | |
787 | * `let*`: remove the final `EVAL` call on the second `ast` argument | |
788 | (third list element). Set `env` (i.e. the local variable passed in | |
789 | as second parameter of `EVAL`) to the new let environment. Set | |
790 | `ast` (i.e. the local variable passed in as first parameter of | |
791 | `EVAL`) to be the second `ast` argument. Continue at the beginning | |
792 | of the loop (no return). | |
793 | * `do`: change the `eval_ast` call to evaluate all the parameters | |
4e7296f9 JM |
794 | except for the last (2nd list element up to but not including |
795 | last). Set `ast` to the last element of `ast`. Continue | |
0f4ca9d1 JM |
796 | at the beginning of the loop (`env` stays unchanged). |
797 | * `if`: the condition continues to be evaluated, however, rather | |
798 | than evaluating the true or false branch, `ast` is set to the | |
799 | unevaluated value of the chosen branch. Continue at the beginning | |
800 | of the loop (`env` is unchanged). | |
801 | ||
802 | * The return value from the `fn*` special form will now become an | |
803 | object/structure with attributes that allow the default invoke case | |
804 | of `EVAL` to do TCO on mal functions. Those attributes are: | |
0f4ca9d1 JM |
805 | * `ast`: the second `ast` argument (third list element) representing |
806 | the body of the function. | |
807 | * `params`: the first `ast` argument (second list element) | |
808 | representing the parameter names of the function. | |
809 | * `env`: the current value of the `env` parameter of `EVAL`. | |
4e7296f9 JM |
810 | * `fn`: the original function value (i.e. what was return by `fn*` |
811 | in step 4). Note that this is deferrable until step 9 when it is | |
812 | needed for the `map` and `apply` core functions). | |
0f4ca9d1 JM |
813 | |
814 | * The default "apply"/invoke case of `EVAL` must now be changed to | |
815 | account for the new object/structure returned by the `fn*` form. | |
816 | Continue to call `eval_ast` on `ast`. The first element is `f`. | |
817 | Switch on the type of `f`: | |
818 | * regular function (not one defined by `fn*`): apply/invoke it as | |
4e7296f9 | 819 | before (in step 4). |
0f4ca9d1 JM |
820 | * a `fn*` value: set `ast` to the `ast` attribute of `f`. Generate |
821 | a new environment using the `env` and `params` attributes of `f` | |
822 | as the `outer` and `binds` arguments and rest `ast` arguments | |
823 | (list elements 2 through the end) as the `exprs` argument. Set | |
824 | `env` to the new environment. Continue at the beginning of the loop. | |
825 | ||
826 | Run some manual tests from previous steps to make sure you have not | |
827 | broken anything by adding TCO. | |
828 | ||
829 | Now go to the top level, run the step 5 tests. | |
830 | ||
831 | ``` | |
832 | make test^quux^step5 | |
833 | ``` | |
834 | ||
835 | Look at the step 5 test file `tests/step5_tco.mal`. The `sum-to` | |
836 | function cannot be tail call optimized because it does something after | |
837 | the recursive call (`sum-to` calls itself and then does the addition). | |
838 | Lispers say that the `sum-to` is not in tail position. The `sum2` | |
839 | function however, calls itself from tail position. In other words, the | |
840 | recursive call to `sum2` is the last action that `sum2` does. Calling | |
841 | `sum-to` with a large value will cause a stack overflow exception in | |
842 | most target languages (some have super-special tricks they use to | |
843 | avoid stack overflows). | |
844 | ||
845 | Congratulations, your mal implementation already has a feature (TCO) | |
846 | that most mainstream languages lack. | |
847 | ||
848 | ||
849 | <a name="step6"></a> | |
850 | ||
851 | ### Step 6: Files and Evil | |
852 | ||
853 | ![step6_file architecture](step6_file.png) | |
854 | ||
855 | In step 5 you added tail call optimization. In this step you will add | |
856 | some string and file operations and give your implementation a touch | |
857 | of evil ... er, eval. And as long as your language supports function | |
858 | closures, this step will be quite simple. However, to complete this | |
859 | step, you must implement string type support, so if you have been | |
860 | holding off on that you will need to go back and do so. | |
861 | ||
862 | Compare the pseudocode for step 5 and step 6 to get a basic idea of | |
863 | the changes that will be made during this step: | |
864 | ``` | |
865 | diff -urp ../process/step5_tco.txt ../process/step6_file.txt | |
866 | ``` | |
867 | ||
ffd31966 JM |
868 | * Copy `step5_tco.qx` to `step6_file.qx`. |
869 | ||
0f4ca9d1 JM |
870 | * Add two new string functions to the core namespaces: |
871 | * `read-string`: this function just exposes the `read_str` function | |
872 | from the reader. If your mal string type is not the same as your | |
873 | target language (e.g. statically typed language) then your | |
874 | `read-string` function will need to unbox (extract) the raw string | |
875 | from the mal string type in order to call `read_str`. | |
876 | * `slurp`: this function takes a file name (string) and returns the | |
877 | contents of the file as a string. Once again, if your mal string | |
878 | type wraps a raw target language string, then you will need to | |
879 | unmarshall (extract) the string parameter to get the raw file name | |
880 | string and marshall (wrap) the result back to a mal string type. | |
881 | ||
6fef8e58 JM |
882 | * In your main program, add a new symbol "eval" to your REPL |
883 | environment. The value of this new entry is a function that takes | |
884 | a single argument `ast`. The closure calls the your `EVAL` function | |
885 | using the `ast` as the first argument and the REPL environment | |
886 | (closed over from outside) as the second argument. The result of | |
887 | the `EVAL` call is returned. This simple but powerful addition | |
888 | allows your program to treat mal data as a mal program. For example, | |
889 | you can now to this: | |
890 | ``` | |
891 | (def! mal-prog (list + 1 2)) | |
892 | (eval mal-prog) | |
893 | ``` | |
0f4ca9d1 JM |
894 | |
895 | * Define a `load-file` function using mal itself. In your main | |
896 | program call the `rep` function with this string: | |
897 | "(def! load-file (fn* (f) (eval (read-string (str \"(do \" (slurp f) \")\")))))". | |
898 | ||
899 | Try out `load-file`: | |
900 | * `(load-file "../tests/incA.mal")` -> `9` | |
901 | * `(inc4 3)` -> `7` | |
902 | ||
903 | The `load-file` function does the following: | |
904 | * Call `slurp` to read in a file by name. Surround the contents with | |
905 | "(do ...)" so that the whole file will be treated as a single | |
906 | program AST (abstract syntax tree). | |
907 | * Call `read-string` on the string returned from `slurp`. This uses | |
908 | the reader to read/convert the file contents into mal data/AST. | |
909 | * Call `eval` (the one in the REPL environment) on the AST returned | |
910 | from `read-string` to "run" it. | |
911 | ||
912 | Now go to the top level, run the step 6 tests. The optional tests will | |
913 | need support from the reader for comments, vectors and hash-maps: | |
914 | ``` | |
915 | make test^quux^step6 | |
916 | ``` | |
917 | ||
918 | Congratulations, you now have a full-fledged scripting language that | |
6fef8e58 JM |
919 | can run other mal programs. The `slurp` function loads a file as |
920 | a string, the `read-string` function calls the mal reader to turn that | |
921 | stirng into data, and the `eval` function takes data and evaluates it | |
922 | as a normal mal program. However, it is important to note that the | |
923 | `eval` function is not just for running external programs. Because mal | |
924 | programs are regular mal data structures, you can dynamically generate | |
925 | or manipulate those data structures before calling `eval` on them. | |
926 | This isomorphisism (same shape) between data and programs is known as | |
927 | "homoiconicity". Lisp languages are homoiconic and this property | |
928 | distinguishes them from most other programming languages. | |
929 | ||
930 | You mal implementation is quite powerful already but the set of | |
931 | functions that are available (from `core.qx`) is fairly limited. The | |
932 | bulk of the functions you will add are described in step 9 and step A, | |
933 | but you will begin to flesh them out over the next few steps to | |
934 | support quoting (step 7) and macros (step 8). | |
0f4ca9d1 JM |
935 | |
936 | ||
45a8b3ca | 937 | #### Deferrable: |
0f4ca9d1 JM |
938 | |
939 | * Add the ability to run another mal program from the command line. | |
940 | Prior to the REPL loop, check if your mal implementation is called | |
941 | with command line arguments. If so, treat the first argument as | |
942 | a filename and use `rep` to call `load-file` on that filename, and | |
943 | finally exit/terminate execution. | |
944 | ||
945 | * Add the rest of the command line arguments to your REPL environment | |
946 | so that programs that are run with `load-file` have access to their | |
6fef8e58 | 947 | calling environmnet. Add a new "\*ARGV\*" (symbol) entry to your REPL |
0f4ca9d1 JM |
948 | environment. The value of this entry should be the rest of the |
949 | command line arguments as a mal list value. | |
950 | ||
951 | ||
952 | <a name="step7"></a> | |
953 | ||
954 | ### Step 7: Quoting | |
955 | ||
956 | ![step7_quote architecture](step7_quote.png) | |
957 | ||
958 | In step 7 you will add the special forms `quote` and `quasiquote` and | |
959 | add supporting core functions `cons` and `concat`. The two quote forms | |
960 | add a powerful abstraction for manipulating mal code itself | |
961 | (meta-programming). | |
962 | ||
0f4ca9d1 JM |
963 | The `quote` special form indicates to the evaluator (`EVAL`) that the |
964 | parameter should not be evaluated (yet). At first glance, this might | |
965 | not seem particular useful but an example of what this enables is the | |
966 | ability for a mal program to refer to a symbol itself rather than the | |
967 | value that it evaluates to. Likewise with lists. For example, consider | |
968 | the following: | |
969 | ||
970 | * `(prn abc)`: this will lookup the symbol `abc` in the current | |
971 | evaluation environment and print it. This will result in error if | |
972 | `abc` is not defined. | |
973 | * `(prn (quote abc))`: this will print "abc" (prints the symbol | |
974 | itself). This will work regardless of whether `abc` is defined in | |
975 | the current environment. | |
976 | * `(prn (1 2 3))`: this will result in an error because `1` is not | |
977 | a function and cannot be applied to the arguments `(2 3)`. | |
978 | * `(prn (quote (1 2 3)))`: this will print "(1 2 3)". | |
979 | * `(def! l (quote (1 2 3)))`: list quoting allows us to define lists | |
980 | directly in the code (list literal). Another way of doing this is | |
981 | with the list function: `(def! l (list 1 2 3))`. | |
982 | ||
983 | The second special quoting form is `quasiquote`. This allows a quoted | |
984 | list to have internal elements of the list that are temporarily | |
985 | unquoted (normal evaluation). There are two special forms that only | |
986 | mean something within a quasiquoted list: `unquote` and | |
987 | `splice-unquote`. These are perhaps best explained with some examples: | |
988 | ||
989 | * `(def! lst (quote (2 3)))` -> `(2 3)` | |
990 | * `(quasiquote (1 (unquote lst)))` -> `(1 (2 3))` | |
991 | * `(quasiquote (1 (splice-unquote lst)))` -> `(1 2 3)` | |
992 | ||
993 | The `unquote` form turns evaluation back on for its argument and the | |
994 | result of evaluation is put in place into the quasiquoted list. The | |
995 | `splice-unquote` also turns evaluation back on for its argument, but | |
996 | the evaluated value must be a list which is then "spliced" into the | |
997 | quasiquoted list. The true power of the quasiquote form will be | |
998 | manifest when it used together with macros (in the next step). | |
999 | ||
ffd31966 JM |
1000 | Compare the pseudocode for step 6 and step 7 to get a basic idea of |
1001 | the changes that will be made during this step: | |
1002 | ``` | |
1003 | diff -urp ../process/step6_file.txt ../process/step7_quote.txt | |
1004 | ``` | |
1005 | ||
1006 | * Copy `step6_file.qx` to `step7_quote.qx`. | |
1007 | ||
0f4ca9d1 JM |
1008 | * Before implementing the quoting forms, you will need to implement |
1009 | * some supporting functions in the core namespace: | |
1010 | * `cons`: this function takes a list as its second | |
1011 | parameter and returns a new list that has the first argument | |
1012 | prepended to it. | |
1013 | * `concat`: this functions takes 0 or more lists as | |
1014 | parameters and returns a new list that is a concatenation of all | |
1015 | the list parameters. | |
1016 | ||
1017 | An aside on immutability: note that neither cons or concat mutate | |
1018 | their original list arguments. Any references to them (i.e. other | |
1019 | lists that they may be "contained" in) will still refer to the | |
1020 | original unchanged value. Mal, like Clojure, is a language which uses | |
1021 | immutable data structures. I encourage you to read about the power and | |
1022 | importance of immutability as implemented in Clojure (from which | |
1023 | Mal borrows most of its syntax and feature-set). | |
1024 | ||
1025 | * Add the `quote` special form. This form just returns its argument | |
1026 | (the second list element of `ast`). | |
1027 | ||
1028 | * Add the `quasiquote` special form. First implement a helper function | |
1029 | `is_pair` that returns true if the parameter is a non-empty list. | |
1030 | Then define a `quasiquote` function. This is called from `EVAL` with | |
1031 | the first `ast` argument (second list element) and then `ast` is set | |
1032 | to the result and execution continues at the top of the loop (TCO). | |
1033 | The `quasiquote` function takes a parameter `ast` and has the | |
1034 | following conditional: | |
1035 | 1. if `is_pair` of `ast` is false: return a new list containing: | |
1036 | a symbol named "quote" and `ast`. | |
1037 | 2. else if the first element of `ast` is a symbol named "unquote": | |
1038 | return the second element of `ast`. | |
1039 | 3. if `is_pair` of first element of `ast` is true and the first | |
1040 | element of first element of `ast` (`ast[0][0]`) is a symbol named | |
1041 | "splice-unquote": return a new list containing: a symbol named | |
1042 | "concat", the second element of first element of `ast` | |
1043 | (`ast[0][1]`), and the result of calling `quasiquote` with the | |
1044 | second through last element of `ast`. | |
1045 | 4. otherwise: return a new list containing: a symbol named "cons", the | |
1046 | result of calling `quasiquote` on first element of `ast` | |
1047 | (`ast[0]`), and result of calling `quasiquote` with the second | |
1048 | through last element of `ast`. | |
1049 | ||
1050 | ||
1051 | Now go to the top level, run the step 7 tests: | |
1052 | ``` | |
1053 | make test^quux^step7 | |
1054 | ``` | |
1055 | ||
1056 | Quoting is one of the more mundane functions available in mal, but do | |
1057 | not let that discourage you. Your mal implementation is almost | |
1058 | complete, and quoting sets the stage for the next very exiting step: | |
1059 | macros. | |
1060 | ||
1061 | ||
45a8b3ca | 1062 | #### Deferrable |
0f4ca9d1 JM |
1063 | |
1064 | * The full names for the quoting forms are fairly verbose. Most Lisp | |
1065 | languages have a short-hand syntax and Mal is no exception. These | |
1066 | short-hand syntaxes are known as reader macros because they allow us | |
1067 | to manipulate mal code during the reader phase. Macros that run | |
1068 | during the eval phase are just called "macros" and are described in | |
1069 | the next section. Expand the conditional with reader `read_form` | |
1070 | function to add the following four cases: | |
1071 | * token is "'" (single quote): return a new list that contains the | |
1072 | symbol "quote" and the result of reading the next form | |
1073 | (`read_form`). | |
1074 | * token is "`" (back-tick): return a new list that contains the | |
1075 | symbol "quasiquote" and the result of reading the next form | |
1076 | (`read_form`). | |
1077 | * token is "~" (tilde): return a new list that contains the | |
1078 | symbol "unquote" and the result of reading the next form | |
1079 | (`read_form`). | |
1080 | * token is "~@" (tilde + at sign): return a new list that contains | |
1081 | the symbol "splice-unquote" and the result of reading the next | |
1082 | form (`read_form`). | |
1083 | ||
1084 | * Add support for quoting of vectors. The `is_pair` function should | |
1085 | return true if the argument is a non-empty list or vector. `cons` | |
1086 | should also accept a vector as the second argument. The return value | |
1087 | is a list regardless. `concat` should support concatenation of | |
1088 | lists, vectors, or a mix or both. The result is always a list. | |
1089 | ||
1090 | ||
1091 | <a name="step8"></a> | |
1092 | ||
1093 | ### Step 8: Macros | |
1094 | ||
1095 | ![step8_macros architecture](step8_macros.png) | |
1096 | ||
1097 | Your mal implementation is now ready for one of the most Lispy and | |
1098 | exciting of all programming concepts: macros. In the previous step, | |
1099 | quoting enabled some simple manipulation data structures and therefore | |
1100 | manipulation of mal code (because the `eval` function from step | |
1101 | 6 turns mal data into code). In this step you will be able to mark mal | |
1102 | functions as macros which can manipulate mal code before it is | |
1103 | evaluated. In other words, macros are user-defined special forms. Or | |
1104 | to look at it another way, macros allow mal programs to redefine | |
1105 | the mal language itself. | |
1106 | ||
1107 | Compare the pseudocode for step 7 and step 8 to get a basic idea of | |
1108 | the changes that will be made during this step: | |
1109 | ``` | |
1110 | diff -urp ../process/step7_quote.txt ../process/step8_macros.txt | |
1111 | ``` | |
1112 | ||
ffd31966 JM |
1113 | * Copy `step7_quote.qx` to `step8_macros.qx`. |
1114 | ||
1115 | ||
0f4ca9d1 JM |
1116 | You might think that the infinite power of macros would require some |
1117 | sort of complex mechanism, but the implementation is actually fairly | |
1118 | simple. | |
1119 | ||
1120 | * Add a new attribute `is_macro` to mal function types. This should | |
1121 | default to false. | |
1122 | ||
1123 | * Add a new special form `defmacro!`. This is very similar to the | |
1124 | `def!` form, but before the evaluated value (mal function) is set in | |
1125 | the environment, the `is_macro` attribute should be set to true. | |
1126 | ||
1127 | * Add a `is_macro_call` function: This function takes arguments `ast` | |
1128 | and `env`. It returns true if `ast` is a list that contains a symbol | |
1129 | as the first element and that symbol refers to a function in the | |
1130 | `env` environment and that function has the `is_macro` attribute set | |
1131 | to true. Otherwise, it returns false. | |
1132 | ||
1133 | * Add a `macroexpand` function: This function takes arguments `ast` | |
1134 | and `env`. It calls `is_macro_call` with `ast` and `env` and loops | |
1135 | while that condition is true. Inside the loop, the first element of | |
1136 | the `ast` list (a symbol), is looked up in the environment to get | |
1137 | the macro function. This macro function is then called/applied with | |
1138 | the rest of the `ast` elements (2nd through the last) as arguments. | |
1139 | The return value of the macro call becomes the new value of `ast`. | |
1140 | When the loop completes because `ast` no longer represents a macro | |
1141 | call, the current value of `ast` is returned. | |
1142 | ||
1143 | * In the evaluator (`EVAL`) before the special forms switch (apply | |
1144 | section), perform macro expansion by calling the `macroexpand` | |
1145 | function with the current value of `ast` and `env`. Set `ast` to the | |
1146 | result of that call. If the new value of `ast` is no longer a list | |
1147 | after macro expansion, then return `ast`, otherwise continue with | |
1148 | the rest of the apply section (special forms switch). | |
1149 | ||
1150 | * Add a new special form condition for `macroexpand`. Call the | |
1151 | `macroexpand` function using the first `ast` argument (second list | |
1152 | element) and `env`. Return the result. This special form allows | |
1153 | a mal program to do explicit macro expansion without applying the | |
1154 | result (which can be useful for debugging macro expansion). | |
1155 | ||
1156 | Now go to the top level, run the step 8 tests: | |
1157 | ``` | |
1158 | make test^quux^step8 | |
1159 | ``` | |
1160 | ||
8a98ef9a JM |
1161 | There is a reasonably good chance that the macro tests will not pass |
1162 | the first time. Although the implementation of macros is fairly | |
1163 | simple, debugging runtime bugs with macros can be fairly tricky. If | |
1164 | you do run into subtle problems that are difficult to solve, let me | |
1165 | recommend a couple of approaches: | |
1166 | ||
1167 | * Use the macroexpand special form to eliminate one of the layers of | |
1168 | indirection (to expand but skip evaluate). This will often reveal | |
1169 | the source of the issue. | |
1170 | * Add a debug print statement to the top of your main `eval` function | |
1171 | (inside the TCO loop) to print the current value of `ast` (hint use | |
1172 | `pr_str` to get easier to debug output). Pull up the step8 | |
1173 | implementation from another language and uncomment its `eval` | |
1174 | function (yes, I give you permission to violate the rule this once). | |
1175 | Run the two side-by-side. The first difference is likely to point to | |
1176 | the bug. | |
1177 | ||
1178 | Congratulations! You now have a Lisp interpreter with a super power | |
1179 | that most non-Lisp languages can only dream of (I have it on good | |
1180 | authority that languages dream when you are not using them). If you | |
1181 | are not already familiar with Lisp macros, I suggest the following | |
1182 | excercise: write a recursive macro that handles postfixed mal code | |
1183 | (with the function as the last parameter instead of the first). Or | |
1184 | not. I have not actually done so myself, but I have heard it is an | |
1185 | interesting excercise. | |
1186 | ||
1187 | In the next step you will add try/catch style exception handling to | |
1188 | your implementation in addition to some new core functions. After | |
1189 | step9 you will be very close to having a fully self-hosting mal | |
1190 | implementation. Let us continue! | |
1191 | ||
0f4ca9d1 | 1192 | |
45a8b3ca | 1193 | ### Deferrable |
0f4ca9d1 JM |
1194 | |
1195 | * Add the following new core functions which are frequently used in | |
1196 | macro functions: | |
1197 | * `nth`: this function takes a list (or vector) and a number (index) | |
1198 | as arguments, returns the element of the list at the given index. | |
1199 | If the index is out of range, this function raises an exception. | |
1200 | * `first`: this function takes a list (or vector) as its argument | |
1201 | and return the first element. If the list (or vector) is empty or | |
1202 | is `nil` then `nil` is returned. | |
1203 | * `rest`: this function takes a list (or vector) as its argument and | |
1204 | returns a new list containing all the elements except the first. | |
1205 | ||
1206 | * In the main program, use the `rep` function to define two new | |
1207 | control structures macros. Here are the string arguments for `rep` | |
1208 | to define these macros: | |
1209 | * `cond`: "(defmacro! cond (fn* (& xs) (if (> (count xs) 0) (list 'if (first xs) (if (> (count xs) 1) (nth xs 1) (throw \"odd number of forms to cond\")) (cons 'cond (rest (rest xs)))))))" | |
1210 | * `or`: "(defmacro! or (fn* (& xs) (if (empty? xs) nil (if (= 1 (count xs)) (first xs) `(let* (or_FIXME ~(first xs)) (if or_FIXME or_FIXME (or ~@(rest xs))))))))" | |
1211 | ||
1212 | ||
ffd31966 JM |
1213 | <a name="step9"></a> |
1214 | ||
1215 | ### Step 9: Try | |
1216 | ||
1217 | ![step9_try architecture](step9_try.png) | |
1218 | ||
2345a3da JM |
1219 | In this step you will implement the final mal special form for |
1220 | error/exception handling: `try*/catch*`. You will also add several core | |
1221 | functions to you implementation. In particular, you will enhance the | |
1222 | functional programming pedigree of you implementation by adding the | |
1223 | `apply` and `map` core functions. | |
1224 | ||
ffd31966 | 1225 | Compare the pseudocode for step 8 and step 9 to get a basic idea of |
dbac60df | 1226 | the changes that will be made during this step: |
ffd31966 JM |
1227 | ``` |
1228 | diff -urp ../process/step8_macros.txt ../process/step9_try.txt | |
1229 | ``` | |
1230 | ||
1231 | * Copy `step8_macros.qx` to `step9_try.qx`. | |
1232 | ||
2345a3da JM |
1233 | * Add the `try*/catch*` special form to the EVAL function. The |
1234 | try catch form looks like this: `(try* A (catch* B C))`. The form | |
1235 | `A` is evaluated, if it throws an exception, then form `C` is | |
1236 | evaluated with a new environment that binds the symbol B to the | |
1237 | value of the exception that was thrown. | |
1238 | * If your target language has built-in try/catch style exception | |
1239 | handling then you are already 90% of the way done. Add a | |
1240 | (native language) try/catch block that calls evaluates `A` within | |
1241 | the try block and catches all exceptions. If an exception is | |
1242 | caught, then translate it to a mal type/value. For native | |
1243 | exceptions this is either the message string or a mal hash-map | |
1244 | that contains the message string and other attributes of the | |
1245 | exception. When a regular mal types/values is used as an | |
1246 | exception, you will probably need to store it within a native | |
1247 | exception type in order to be able to convey/transport it using | |
1248 | the native try/catch mechanism. Then you will extract the mal | |
1249 | type/value from the native exception. Create a new mal environment | |
1250 | that binds B to the value of the exception. Finally, evaluate `C` | |
1251 | using that new environment. | |
1252 | * If your target language does not have built-in try/catch style | |
1253 | exception handling then you have some extra work to do. One of the | |
1254 | most straightforward approaches is to create a a global error | |
1255 | variable that stores the thrown mal type/value. The complication | |
1256 | is that there are a bunch of places where you must check to see if | |
1257 | the global error state is set and return without proceeding. The | |
1258 | rule of thumb is that this check should happen at the top of your | |
1259 | EVAL function and also right after any call to EVAL (and after any | |
1260 | function call that might happen to call EVAL further down the | |
1261 | chain). Yes, it is ugly, but you were warned in the section on | |
1262 | picking a language. | |
1263 | ||
1264 | * Add the `throw` core function. | |
1265 | * If your language supports try/catch style exception handling, then | |
1266 | this function takes a mal type/value and throws/raises it as an | |
1267 | exception. In order to do this, you may need to create a custom | |
1268 | exception object that wraps a mal value/type. | |
1269 | * If your language does not support try/catch style exception | |
1270 | handling, then set the global error state to the mal type/value. | |
1271 | ||
1272 | * Add the `apply` and `map` core functions. In step 5, if you did not | |
1273 | add the original function (`fn`) to the structure returned from | |
bd62ff74 JM |
1274 | `fn*`, the you will need to do so now. |
1275 | * `apply`: takes at least two arguments. The first argument is | |
1276 | a function and the last argument is list (or vector). The | |
1277 | arguments between the function and the last arguemnt (if there are | |
1278 | any) are concatenated with the final argument to create the | |
1279 | arguments that are used to call the function. The apply | |
1280 | function allows a function to be called with arguments that are | |
1281 | contained in a list (or vector). In other words, `(apply F A B [C | |
1282 | D])` is equivalent to `(F A B C D)`. | |
1283 | * `map`: takes a function and a list (or vector) and evaluates the | |
1284 | function against every element of the list (or vector) one at | |
1285 | a time and returns the results as a list. | |
2345a3da JM |
1286 | |
1287 | * Add some type predicates core functions. In Lisp, predicates are | |
1288 | functions that return true/false (or true value/nil) and typically | |
bd62ff74 JM |
1289 | end in "?" or "p". |
1290 | * `nil?`: takes a single argument and returns true (mal true value) | |
1291 | if the argument is nil (mal nil value). | |
1292 | * `true?`: takes a single argument and returns true (mal true value) | |
1293 | if the argument is a true value (mal true value). | |
1294 | * `false?`: takes a single argument and returns true (mal true | |
1295 | value) if the argument is a false value (mal false value). | |
1296 | * `symbol?`: takes a single argument and returns true (mal true | |
1297 | value) if the argument is a symbol (mal symbol value). | |
1298 | ||
1299 | Now go to the top level, run the step 9 tests: | |
1300 | ``` | |
1301 | make test^quux^step9 | |
1302 | ``` | |
1303 | ||
1304 | Your mal implementation is now essentially a fully featured Lisp | |
1305 | interpreter. But if you stop now you will miss one of the most | |
1306 | satisfying and enlightening aspects of creating a mal implementation: | |
1307 | self-hosting. | |
2345a3da JM |
1308 | |
1309 | ### Deferrable | |
1310 | ||
e37d9b49 JM |
1311 | * Add the following new core functions: |
1312 | * `symbol`: takes a string and returns a new symbol with the string | |
1313 | as its name. | |
1314 | * `keyword`: takes a string and returns a keyword with the same name | |
1315 | (usually just be prepending the special keyword | |
1316 | unicode symbol). This function should also detect if the argument | |
1317 | is already a keyword and just return it. | |
1318 | * `keyword?`: takes a single argument and returns true (mal true | |
1319 | value) if the argument is a keyword, otherwise returns false (mal | |
1320 | false value). | |
1321 | * `vector`: takes a variable number of arguments and returns | |
1322 | a vector containing those arguments. | |
1323 | * `vector?`: takes a single argument and returns true (mal true | |
1324 | value) if the argument is a vector, otherwise returns false (mal | |
1325 | false value). | |
1326 | * `hash-map`: takes a variable but even number of arguments and | |
1327 | returns a new mal hash-map value with keys from the odd arguments | |
1328 | and values from the even arguments respectively. This is basically | |
1329 | the functional form of the `{}` reader literal syntax. | |
1330 | * `map?`: takes a single argument and returns true (mal true | |
1331 | value) if the argument is a hash-map, otherwise returns false (mal | |
1332 | false value). | |
1333 | * `assoc`: takes a hash-map as the first argument and the remaining | |
1334 | arguments are odd/even key/value pairs to "associate" (merge) into | |
1335 | the hash-map. Note that the original hash-map is unchanged | |
1336 | (remember, mal values are immutable), and a new hash-map | |
1337 | containing the old hash-maps key/values plus the merged key/value | |
1338 | arguments is returned. | |
1339 | * `dissoc`: takes a hash-map and a list of keys to remove from the | |
1340 | hash-map. Again, note that the original hash-map is unchanged and | |
1341 | a new hash-map with the keys removed is returned. Key arguments | |
1342 | that do not exist in the hash-map are ignored. | |
1343 | * `get`: takes a hash-map and a key and returns the value of looking | |
1344 | up that key in the hash-map. If the key is not found in the | |
1345 | hash-map then nil is returned. | |
1346 | * `contains?`: takes a hash-map and a key and returns true (mal true | |
1347 | value) if the key exists in the hash-map and false (mal false | |
1348 | value) otherwise. | |
1349 | * `keys`: takes a hash-map and returns a list (mal list value) of | |
1350 | all the keys in the hash-map. | |
1351 | * `vals`: takes a hash-map and returns a list (mal list value) of | |
1352 | all the values in the hash-map. | |
1353 | * `sequential?`: takes a single arguments and returns true (mal true | |
1354 | value) if it is a list or a vector, otherwise returns false (mal | |
1355 | false value). | |
ffd31966 JM |
1356 | |
1357 | ||
8569b2af | 1358 | <a name="stepA"></a> |
ffd31966 | 1359 | |
bd62ff74 | 1360 | ### Step A: Mutation, Self-hosting and Interop |
ffd31966 | 1361 | |
90f618cb | 1362 | ![stepA_mal architecture](stepA_mal.png) |
ffd31966 | 1363 | |
bd62ff74 JM |
1364 | You have reached the final step of your mal implementation. This step |
1365 | is kind of a catchall for things that did not fit into other steps. | |
1366 | But most importantly, the changes you make in this step will unlock | |
1367 | the magical power known as "self-hosting". You might have noticed | |
1368 | that one of the languages that mal is implemented in is "mal". Any mal | |
1369 | implementation that is complete enough can run the mal implementation | |
1370 | of mal. You might need to pull out your hammock and ponder this for | |
1371 | a while if you have never built a compiler or interpreter before. Look | |
1372 | at the step source files for the mal implementation of mal (it is not | |
1373 | cheating now that you have reached step A). | |
1374 | ||
1375 | If you deferred the implementation of keywords, vectors and hash-maps, | |
1376 | now is the time to go back and implement them if you want your | |
1377 | implementation to self-host. | |
1378 | ||
ffd31966 JM |
1379 | Compare the pseudocode for step 9 and step A to get a basic idea of |
1380 | the changes that will be made during this step: | |
1381 | ``` | |
dbac60df | 1382 | diff -urp ../process/step9_try.txt ../process/stepA_mal.txt |
ffd31966 JM |
1383 | ``` |
1384 | ||
90f618cb | 1385 | * Copy `step9_try.qx` to `stepA_mal.qx`. |
ffd31966 | 1386 | |
bd62ff74 JM |
1387 | * Add meta-data support to mal functions. TODO. Should be separate |
1388 | from the function macro flag. | |
1389 | ||
1390 | * Add atom data type and supporting core functions. This is the only | |
1391 | mal data type/value that is mutable. TODO | |
1392 | ||
1393 | * Add the `readline` core function. TODO | |
1394 | ||
1395 | ||
1396 | Now go to the top level, run the step 9 tests: | |
1397 | ``` | |
1398 | make test^quux^step9 | |
1399 | ``` | |
1400 | ||
1401 | Once you have passed all the non-optional step A tests, it is time to | |
1402 | try self-hosting. Run your step A implementation as normal, but use | |
1403 | the file argument mode you added in step 6 to run a each of the step | |
1404 | from the mal implementation: | |
1405 | ``` | |
1406 | ./stepA_mal.qx ../mal/step1_read_print.mal | |
1407 | ./stepA_mal.qx ../mal/step2_eval.mal | |
1408 | ... | |
1409 | ./stepA_mal.qx ../mal/step9_try.mal | |
1410 | ./stepA_mal.qx ../mal/stepA_mal.mal | |
1411 | ``` | |
1412 | ||
1413 | There is a very good change that you will encounter an error at some | |
1414 | point while trying to run the mal in mal implementation steps above. | |
1415 | Debugging failures that happen while self-hosting is MUCH more | |
1416 | difficult and mind bending. One of the best approaches I have | |
1417 | personally found is to add prn statements to the mal implemenation | |
1418 | step (not your own implementation of mal) that is causing problems. | |
1419 | ||
1420 | Another approach I have frequently used is to pull out the code from | |
1421 | the mal implementation that is causing the problem and simplify it | |
1422 | step by step until you have a simple piece of mal code that still | |
1423 | reproduces the problem. Once the reproducer is simple enough you will | |
1424 | probably know where in your own implementation that problem is likely | |
1425 | to be. Please add your simple reproducer as a test case so that future | |
1426 | implementers will fix similar issues in their code before they get to | |
1427 | self-hosting when it is much more difficult to track down and fix. | |
1428 | ||
1429 | Once you can manually run all the self-hosted steps, it is time to run | |
1430 | all the tests in self-hosted mode: | |
1431 | ``` | |
1432 | make MAL_IMPL=quux test^mal | |
1433 | ``` | |
1434 | ||
1435 | When you run into problems (which you almost certainly will), use the | |
1436 | same process described above to debug them. | |
1437 | ||
1438 | Congratulations!!! When all the tests pass, you should pause for | |
1439 | a moment and consider what you have accomplished. You have implemented | |
1440 | a Lisp interpreter that is powerful and complete enough to run a large | |
1441 | mal program which is itself an implementation of the mal language. You | |
1442 | might even be asking if you can continue the "inception" by using your | |
1443 | implementation to run a mal implementation which itself runs the mal | |
1444 | implementation. | |
1445 | ||
1446 | ### Optional | |
1447 | ||
1448 | * Add metadata support to composite data types, symbols and native | |
1449 | functions. TODO | |
1450 | * `time-ms` TODO | |
1451 | * `conj` TODO | |
ffd31966 JM |
1452 | |
1453 | ||
0f4ca9d1 JM |
1454 | ## TODO: |
1455 | ||
1456 | * simplify: "X argument (list element Y)" -> ast[Y] | |
0f4ca9d1 JM |
1457 | * list of types with metadata: list, vector, hash-map, mal functions |
1458 | * more clarity about when to peek and poke in read_list and read_form | |
1459 | * tokenizer: use first group rather than whole match (to eliminate | |
1460 | whitespace/commas) |