max/lean4-htt
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 2
15293 commits 1 branch 0 tags 5.3 GiB
Commit graph

264 commits

Author SHA1 Message Date
Sebastian Ullrich
e5aaf391ff fix(library/init/lean/parser/syntax): macro_scopes.flip 2018-12-19 15:04:48 +01:00
Sebastian Ullrich
5b44a6f93d feat(library/init/lean/parser/parsec): structured doc_comment syntax nodes 2018-12-19 14:42:33 +01:00
Sebastian Ullrich
0cf88598d2 feat(library/init/lean/parser/declaration): stricter grammar for universe parameters and non-optional declaration types 2018-12-19 14:41:17 +01:00
Sebastian Ullrich
4f7be93e87 feat(library/init/lean): remove support for section aliases 2018-12-18 17:04:04 +01:00
Sebastian Ullrich
0a63d39247 refactor(library/init/lean): syntax.mk_app'/mk_app ~> syntax.mk_app/mk_capp 2018-12-14 17:37:37 +01:00
Sebastian Ullrich
306da89551 feat(library/init/lean/expander): expand universe( variable)s to multiple universe( variable) commands 2018-12-07 10:31:14 +01:00
Sebastian Ullrich
e7a6746b6a refactor(library/init/lean): share error function between expander and elaborator 2018-12-06 17:03:01 +01:00
Sebastian Ullrich
9c9e642210 feat(library/init/lean/elaborator): universe operators, subtype 2018-12-06 13:23:12 +01:00
Sebastian Ullrich
4b3995fac3 refactor(library/init/lean/parser/term): factor out opt_type parser 2018-12-06 13:23:12 +01:00
Sebastian Ullrich
143ac0e58a feat(library/init/lean/expander): also normalize bracketed binders in e.g. declarations 2018-12-06 13:23:12 +01:00
Sebastian Ullrich
79b7a29d3f fix(library/init/lean/parser/syntax): syntax.to_format: reverse scope stack 2018-12-04 13:44:46 +01:00
Sebastian Ullrich
ad27da3209 perf(library/init/lean/parser/syntax): optimize syntax.flip_scopes 2018-12-04 13:16:45 +01:00
Sebastian Ullrich
c3f8e219c0 feat(library/init/lean/expander): preresolve global references (not actually used anywhere yet) 2018-11-27 18:02:51 +01:00
Sebastian Ullrich
d9dc4edd31 refactor(library/init/lean/parser/combinators): improve sep_by view 2018-11-27 17:11:14 +01:00
Sebastian Ullrich
ee9c26947c feat(library/init/lean/elaborator): structure instances 2018-11-22 17:28:19 +01:00
Sebastian Ullrich
32e3e8a336 feat(library/init/lean/elaborator): sort_app, explicit 2018-11-21 18:13:38 +01:00
Sebastian Ullrich
cc93a2eb89 feat(library/init/lean/expander): extend "simple binders" to pi, expand pi 2018-11-21 18:13:38 +01:00
Sebastian Ullrich
a4b9cc83b8 feat(library/init/lean/{expander,elaborator}): progress 2018-11-21 18:13:38 +01:00
Sebastian Ullrich
222fff8862 refactor(library/init/lean/elaborator): introduce simple lambda binder 2018-11-21 18:13:38 +01:00
Sebastian Ullrich
c8eaee74b4 feat(frontends/lean,library/init/lean/parser/combinators): add node_longest_choice! macro 2018-11-19 18:02:28 +01:00
Sebastian Ullrich
d10ff29bea fix(library/init/lean/parser/term): placeholder binder after bracketed binder 2018-11-19 13:38:45 +01:00
Sebastian Ullrich
cd287629cb feat(library/init/lean/elaborator): initial version of translating syntax trees to preterms 2018-11-19 13:38:45 +01:00
Sebastian Ullrich
9e8dfbad79 perf(library/init/lean/parser): reintroduce separate syntax.ident constructor with much simpler and more efficient encoding 2018-11-18 16:55:04 +01:00
Sebastian Ullrich
68dad5e77c chore(library/init/lean/parser/token): raw: unnecessary try 2018-11-18 16:11:41 +01:00
Leonardo de Moura
7937f1e010 chore(library/init/util): remove util.lean 
At least initially, we will not support these primitives in the new compiler.
 2018-11-15 16:27:52 -08:00
Sebastian Ullrich
600b7ceb28 fix(library/init/lean/parser/syntax): make sure macro scopes are not pushed inside of ident nodes 2018-11-14 17:23:32 +01:00
Leonardo de Moura
182812a278 chore(library/init/lean/parser/token): make sure generate code does contain partially applied reader_t.or_else and reader_t.bind 
@kha This one was crazy, the compiler created a join point for the
continuation of the match-expression. Each case of the match was
invoking the join point with a different parser. Two of the branches
were the partially applied `reader_t.bind` and `reader_t.orelse`.
This change did not improve the performance much, but it makes sure we
don't waste time trying to figure out why we have these two partial
applications in the call graph.
 2018-11-08 17:39:46 -08:00
Leonardo de Moura
6eca6bdb0e perf(library/init/lean/parser/term): avoid eta-expansion issues 
cc @kha
 2018-11-08 16:21:37 -08:00
Leonardo de Moura
1cb22b6801 perf(library/init/lean/parser/command): move common command parsers to the beginning of the list 
@kha This modification saved 150k object allocations on my machine.

BTW, the function
```
def command_parser.run (commands : list command_parser) (p : command_parser)
  : parser_t command_parser_config id syntax :=
λ cfg, (p.run cfg).run_parsec $ λ _, any_of $ commands.map (λ p, p.run cfg)
```
is also affected by the problem I described at Zulip. It is another
example where eager eta-expansion is bad. Every time we call it, we
will create approx. 20 closures and 20 cons memory cells. We have at
least 600 commands in core.lean. So, just the `map` nested there will
generate 24k memory allocations. Moreover, the problem will get worse as we add
more commands.
 2018-11-08 14:45:34 -08:00
Leonardo de Moura
ad9aa3bc19 perf(library/init/lean/parser/token): optimize symbol 
Same trick used at `str`.

@kha: I also replaced `"token " ++ sym'` with `sym'`. We were creating
thousands of strings of the form "token ...".
 2018-11-08 14:05:09 -08:00
Leonardo de Moura
0dfeb8a79c perf(library/init/lean/parser/parsec): optimize str and raw_str 
Both `str` amd `raw_str` are used with string literals. This commit
makes sure we don't need to recompute the nested term
`dlist.singleton (repr s)`. This modification saves .2 secs when
parsing `core.lean` on my MacBook.

cc @kha
 2018-11-08 11:18:21 -08:00
Sebastian Ullrich
eaeb0a40a5 chore(library/init/lean/parser): do not expose the parser cache as monad_state 2018-11-08 16:01:19 +01:00
Sebastian Ullrich
bd70dc1fc9 perf(library/init/lean/parser): move backtrackable state from parser_core_t to module_parser_m 2018-11-08 15:58:41 +01:00
Sebastian Ullrich
41f4a34d4b feat(library/init/lean/parser/token): cache hit statistics 2018-11-08 14:52:54 +01:00
Sebastian Ullrich
d77c6169f4 chore(library/init/lean/parser/combinators): node: small simplification 2018-11-08 10:43:17 +01:00
Sebastian Ullrich
7fdfdb1784 chore(library/init/lean/parser): remove unnecessary class constraints 2018-11-06 21:45:08 +01:00
Sebastian Ullrich
090c4c0ce7 feat(library/init/lean/syntax): add lazily propagated macro scopes to syntax_node 2018-11-06 16:46:50 +01:00
Sebastian Ullrich
fc4dcd19cf refactor(library/init/lean/parser): use ident.view in syntax.to_format 2018-11-05 17:57:00 +01:00
Sebastian Ullrich
8c27f0aac6 refactor(frontends/lean/elaborator,library/init/lean): finish no_kind refactoring 2018-11-04 20:24:40 +01:00
Sebastian Ullrich
93fc1fd1de fix(library/init/lean/parser/combinators): node: do not wrap error in first argument 2018-10-30 17:43:05 +01:00
Sebastian Ullrich
74bd182bc8 perf(library/init/lean/parser/combinators): make node semireducible 2018-10-30 17:43:05 +01:00
Sebastian Ullrich
0d7ea62f9a perf(frontends/lean/elaborator,library/init/lean): put out_params first to benefit from instance pre-filtering 2018-10-30 17:43:05 +01:00
Sebastian Ullrich
2e0d6649be perf(library/init/lean/parser/basic): make basic_parser_m semireducible 2018-10-30 17:43:05 +01:00
Sebastian Ullrich
188ab30a9c fix(library/init/lean/parser/declaration): precedence for attribute arguments 2018-10-30 17:43:05 +01:00
Leonardo de Moura
cc3767e6a5 refactor(library/init/data): avoid indirection at rbmap 
Now, the nodes in a `rbmap` contain the key and value, and we avoid
one level of indirection. `rbmap`s are more common than `rbtree`.
We implement `rbtree A` as `rbmap A unit`.
 2018-10-26 17:14:09 -07:00
Leonardo de Moura
96ef30e827 chore(library/init/lean/parser/parsec): remove private 
Motivation: allow us to change the attribute in other files
 2018-10-25 10:55:32 -07:00
Leonardo de Moura
370f67b27b feat(library/init/lean/parser/parsec): mark *_mk_res functions as [inline_if_reduce] 2018-10-25 10:10:54 -07:00
Sebastian Ullrich
08d8856112 perf(library/init/lean/parser/token): replace longest_match in token with custom implementation 2018-10-22 22:33:58 +02:00
Sebastian Ullrich
163f996d00 perf(library/init/lean/parser): inline some trivial functions 2018-10-22 22:23:54 +02:00
Leonardo de Moura
19be59065a perf(library/init/lean/parser/parsec): mark bind_mk_res and orelse_mk_res with @[inline] 
A few weeks ago, it was not feasible to inline `bind_mk_res` and
`orelse_mk_res` because the compilation time would increase a lot.
Since then I have improved the heuristics for deciding whether to float
`cases_on` or not.
So, I tried today to mark them with `@[inline]` again.
The corelib build time increased only 1.2 secs, but the `parser1.lean` runtime improved:
Before:
```
num. allocated objects:  18025367
num. allocated closures: 2988476
```
After:
```
num. allocated objects:  15774515
num. allocated closures: 2488695
```
I used my desktop to collect the numbers above.
 2018-10-22 11:11:21 -07:00