Commit Graph

65 Commits

Author SHA1 Message Date
Stevie Strickland
880adecea4 I think that the contract changes have been merged appropriately.
svn: r12412
2008-11-12 16:44:46 +00:00
Matthew Flatt
95bedb17fb change contracts to reflect context via #%variable-reference instead of syntax objects; add id-less #%variable-reference form; drop link to unneeded CPort objects in delayed thunks; streamline some modules that tend to show up in deep phases
svn: r12231
2008-11-04 14:24:12 +00:00
Stevie Strickland
69bcae8d7b I... think this is merged correctly. It asked me about a difference between
deleting a svn:mergeinfo entry and adding to the existing one, and... hell,
I dunno.

svn: r12094
2008-10-22 17:46:47 +00:00
Robby Findler
122f8d41dc PR 9852
svn: r12076
2008-10-20 20:08:05 +00:00
Stevie Strickland
8668c6301b Ah, well. Unless magic happens, I choose the Rays, because seriously?
I may have a love for the NL, but forget the Phillies.

Just glad LA got knocked out first so we didn't have to hear about Manny
making it and Boston not.

svn: r12050
2008-10-17 02:08:29 +00:00
Robby Findler
eb7d9be3f0 fixed the contract? predicate so that it recognizes thigns that aren't yet turned into contract structs
svn: r12043
2008-10-15 12:54:37 +00:00
Stevie Strickland
41b85b19de On the plane I tried to avoid verifying the contracts when it's not needed.
svn: r11783
2008-09-17 02:52:31 +00:00
Stevie Strickland
61399c87d7 Trunkin', I'm a goin' home
Whoa whoa baby, back where I belong,
Back home, sit down and patch my bones
And get back trunkin' on

svn: r11768
2008-09-15 22:50:55 +00:00
Robby Findler
d165c586af PR 9760
svn: r11766
2008-09-15 21:58:59 +00:00
Stevie Strickland
402deaafe5 Another error case.
svn: r11765
2008-09-15 21:52:02 +00:00
Stevie Strickland
32e0f2d318 One more slight change here.
svn: r11764
2008-09-15 21:49:13 +00:00
Stevie Strickland
e76da360e5 Rework this somewhat so you get appropriate error messages with or without
#:type.

svn: r11763
2008-09-15 21:47:24 +00:00
Stevie Strickland
deccda53ea Add duplicate checking to exports list.
svn: r11761
2008-09-15 19:53:41 +00:00
Stevie Strickland
e40c856660 Do head expansion, check to make sure exported identifiers were defined
inside the with-contract form.

svn: r11760
2008-09-15 19:41:54 +00:00
Stevie Strickland
4ac235f534 Trunk merging, taking care to integrate Robby's changes appropriately. Only
change from what he did was that I added a name arg to the verify/contract
macro (he already has coerce-contract take a name appropriately, so this
was very simple, yay!).

svn: r11737
2008-09-13 23:17:58 +00:00
Robby Findler
85e489219c a bunch more values are now converted into contracts automatically
svn: r11729
2008-09-13 22:35:09 +00:00
Stevie Strickland
2d2b667bd8 Merging the changes from trunk into here. Adding back the changes that
allow for reporting the position of the contract that failed.

svn: r11728
2008-09-13 17:05:00 +00:00
Robby Findler
06a4d0df4a got rid of define-struct/prop since scheme/base has a define-struct that does all that (and more) now
svn: r11727
2008-09-13 16:29:01 +00:00
Stevie Strickland
e5539b1766 Trunk merging, check it and see
svn: r11726
2008-09-13 16:06:35 +00:00
Stevie Strickland
0870c7ae1d Actually, move the old-style define/contract to its own file in mzlib,
and change around mzlib/contract.ss appropriately.

svn: r11715
2008-09-13 02:51:09 +00:00
Stevie Strickland
16bce22386 Slight change just because we're now printing what kind of thing triggered
the broken contract.

svn: r11714
2008-09-13 02:11:05 +00:00
Stevie Strickland
c484131597 Okay, here's the old stuff put back, will fix up the unit tests when I get
home.

svn: r11709
2008-09-12 23:51:51 +00:00
Stevie Strickland
41ee6c8ac7 Hello, McFly, hello?!
I can't believe there wasn't already a test in there that checked this kind
of thing.  Wait, of course there couldn't have been, because it was specific
to define/contract and with-contract, and _I'm_ writing those tests.  Ah,
well, fixed! :p

svn: r11707
2008-09-12 23:25:58 +00:00
Stevie Strickland
a343268083 Just to make sure the srcloc info is correct.
svn: r11706
2008-09-12 22:59:51 +00:00
Stevie Strickland
e83a72d6ae * Fix converting modules whose source is (list 'quote 'name) to "module 'name"
in contract messages
 * Differentiate between define/contract functions and normal definitions
 * Fix up contract tests

svn: r11700
2008-09-12 20:46:41 +00:00
Stevie Strickland
4e8064e8a5 Change it so that it uses strings instead of symbols here.
svn: r11697
2008-09-12 19:01:13 +00:00
Stevie Strickland
19cfe4e49e I hate this kind of error checking.
svn: r11696
2008-09-12 18:55:11 +00:00
Stevie Strickland
019c2c8c34 Handle strings as blame, now add a description of the type of contract
boundary that triggered the contract error.

svn: r11695
2008-09-12 18:49:08 +00:00
Stevie Strickland
b309ced851 Appropriately check the context.
svn: r11693
2008-09-12 18:17:40 +00:00
Stevie Strickland
3b44bca1d5 Remove the introducer, which isn't being used anyway.
svn: r11692
2008-09-12 18:13:20 +00:00
Stevie Strickland
ac576f5085 There's nothing quasi about the body, so just go with regular syntax
(and change the quote-syntax uses to just #')

svn: r11690
2008-09-12 17:09:42 +00:00
Stevie Strickland
e53d5c43ff Just indention.
svn: r11689
2008-09-12 16:59:00 +00:00
Stevie Strickland
ea206c7b3d Go back to the old style.
svn: r11688
2008-09-12 16:48:33 +00:00
Stevie Strickland
e2c770ab45 So we _can_ do this, but since provides/requires are between modules, and
thus it's the enclosing module that made the contract here, we should
blame it for the misuse, not the contract boundary in which the call was
made (though further thought should be put towards this).

svn: r11687
2008-09-12 16:39:11 +00:00
Stevie Strickland
f191636cc7 Use the same info here as the provide/contract -contract uses do.
svn: r11685
2008-09-12 16:20:38 +00:00
Stevie Strickland
a44810f4f5 More error checking, shouldn't have allowed (define/contract x number? 1 2),
for example.

svn: r11680
2008-09-12 15:19:29 +00:00
Stevie Strickland
33d3cb7cd0 Have provide/contract-transformers appropriately check to see whether we
are within a with-contract scope.  Not quite sure if this code is 100%
correct (in terms of being written apprropiately), but it does the job for now.

svn: r11676
2008-09-12 14:51:26 +00:00
Stevie Strickland
68aa941b87 Since we allow non-identifiers here now (as long as they'd be appropriate
name + argument lists a la define), this case will never fire.

svn: r11662
2008-09-11 23:25:43 +00:00
Stevie Strickland
cc6d39e217 More error reporting.
svn: r11661
2008-09-11 23:19:34 +00:00
Stevie Strickland
f4dd7e85fe We should make sure the error here at least has define/contract in it (even
if it's the wrong shape due to leaving the contract out in the call), and we
_should_ check to see if we're in a expression context.  (Can't remember what
reasoning Ryan had about not doing this, but it seems to me that we still
need to check, since this expands into defines of various sorts.)

svn: r11660
2008-09-11 23:11:38 +00:00
Stevie Strickland
e1f430df72 We should use the name of the form in generating the error.
svn: r11659
2008-09-11 22:58:25 +00:00
Stevie Strickland
249f3db1b5 Add identifier checking and contract verification where appropriate.
svn: r11658
2008-09-11 22:43:19 +00:00
Stevie Strickland
430374358e Allow unprotected identifiers to be exported from with-contract, plus
do some error checking on that list.

svn: r11651
2008-09-11 21:01:14 +00:00
Stevie Strickland
1621335290 * Make define/contract have an implicit begin
* Change with-contract's implicit begin into begin-with-definitions

svn: r11650
2008-09-11 20:38:09 +00:00
Stevie Strickland
eb676359c9 There might be a simpler way of writing this, but my kung macro isn't yet
up to par if so.

svn: r11649
2008-09-11 17:49:29 +00:00
Stevie Strickland
16399b7827 Add in nicer version for function definitions.
svn: r11648
2008-09-11 17:36:43 +00:00
Stevie Strickland
c1fa0fe0ed Now have define/contract use the new with-contract form.
svn: r11647
2008-09-11 17:01:01 +00:00
Stevie Strickland
e5ed38fedc This wasn't used in define/contract either.
svn: r11646
2008-09-11 16:52:08 +00:00
Stevie Strickland
cb07ceefff Thanks to Sam and Carl, finally got this working like I wanted it.
svn: r11645
2008-09-11 16:51:36 +00:00
Stevie Strickland
8cbc41042e Pull out the querying of current-contract-region (even though it doesn't
work), as we shouldn't be forming the if clause (here a cond) in the syntax
(as it should be evaluated at expansion time).

svn: r11642
2008-09-11 13:51:03 +00:00