80a98ef29c
Couple of changes:
...
- Update readme, changelog, cargo version number in preparation for release
- Remove panicking behavior on recursive callback calls, add additional error
variant for recursive callback errors.
2017-12-04 01:47:04 -05:00
0c644e7136
more reorganization in an attempt to shrink the size of lua.rs
2017-12-04 01:04:12 -05:00
a44b6b5170
Move function and thread into their own modules, auto-formatting
2017-12-04 00:57:39 -05:00
51838f3509
Include garbage collector error type, remove unnecessary setmetatable wrapper
2017-12-04 00:35:13 -05:00
d76935e683
I *THINK* this might actually be it, is rlua 'm' safe now?
2017-12-04 00:15:20 -05:00
d0ff10b528
I believe this is all the external API changes necessary for 'm' safety
2017-12-03 23:45:00 -05:00
e80e7d4540
missing push_string calls in util
2017-12-03 23:25:03 -05:00
67e8907f19
Couple of changes in preparation for 'm' safety:
...
- auto formatting
- add gc control to ffi
- add gc_guard to util functions
- use gc_guard to make util error handling functions never trigger __gc
metamethod Lua errors even without __gc metatable wrapper
- sort of a technicality, don't call luaL_requiref outside of the Lua
constructor, as it could trigger the garbage collector when user code has had
a chance to set __gc metamethods. Changes the API to load the debug table.
2017-12-03 23:01:03 -05:00
0bd676aa81
more refactoring
2017-12-03 21:19:32 -05:00
c95f591935
remove specific protected functions in favor of generic protect_lua_call
2017-12-03 20:29:41 -05:00
37a3145ced
missed stack size fix
2017-12-03 20:15:57 -05:00
a490229f31
More refactoring towards mem error safety
2017-12-03 20:10:45 -05:00
0909ca34fc
auto formatting
2017-12-03 18:25:53 -05:00
e41c72d642
more refactoring trying to find a workable path to 'm' error safety
2017-12-03 18:15:31 -05:00
5742e3f20a
still making small structural changes
...
slowly trying to refactor things until using all the protected calls in
protected_ffi is workable
2017-12-02 19:16:57 -05:00
f51a822738
auto-formatting
2017-12-02 18:56:14 -05:00
8a6161b16f
new strategy for protected ffi calls
2017-12-02 18:37:17 -05:00
b7c80ec066
make some things private
2017-12-02 17:47:00 -05:00
2b7e89e7c6
move error / panic metatable creation
2017-12-02 17:13:46 -05:00
fa1703d3d1
split macros into their own file
2017-12-02 17:04:33 -05:00
8a7e03978b
Experimental protected versions of all used 'm' erroring functions
2017-12-02 15:41:53 -05:00
cfa2b34a7b
Bump version to 0.9.7, update changelog
2017-11-08 14:07:48 -05:00
cf013a6a43
Merge pull request #55 from Timidger/bugfix/setting_metatable_nil
...
Fixes error when setting metatable to nil
2017-11-08 00:32:38 -05:00
f9c451bd26
Add test ensuring behaviour is correct
2017-11-07 20:13:52 -08:00
8ac27877e4
Only get gc on setmetatable if metatable isn't nil
2017-11-07 20:13:30 -08:00
33eff0e132
Merge pull request #54 from Timidger/feature/debug-unsafe-lib
...
Added ability to load debug lib, Fixes #52
2017-11-04 02:57:03 -04:00
a137086064
Check that we have enough room on stack for module
2017-10-29 15:26:40 -07:00
71fdca5477
Fixed test to use Lua assert
2017-10-29 15:24:54 -07:00
316c6976e4
Added pop after loading lua library
2017-10-29 14:54:55 -07:00
26b7901ab3
Added test for loading unsafe debug library in Lua
2017-10-29 14:53:30 -07:00
2658433bd1
Added ability to load debug lib, Fixes #52
2017-10-28 20:46:24 -07:00
00a3153ae0
Bump version / changelog for 0.9.6
2017-10-26 16:53:14 -04:00
892069edd6
Correctly wrap external errors so that Debug trait and Error::cause are correct
2017-10-26 16:49:16 -04:00
8592af1b16
Bump version to 0.9.5, update changelog
2017-10-25 14:55:05 -04:00
deab88a851
Fix xpcall return values, make the tests a lot better.
2017-10-25 14:51:10 -04:00
773bf3e9ba
Fix some clippy lints, possible edge case API incompatibility around HashMap
2017-10-24 16:15:57 -04:00
4bbeeb2b2d
Update to 0.9.4, update changelog
2017-10-24 08:59:55 -04:00
f634ce1101
It turns out, luaL_ref uses 1 extra internal stack space
...
All my check_stack calls are wrong, and this blows up very easily in situations
with recursive data structures.
2017-10-24 08:53:26 -04:00
e7661a5c27
auto-formatting
2017-10-23 16:42:20 -04:00
8c30acec4e
Possibly better stack behavior
2017-10-23 15:17:46 -04:00
ba183b0119
Fix quadratic behavior in bind
2017-10-23 14:50:47 -04:00
4a7881f652
Whoops, messed up the version in the changelog.
2017-10-14 19:33:53 -04:00
401c06cfe6
Bump cargo version to 0.9.3, update changelog.
2017-10-14 19:27:24 -04:00
8a2cfeb00d
Update readme regarding plans for panic / abort safety and other clarifications
2017-10-14 18:53:05 -04:00
4b7a3403bc
Cleanups of userdata handling, particularly around callbacks
...
First, make sure that `add_methods` cannot trigger another userdata registry
insert, causing an unintended panic. Second, remove `RefCell` surrounding
userdata hashmap, as this change makes it no longer needed. Third, add a
`RefCell` around `Callback` because FnMut means that callbacks cannot recurse
into themselves, and panic appropriately when this happens. This should
eventually be turned into an error.
2017-10-14 18:26:09 -04:00
c5a4dfd7eb
more reorganization, move simple type defines to types.rs module
2017-09-30 01:27:18 -04:00
8324d9e648
move userdata tests into userdata.rs
2017-09-30 01:13:58 -04:00
7a45490aca
crudely move LightUserData/UserData/AnyUserData to their own module
2017-09-30 01:08:08 -04:00
cf764fbabe
move table tests to table.rs
2017-09-30 00:53:28 -04:00
f785180a77
Merge pull request #45 from jonas-schievink/string
...
Extend String API
2017-09-30 00:43:19 -04:00
cc502379c1
Print CallbackError in a better way
...
CallbackError now, instead of displaying the cause description, instead prints
"callback error: <traceback>". Since the cause is already in the cause chain of
the error, this avoids repeatedly printing the cause of callback errors along
the chain, and also actually prints the callback when using Display on each
error in the chain.
2017-09-26 11:36:50 -04:00
ef538d8757
Update rlua to 0.9.2, add SOME kind of changelog.
2017-09-24 19:32:26 -04:00
61dc2823a6
Merge pull request #47 from jonas-schievink/base--
...
Don't load the base library into the "base" global
2017-09-16 21:09:33 -04:00
a7a080b50a
Don't load the base library into the "base" global
...
The stock Lua interpreter doesn't do this either. AFAIK the "base" library is supposed to be loaded into "_G" only as it contains functions like `assert` and `error`.
2017-09-17 00:07:19 +02:00
4b035af0e8
Merge pull request #44 from jonas-schievink/linkfix
...
Fix links in `Variadic` docs
2017-09-15 19:43:12 -04:00
a5b6d8fc85
Add `String::as_bytes_with_nul`
...
This cannot be accomplished without using unsafe code, which justifies this addition in my opinion.
Also changes "null" to "nul" to be in sync with `std::ffi` docs. Naming is derived from `CStr::to_bytes_with_nul`, using `as_*` instead of `to_*` since this isn't doing any computation.
2017-09-15 23:33:28 +02:00
0e4414fc0b
Impl `AsRef<[u8]>` and generic `PartialEq` for `String`
...
Tests are also moved to the new string.rs file to ensure related functionality is in one place.
2017-09-15 23:10:53 +02:00
823dd39e5d
Move string and table wrappers into own files
2017-09-15 22:03:14 +02:00
14d8b4a0a2
Fix links in `Variadic` docs
2017-09-15 20:56:46 +02:00
121bac8394
Merge pull request #43 from jonas-schievink/dire-dire-docs
...
More documentation work
2017-09-14 23:27:57 -04:00
22cbed0240
More documentation work
2017-09-14 22:59:59 +02:00
d862c0f08e
Add methods to get/set metatables on Table
...
Also make Table::raw_get actually use raw get function instead of lua_gettable!
2017-09-10 19:14:51 -04:00
13dc68c36d
Rename 'lua lifetimes to actually be 'lua
2017-09-09 21:29:22 -04:00
ec088300f2
Specify `gcc` crate dependency with minimum bound on version with `Build`
2017-09-09 21:04:57 -04:00
94ee0aaf16
Merge pull request #42 from thelearnerofcode/master
...
Switched gcc::Config to gcc:Build in build.rs.
2017-09-09 20:58:07 -04:00
21bb04da36
Switched gcc::Config to gcc:Build in build.rs.
2017-09-09 15:03:35 -04:00
5bf228fbd0
Fixup grammar error in doc comments.
2017-08-03 17:07:54 -04:00
50436d3463
autoformatting
2017-08-03 02:03:28 -04:00
1b082b17bc
Don't refer to renamed prelude versions of types, refer to their normal name
2017-08-02 18:21:47 -04:00
1efef92570
I guess I need to bump the cargo version even to change the badge.
2017-08-02 17:43:44 -04:00
fffeb9701b
Badge has wrong url, I guess it doesn't expect the github portion
2017-08-02 17:42:58 -04:00
b42d613c5b
Update Cargo.toml for 0.9, add travis-ci badge
2017-08-02 17:37:29 -04:00
d5552c494a
Typo fix
2017-08-02 17:26:25 -04:00
93e8d7bc5c
Merge pull request #37 from jonas-schievink/patch-1
...
Deny warnings on Travis
2017-08-02 17:17:06 -04:00
bc3add1a8f
Deny warnings on Travis
...
Always treating warnings as errors makes development annoying ("ugh I don't want to fix this right now"), but ignoring them quickly leaves them unfixed and shines a bad light on the project when other people want build it ("hmm, this crate's a little to... yellow for my taste"). I've found that this is pretty much a perfect middle ground (even though rlua doesn't require CI to pass before merging PRs).
2017-08-02 23:13:02 +02:00
ffc9b032e7
Riiight, doctests don't even *show* warnings by default, this flag is important.
2017-08-02 17:12:34 -04:00
cac4141871
Update README for 0.9
2017-08-02 17:04:44 -04:00
b03397137b
deny(warnings) on doctests I don't think is necessary anymore?
2017-08-02 16:37:55 -04:00
7c32ace204
Whoops, was testing removing that and placed it back in the wrong place
2017-08-02 16:37:30 -04:00
2553623b65
Provide custom allocators that ensure that OOM results in an abort
...
(closes unsafety hole)
2017-08-02 16:33:48 -04:00
16f57d18e5
Merge pull request #34 from jonas-schievink/better-error
...
[WIP] Enhanced errors
2017-08-02 16:00:08 -04:00
c3c7d8c093
setmetatable should return the table
2017-08-02 15:56:16 -04:00
dd1d335cee
Handle LUA_ERRERR
2017-08-02 21:41:35 +02:00
9c34d4b99f
Fix soundness problems with rlua
...
setmetatable now wraps a __gc method in a cclosure that aborts on error, also
'debug' library is no longer provided. We could provide just the subset of the
debug library that is sound, though.
2017-08-02 14:36:54 -04:00
820c38e806
Rename Active / Dead to better reflect their behavior
2017-08-02 13:05:26 -04:00
b458f91aac
Merge pull request #35 from jonas-schievink/repl
...
Use rustyline for the REPL
2017-08-02 11:21:08 -04:00
f9f3d05804
Fix argument bugs with pcall / xpcall, add tests for it
2017-08-02 10:42:18 -04:00
95698735a2
Fix multiline inputs
2017-08-02 14:36:10 +02:00
3f20319a84
Use rustyline for the REPL
...
This makes the REPL more usable as you can now edit lines and recall
previously executed statements.
2017-08-02 14:32:45 +02:00
70f05ac068
Remove ErrorError
2017-08-02 14:01:00 +02:00
a43bfd8f63
Remove expected field from FromLuaConversionError
2017-08-02 13:56:28 +02:00
f6cefca916
Remove stray 'for<'a>' of unused 'a lifetime
2017-08-01 18:35:09 -04:00
a1626399c4
Fix compilation
2017-08-01 23:43:51 +02:00
cb61d53ba5
Fix error message and shorten code
2017-08-01 23:42:42 +02:00
ed6e7ed5e7
Don't dispose whole traceback when it's not utf-8
2017-08-01 23:38:12 +02:00
38a13c906e
Remove "lua" prefix from error messages
...
Since this is `rlua::Error`, it should be clear that it refers to
Lua-related errors. Downstream crates want to define their own error
enums, which can add a prefix like "lua error:" to disambiguate.
2017-08-01 23:23:31 +02:00
bf76e41487
Merge IncompleteStatement into SyntaxError
...
Both are a form of syntax error, this reflects that better. No
functionality is lost, incomplete inputs are moved to a bool field of
SyntaxError.
2017-08-01 23:23:31 +02:00
ed0565c176
Fix handling of `CallbackError`
...
Previously, the traceback would be printed, but not the actual error.
I've removed traceback printing completely, not sure if that's a good
idea. A `Display` impl that outputs multiple lines feels weird.
2017-08-01 23:23:31 +02:00
7e250dacce
Enhance error messages
2017-08-01 23:23:31 +02:00
64ac8b4679
Merge pull request #32 from chucklefish/tuples
...
Major API changes before 0.9
2017-08-01 14:52:46 -04:00
kyren