Alexandre Terrasa [Fri, 9 May 2014 17:28:06 +0000 (13:28 -0400)]
benches: remove alternative typings from benches
Signed-off-by: Alexandre Terrasa <alexandre@moz-code.org>
Alexandre Terrasa [Wed, 30 Apr 2014 21:36:33 +0000 (17:36 -0400)]
tests: remove pep8analysis from skip list
Signed-off-by: Alexandre Terrasa <alexandre@moz-code.org>
Alexandre Terrasa [Wed, 30 Apr 2014 20:47:11 +0000 (16:47 -0400)]
layout_builders: removed module
Achievement Unlocked: remove one the most complex (and crappy) module from the compiler +10 nitCoins
Signed-off-by: Alexandre Terrasa <alexandre@moz-code.org>
Alexandre Terrasa [Wed, 30 Apr 2014 20:40:56 +0000 (16:40 -0400)]
separate_erasure_compiler: remove usage of MClass and MProperty LayoutMPropertyLayout
Signed-off-by: Alexandre Terrasa <alexandre@moz-code.org>
Alexandre Terrasa [Wed, 30 Apr 2014 20:29:56 +0000 (16:29 -0400)]
separate_compiler: remove usage of MpropertyLayout
Signed-off-by: Alexandre Terrasa <alexandre@moz-code.org>
Alexandre Terrasa [Wed, 30 Apr 2014 20:19:50 +0000 (16:19 -0400)]
separate_compiler: remove usage of ResolutionLayout
Signed-off-by: Alexandre Terrasa <alexandre@moz-code.org>
Alexandre Terrasa [Wed, 30 Apr 2014 20:14:47 +0000 (16:14 -0400)]
separate_compiler: remove usage of TypingLayout
Signed-off-by: Alexandre Terrasa <alexandre@moz-code.org>
Alexandre Terrasa [Wed, 30 Apr 2014 20:04:56 +0000 (16:04 -0400)]
separate_erasure_compiler: migrate virtual types coloring
Signed-off-by: Alexandre Terrasa <alexandre@moz-code.org>
Alexandre Terrasa [Wed, 30 Apr 2014 20:04:44 +0000 (16:04 -0400)]
separate_erasure_compiler: migrate class coloring
Signed-off-by: Alexandre Terrasa <alexandre@moz-code.org>
Alexandre Terrasa [Wed, 30 Apr 2014 18:05:09 +0000 (14:05 -0400)]
separate_compiler: migrate resolution table coloring
Signed-off-by: Alexandre Terrasa <alexandre@moz-code.org>
Alexandre Terrasa [Wed, 30 Apr 2014 17:56:43 +0000 (13:56 -0400)]
separate_compiler: migrate mprops coloring
Signed-off-by: Alexandre Terrasa <alexandre@moz-code.org>
Alexandre Terrasa [Thu, 8 May 2014 01:42:44 +0000 (21:42 -0400)]
separate_compiler: start migration from layout_builder to coloring on type tables
Signed-off-by: Alexandre Terrasa <alexandre@moz-code.org>
Alexandre Terrasa [Thu, 8 May 2014 01:42:10 +0000 (21:42 -0400)]
coloring: introduce new module for class and property coloring in Nit execution engines
This module will replace layout_builders
Signed-off-by: Alexandre Terrasa <alexandre@moz-code.org>
Alexandre Terrasa [Thu, 8 May 2014 00:28:03 +0000 (20:28 -0400)]
poset: introduce POSet::linearize
Signed-off-by: Alexandre Terrasa <alexandre@moz-code.org>
Alexandre Terrasa [Wed, 30 Apr 2014 17:10:50 +0000 (13:10 -0400)]
nitc/nitc-e: remove typing options
Signed-off-by: Alexandre Terrasa <alexandre@moz-code.org>
Alexis Laferrière [Wed, 30 Apr 2014 18:02:38 +0000 (14:02 -0400)]
lib/gtk: improve documentation on main modules
Signed-off-by: Alexis Laferrière <alexis.laf@xymus.net>
Alexis Laferrière [Wed, 30 Apr 2014 18:04:33 +0000 (14:04 -0400)]
lib/gtk: fix missing pkgconfig declaration in v3_6 and v3_8
Signed-off-by: Alexis Laferrière <alexis.laf@xymus.net>
Alexis Laferrière [Wed, 30 Apr 2014 17:59:58 +0000 (13:59 -0400)]
lib/gtk: move gtk3_8 to gtk::v3_8
Signed-off-by: Alexis Laferrière <alexis.laf@xymus.net>
Alexis Laferrière [Wed, 30 Apr 2014 17:58:06 +0000 (13:58 -0400)]
lib/gtk: move gtk3_6 to gtk::v3_6
Signed-off-by: Alexis Laferrière <alexis.laf@xymus.net>
Alexis Laferrière [Tue, 29 Apr 2014 23:48:29 +0000 (19:48 -0400)]
lib/gtk: move default gtk module as a project
Signed-off-by: Alexis Laferrière <alexis.laf@xymus.net>
Jean Privat [Tue, 29 Apr 2014 00:51:26 +0000 (20:51 -0400)]
Merge: introduce Nith, the ligHt Nit compiler
It is just nitg with less features, thus should be faster to compile and should compile faster.
The result of the compilation should be the same however.
Independently, the series implements phases for the various nitg engines
Pull-Request: #385
Reviewed-by: Alexandre Terrasa <alexandre@moz-code.org>
Jean Privat [Mon, 28 Apr 2014 23:58:37 +0000 (19:58 -0400)]
tests: add nith
Signed-off-by: Jean Privat <jean@pryen.org>
Jean Privat [Mon, 28 Apr 2014 22:51:27 +0000 (18:51 -0400)]
tests: update nitmetric becase the name of the mainmodule has changed
Signed-off-by: Jean Privat <jean@pryen.org>
Jean Privat [Mon, 28 Apr 2014 22:45:17 +0000 (18:45 -0400)]
Merge: Interactive parser
Just to play with `parse_something`
~~~
$ test_parser -i -n
--> hello world
Got ABlockExpr
--> "hello world"
Got AStringExpr
--> for x in [1,2,3] do
... print x
... end
Got ABlockExpr
--> fun foo do
... print "hello"
... end
Got AModule
--> for x in in in p
for x in in in p
^: Syntax error: unexpected keyword 'in'.
--> :q
~~~
Pull-Request: #424
Reviewed-by: Alexis Laferrière <alexis.laf@xymus.net>
Reviewed-by: Alexandre Terrasa <alexandre@moz-code.org>
Jean Privat [Mon, 28 Apr 2014 20:33:30 +0000 (16:33 -0400)]
Merge: Customize the generated Android project
Annotations:
* `app_name("My Name")` will be used to set the project full name and the shortcut name on Android.
* `app_version(1, 5)` or `app_version(1, 5, git_revision)` sets the project `versionName` in the manifest. It does not set the `versionCode`, this will require a way to track the project.
* `java_package` specifies the package of the generated APK (the package should be unique to each applications). The plan is to also use it to specify the Java namespace of the generated code.
* `android_manifest("some lines")` adds custom lines to the manifest. And `android_manifest_application("some lines")` adds lines to the application sub section of the manifest.
I'll have to add this information as a manual somewhere. Should it be in the documentation of the `android` module or on the website?
Pull-Request: #427
Reviewed-by: Jean Privat <jean@pryen.org>
Reviewed-by: Alexandre Terrasa <alexandre@moz-code.org>
Jean Privat [Mon, 28 Apr 2014 18:07:52 +0000 (14:07 -0400)]
abs_comp: move call of `supports_libunwind` in the right place
Signed-off-by: Jean Privat <jean@pryen.org>
Jean Privat [Mon, 28 Apr 2014 18:05:58 +0000 (14:05 -0400)]
modelbuilder: name the main module like the first module.
It is the way done by the compilers, and a lot of parts use
`mainmodule.name` to name things when `-m` is used.
So, in order to not broke the compilers parts,
generalize the idea in phases.
Note: the implicit main-module have no projets associated, so things
should not use "<main>" to identify then.
Signed-off-by: Jean Privat <jean@pryen.org>
Jean Privat [Mon, 28 Apr 2014 17:49:20 +0000 (13:49 -0400)]
Merge: Fix to inkscape_tools and introduce Float::ceil|floor
Previously image generation would cut off some extremities of the source drawing.
Pull-Request: #426
Reviewed-by: Jean Privat <jean@pryen.org>
Reviewed-by: Alexandre Terrasa <alexandre@moz-code.org>
Reviewed-by: Lucas Bajolet <r4pass@hotmail.com>
Jean Privat [Mon, 28 Apr 2014 16:45:47 +0000 (12:45 -0400)]
Merge: Benchs nullables
Allow us to compare engine performances with or without isset-checks before non-nullable attribute accesses
Pull-Request: #419
Reviewed-by: Jean Privat <jean@pryen.org>
Reviewed-by: Alexis Laferrière <alexis.laf@xymus.net>
Reviewed-by: Lucas Bajolet <r4pass@hotmail.com>
Alexis Laferrière [Sun, 27 Apr 2014 23:10:16 +0000 (19:10 -0400)]
examples/mnit_simple: use vibration and android_manifest annotation
Signed-off-by: Alexis Laferrière <alexis.laf@xymus.net>
Alexis Laferrière [Fri, 25 Apr 2014 21:01:44 +0000 (17:01 -0400)]
examples/mnit_simple: use android project annotations
Signed-off-by: Alexis Laferrière <alexis.laf@xymus.net>
Alexis Laferrière [Sun, 27 Apr 2014 19:34:01 +0000 (15:34 -0400)]
android: use custom manifest lines
Signed-off-by: Alexis Laferrière <alexis.laf@xymus.net>
Alexis Laferrière [Sun, 27 Apr 2014 19:06:43 +0000 (15:06 -0400)]
android: use the metadata to customize the generated Android project
Signed-off-by: Alexis Laferrière <alexis.laf@xymus.net>
Alexis Laferrière [Mon, 28 Apr 2014 13:14:19 +0000 (09:14 -0400)]
android: intro annotation to customize Android manifest
Signed-off-by: Alexis Laferrière <alexis.laf@xymus.net>
Jean Privat [Mon, 28 Apr 2014 15:49:05 +0000 (11:49 -0400)]
nith: introduce `nith` the ligHt Nit compiler
nith is just nitg with less features so, less phases to run,
and less code to compile (for a boostrap or test perspective).
Some numbers, note: `--skip-dead-methods` is used to produce less C code.
~~~
$ time nitg nitg.nit --skip-dead-methods
user 0m8.080s
$ time nitg nith.nit --skip-dead-methods
user 0m6.920s
~~~
Signed-off-by: Jean Privat <jean@pryen.org>
Jean Privat [Mon, 28 Apr 2014 15:34:12 +0000 (11:34 -0400)]
nitg: introduce phases for Compiler classes
Move the compiler invocation in each specific module (with phases)
Move the command-lice processing to abstract_compiler.
What remain in final `nitg` modules is the imports and the selection of
the default engine.
Signed-off-by: Jean Privat <jean@pryen.org>
Jean Privat [Mon, 28 Apr 2014 15:21:02 +0000 (11:21 -0400)]
Merge: tests: better handling of tests without a sav file
In tests, do not tag tests without sav files with *skipped*.
Instead: an empty `.res` file is a *success*, otherwise a non-empty `.res` is a *failure*.
In order to enable the correct behavior for all tests, two special `.skip` files are added to catch regressions in the Nit code.
`cc.skip` skips the C compilation and is used for programs that require a specific but unavailable C library.
Tests in this file are compiled with the `--no-cc` option.
`exec.skip` skips the execution and is used for program whose execution is unsuitable or unreproducible.
Programs that rely on the environ var `NIT_TESTING` in their code may want to drop it and just add themselves to the `exec.skip`
Pull-Request: #423
Reviewed-by: Alexandre Terrasa <alexandre@moz-code.org>
Jean Privat [Mon, 28 Apr 2014 15:20:02 +0000 (11:20 -0400)]
tests: add more stuff from examples and contrib
Thanks to the cc.skip and exec.skip files, suitable tests are available
Signed-off-by: Jean Privat <jean@pryen.org>
Alexis Laferrière [Mon, 28 Apr 2014 14:43:28 +0000 (10:43 -0400)]
android: intro annotation to get project version
Signed-off-by: Alexis Laferrière <alexis.laf@xymus.net>
Alexis Laferrière [Thu, 17 Apr 2014 17:16:55 +0000 (13:16 -0400)]
android: intro annotations to gather project metadata
Signed-off-by: Alexis Laferrière <alexis.laf@xymus.net>
Jean Privat [Mon, 28 Apr 2014 13:56:34 +0000 (09:56 -0400)]
Merge: Nitgs semiglobal
The series add some optimizations on nitg-s and nitg-e engines
Some numbers with nitg
~~~
$ nitg nitg.nit
$ time nitg nitg.nit
user 0m12.424s
$ nitg --erasure --inline-coloring-numbers nitg.nit
$ time nitg nitg.nit
user 0m11.672s
$ nitg --erasure --rta --semi-global nitg.nit
$ time nitg nitg.nit
user 0m10.004s
~~~
Pull-Request: #425
Reviewed-by: Alexis Laferrière <alexis.laf@xymus.net>
Reviewed-by: Alexandre Terrasa <alexandre@moz-code.org>
Alexandre Terrasa [Wed, 26 Mar 2014 21:31:39 +0000 (17:31 -0400)]
benches: add benches about nullables things
Signed-off-by: Alexandre Terrasa <alexandre@moz-code.org>
Alexandre Terrasa [Tue, 25 Mar 2014 19:43:15 +0000 (15:43 -0400)]
nitc: add option --count-isset-checks to count all compiled isset checks on non-nullable attributes
Signed-off-by: Alexandre Terrasa <alexandre@moz-code.org>
Alexandre Terrasa [Tue, 25 Mar 2014 19:32:47 +0000 (15:32 -0400)]
nitc: renamed --no-check-initialization in --no-check-attr-isset to follow the new specification of nullable attributes
Signed-off-by: Alexandre Terrasa <alexandre@moz-code.org>
Jean Privat [Sun, 27 Apr 2014 21:21:41 +0000 (17:21 -0400)]
Merge: lib: fix broken assert in FStream::opens
The old assert didn't work since the return was always `!= null` in Nit. Now we use instead the C value of `_file`.
Pull-Request: #403
Reviewed-by: Jean Privat <jean@pryen.org>
Reviewed-by: Alexandre Terrasa <alexandre@moz-code.org>
Jean Privat [Sun, 27 Apr 2014 20:15:48 +0000 (16:15 -0400)]
Merge: Java FFI on Android
The FFI needed some modifications to work on Android:
* Android does not implement the full JNI (so we deactivate them)
* Its implementation does not expose the same signatures as OpenJDK's
* Provide our own load_jclass because we need a specific class loader (JNI's FindClass is no enough on Android)
* Using Dalvik's long corrupted data going through Java (more debugging will be needed, anyway we can use Ints to hold pointers to C objects)
In the API, intro `App::native_activity` as a shortcut to get the main android Activity.
Update mnit_simple to use Java on Android.
Pull-Request: #418
Reviewed-by: Jean Privat <jean@pryen.org>
Alexis Laferrière [Sun, 27 Apr 2014 19:38:46 +0000 (15:38 -0400)]
contrib/inkscape_tools: remove all to_i_strip_e
Signed-off-by: Alexis Laferrière <alexis.laf@xymus.net>
Alexis Laferrière [Sun, 27 Apr 2014 19:03:59 +0000 (15:03 -0400)]
niti: implement Pointer:address_is_null as "intern" for niti executing niti
Signed-off-by: Alexis Laferrière <alexis.laf@xymus.net>
Alexis Laferrière [Sun, 27 Apr 2014 17:54:21 +0000 (13:54 -0400)]
Merge branch 'master' into java-ffi-android
Alexis Laferrière [Sun, 27 Apr 2014 17:52:27 +0000 (13:52 -0400)]
lib: fix find_class_loader with latest FFI spec
Signed-off-by: Alexis Laferrière <alexis.laf@xymus.net>
Jean Privat [Sun, 27 Apr 2014 16:07:31 +0000 (12:07 -0400)]
Merge: github_merge: copy the merge message for commit when conflict
so the following commit will be pre-rendered
Pull-Request: #422
Jean Privat [Sun, 27 Apr 2014 15:56:47 +0000 (11:56 -0400)]
nitgs: add --no-colo-dead-methods to explicitely disable coloring of dead methods
Because of coloring bug #375 ; having less useless entries in tables
increase the likeness of the bug.
Signed-off-by: Jean Privat <jean@pryen.org>
Jean Privat [Wed, 16 Apr 2014 20:25:41 +0000 (16:25 -0400)]
sepcomp: do not colorize dead methods
When RTA is available, dead methods do not need to be colored.
Signed-off-by: Jean Privat <jean@pryen.org>
Jean Privat [Mon, 7 Apr 2014 18:25:44 +0000 (14:25 -0400)]
sepcomp: add --semi-global to activate all semi-global optimizations
Some numbers:
Note that `user` is the bootstrapped time, therefore we measure both the
compilation time and the execution time.
I suspect this is stupid from the point-of-view of any scientific methodology;
but, anyway, it is numbers and people like numbers.
~~~
$ nitg --separate nitg.nit
user 0m11.228s
$ nitg --separate --semi-global nitg.nit
user 0m9.736s
$ nitg --erasure nitg.nit
user 0m7.800s
$ nitg --erasure --rta nitg.nit
user 0m8.280s
$ nitg --erasure --rta --semi-global nitg.nit
user 0m7.648s
~~~
Signed-off-by: Jean Privat <jean@pryen.org>
Jean Privat [Mon, 7 Apr 2014 18:11:14 +0000 (14:11 -0400)]
sepcomp: add --skip-dead-methods
This avoid the generation and the C-compilation of dead methods
Signed-off-by: Jean Privat <jean@pryen.org>
Jean Privat [Mon, 7 Apr 2014 18:09:18 +0000 (14:09 -0400)]
nitge: learn to use RTA if --rta is given
A new option `--rta` activate RTA optimization and
enable the use of some semi-global ones (like --direct-call-monomorph).
Signed-off-by: Jean Privat <jean@pryen.org>
Jean Privat [Mon, 7 Apr 2014 18:05:56 +0000 (14:05 -0400)]
nitge: optimize compile_class_to_c for dead classes
Classes that are dead (interface and abstract) do not need the full
class structure (VFT an cie). Only fields for typing are keept.
Signed-off-by: Jean Privat <jean@pryen.org>
Jean Privat [Wed, 16 Apr 2014 17:24:14 +0000 (13:24 -0400)]
nitgs: optimize compile_class_to_c for dead methods
Signed-off-by: Jean Privat <jean@pryen.org>
Alexis Laferrière [Sun, 27 Apr 2014 12:48:29 +0000 (08:48 -0400)]
contrib/inkscape_tools: use floor and ceil to better cut images
Signed-off-by: Alexis Laferrière <alexis.laf@xymus.net>
Alexis Laferrière [Sun, 27 Apr 2014 12:46:47 +0000 (08:46 -0400)]
lib/math: intro ceil and floor
Signed-off-by: Alexis Laferrière <alexis.laf@xymus.net>
Jean Privat [Sun, 27 Apr 2014 01:40:40 +0000 (21:40 -0400)]
tests: tests -i on test_parser
Signed-off-by: Jean Privat <jean@pryen.org>
Jean Privat [Sun, 27 Apr 2014 01:34:32 +0000 (21:34 -0400)]
test_parser: add option `-i` for interactive parsing
Signed-off-by: Jean Privat <jean@pryen.org>
Jean Privat [Sun, 27 Apr 2014 01:33:45 +0000 (21:33 -0400)]
parser_util: add `interactive_parse` to have a simple AST from the user
Signed-off-by: Jean Privat <jean@pryen.org>
Jean Privat [Sun, 27 Apr 2014 01:01:52 +0000 (21:01 -0400)]
Merge: Fix some AST bugs
`transform` and `rta` have some issues that lead to inconsistant information on the AST.
This series fixes them.
Note: Most commits extracted (cherry-picked) from #386 and #367
Pull-Request: #404
Reviewed-by: Alexis Laferrière <alexis.laf@xymus.net>
Jean Privat [Sun, 27 Apr 2014 01:00:43 +0000 (21:00 -0400)]
tests: niti skips test_markdown_args1
It borderline timeouts.
Signed-off-by: Jean Privat <jean@pryen.org>
Jean Privat [Sat, 26 Apr 2014 03:06:22 +0000 (23:06 -0400)]
tests: remove empty sav files
since now, empty res and no sav is an OK test
Signed-off-by: Jean Privat <jean@pryen.org>
Jean Privat [Sat, 26 Apr 2014 02:21:31 +0000 (22:21 -0400)]
tests: a res without sav is OK if empty else is failed
This will simplify the addition of sav, that are required only if not
empty.
Moreover, a missing non-empty sav will triggers a FAIL instead of a
skip.
Signed-off-by: Jean Privat <jean@pryen.org>
Jean Privat [Sat, 26 Apr 2014 03:05:22 +0000 (23:05 -0400)]
tests: add missing sav
Signed-off-by: Jean Privat <jean@pryen.org>
Jean Privat [Sat, 26 Apr 2014 03:04:07 +0000 (23:04 -0400)]
model: fatal error, instead of abort if more that one primitive class
Signed-off-by: Jean Privat <jean@pryen.org>
Jean Privat [Sat, 26 Apr 2014 03:02:01 +0000 (23:02 -0400)]
tests: add `cc.skip` to skip the C compilation of some programs
Exemple: unavailable C libraries.
Signed-off-by: Jean Privat <jean@pryen.org>
Jean Privat [Sat, 26 Apr 2014 02:20:20 +0000 (22:20 -0400)]
tests: niti fully skips tests in exec.skip
Signed-off-by: Jean Privat <jean@pryen.org>
Jean Privat [Sat, 26 Apr 2014 02:09:35 +0000 (22:09 -0400)]
tests: add `exec.skip` to compile but not execute tests
Signed-off-by: Jean Privat <jean@pryen.org>
Jean Privat [Sat, 26 Apr 2014 03:26:35 +0000 (23:26 -0400)]
tests: update sav/error_needed_method_alt4
Signed-off-by: Jean Privat <jean@pryen.org>
Jean Privat [Sat, 26 Apr 2014 01:24:07 +0000 (21:24 -0400)]
Merge: Metrics nullables
Add metrics about nullables usage in nit_metrics.
Also fix some bugs in metrics collect
Pull-Request: #420
Reviewed-by: Jean Privat <jean@pryen.org>
Jean Privat [Sat, 26 Apr 2014 01:11:04 +0000 (21:11 -0400)]
github_merge: copy the merge message for commit when conflict
so the following commit will be pre-rendered
Signed-off-by: Jean Privat <jean@pryen.org>
Alexis Laferrière [Tue, 22 Apr 2014 16:58:18 +0000 (12:58 -0400)]
niti: add support partial for NativeFile::address_is_null
Signed-off-by: Alexis Laferrière <alexis.laf@xymus.net>
Jean Privat [Fri, 25 Apr 2014 19:20:23 +0000 (15:20 -0400)]
Merge: Src is a project
Improve nitls to be more versatile, and modulebuilder to detect projects in `src` directories
Pull-Request: #411
Reviewed-by: Alexis Laferrière <alexis.laf@xymus.net>
Alexis Laferrière [Sat, 15 Mar 2014 12:26:24 +0000 (08:26 -0400)]
examples/mnit_simple: test Android log and toasts
Signed-off-by: Alexis Laferrière <alexis.laf@xymus.net>
Alexis Laferrière [Tue, 8 Apr 2014 13:12:26 +0000 (09:12 -0400)]
lib/mnit_android: intro App::native_activity
Signed-off-by: Alexis Laferrière <alexis.laf@xymus.net>
Alexis Laferrière [Thu, 10 Apr 2014 12:52:34 +0000 (08:52 -0400)]
nitg: use load_jclass in Java FFI
Signed-off-by: Alexis Laferrière <alexis.laf@xymus.net>
Alexis Laferrière [Mon, 17 Mar 2014 04:13:57 +0000 (00:13 -0400)]
lib: intro the load_jclass method
Signed-off-by: Alexis Laferrière <alexis.laf@xymus.net>
Alexis Laferrière [Sun, 16 Mar 2014 21:22:31 +0000 (17:22 -0400)]
android: enable code in AndroidManifest
Signed-off-by: Alexis Laferrière <alexis.laf@xymus.net>
Alexis Laferrière [Sat, 15 Mar 2014 12:24:14 +0000 (08:24 -0400)]
android: support FFI with Java
Signed-off-by: Alexis Laferrière <alexis.laf@xymus.net>
Alexis Laferrière [Fri, 25 Apr 2014 19:17:55 +0000 (15:17 -0400)]
java ffi: use Java int to store C pointers
While long worked with OpenJDK on linux, it did not on Android.
Signed-off-by: Alexis Laferrière <alexis.laf@xymus.net>
Jean Privat [Wed, 16 Apr 2014 20:15:52 +0000 (16:15 -0400)]
typing: remove `raw_arguments` as a cache
Because of transform and other potential optimization using astbuilder,
the typing phase cannot stores the raw_arguments and expect it is still
valid after subsequent phases.
The simplest way is to make `raw_arguments` a method that recollect the
correct nodes.
Alternative using some kind of cache invalidation seems to complex for not
a real gain.
Signed-off-by: Jean Privat <jean@pryen.org>
Jean Privat [Fri, 25 Apr 2014 13:16:03 +0000 (09:16 -0400)]
transform: disable transform of ASuperstringExpr
Signed-off-by: Jean Privat <jean@pryen.org>
Jean Privat [Wed, 16 Apr 2014 17:28:13 +0000 (13:28 -0400)]
rta: do not try to remove things from a set in iteration
Instead, register the things to remove then remove them once the
iteration is over,
Signed-off-by: Jean Privat <jean@pryen.org>
Jean Privat [Mon, 7 Apr 2014 14:49:05 +0000 (10:49 -0400)]
rta: add_monomorphic_send mark the mproperty as live
Signed-off-by: Jean Privat <jean@pryen.org>
Jean Privat [Tue, 22 Apr 2014 13:29:02 +0000 (09:29 -0400)]
rta: guard `add_send` with a specific set, instead of overusing `live_methods`.
Eg. a monomorph method could be live, without being subject to try_send.
Signed-off-by: Jean Privat <jean@pryen.org>
Jean Privat [Wed, 26 Mar 2014 07:48:19 +0000 (03:48 -0400)]
transform: do not detach things while itering
This invalidates the underlying iterator.
Signed-off-by: Jean Privat <jean@pryen.org>
Alexandre Terrasa [Fri, 25 Apr 2014 02:31:31 +0000 (22:31 -0400)]
metrics: update tests
Signed-off-by: Alexandre Terrasa <alexandre@moz-code.org>
Jean Privat [Tue, 22 Apr 2014 20:28:23 +0000 (16:28 -0400)]
modelbuilder: a directory named `src` triggers more heuristics
A MGroup is determined by the fact that a directory is named `src`
Thus, current project that put their module in a src directory can be
correctly identified as MProject
~~~
$ nitls src/nit*.nit examples/mnit_dino/src/*.nit -t
mnit_dino (examples/mnit_dino/src)
|--dino (examples/mnit_dino/src/dino.nit)
|--dino_android (examples/mnit_dino/src/dino_android.nit)
|--dino_linux (examples/mnit_dino/src/dino_linux.nit)
|--fancy_dino (examples/mnit_dino/src/fancy_dino.nit)
|--game_logic (examples/mnit_dino/src/game_logic.nit)
|--graphism (examples/mnit_dino/src/graphism.nit)
`--splash (examples/mnit_dino/src/splash.nit)
nit (src)
|--nit (src/nit.nit)
|--nit_version (src/nit_version.nit)
|--nitdbg_client (src/nitdbg_client.nit)
|--nitdbg_server (src/nitdbg_server.nit)
|--nitdbg_websocket_server (src/nitdbg_websocket_server.nit)
|--nitdoc (src/nitdoc.nit)
|--nitg (src/nitg.nit)
|--nitlight (src/nitlight.nit)
|--nitls (src/nitls.nit)
|--nitmetrics (src/nitmetrics.nit)
|--nitunit (src/nitunit.nit)
`--nitx (src/nitx.nit)
~~~
Signed-off-by: Jean Privat <jean@pryen.org>
Jean Privat [Fri, 25 Apr 2014 02:28:10 +0000 (22:28 -0400)]
Merge: improve nitls
Improve nitls to be more versatile.
Pull-Request: #413
Reviewed-by: Alexis Laferrière <alexis.laf@xymus.net>
Jean Privat [Fri, 25 Apr 2014 02:28:04 +0000 (22:28 -0400)]
Merge: Improve nitunit
More robust, verbose, helpfull and now understand the fences `~~~`
Also fix some existing nitunit tests.
Pull-Request: #415
Reviewed-by: Alexis Laferrière <alexis.laf@xymus.net>
Jean Privat [Fri, 25 Apr 2014 02:25:47 +0000 (22:25 -0400)]
nitls: parse options before instantiating modelbuilder
Because things, eg paths, may depend on options,
Signed-off-by: Jean Privat <jean@pryen.org>
Alexandre Terrasa [Tue, 25 Mar 2014 21:28:26 +0000 (17:28 -0400)]
metrics: display sums in to_console
Signed-off-by: Alexandre Terrasa <alexandre@moz-code.org>
Alexandre Terrasa [Tue, 25 Mar 2014 21:27:41 +0000 (17:27 -0400)]
metrics: count nullable attributes
Signed-off-by: Alexandre Terrasa <alexandre@moz-code.org>
Alexandre Terrasa [Tue, 25 Mar 2014 21:26:25 +0000 (17:26 -0400)]
metrics: compute number of properties and number of attributes in mclasses_metrics
Signed-off-by: Alexandre Terrasa <alexandre@moz-code.org>
Alexandre Terrasa [Tue, 25 Mar 2014 21:25:41 +0000 (17:25 -0400)]
model_utils: add MAttribute::is_nullable
Signed-off-by: Alexandre Terrasa <alexandre@moz-code.org>
Alexandre Terrasa [Tue, 25 Mar 2014 21:25:05 +0000 (17:25 -0400)]
model_utils: add attributes count (local + inherited) in MClass
nitlanguage / nit.git / log