Feature/48 as operator #15

ppaulweber · 2017-11-18T01:03:22Z

see: casm-lang/libcasm-fe#93

* related to sealangdotorg/sea#48

* also added missing type test cases like file, port etc. * related to sealangdotorg/sea#48

* fixed incorrect test case description

* related to sealangdotorg/sea#48

emmanuel099 · 2017-11-18T12:27:17Z

test/operator/or/bit/11/zz.casm

@@ -55,8 +55,8 @@ rule foo =
    let b : Bit'16 =    0 in
    let c = a or b in
    {
-        assert( c  = asBit( 0, 16 )) 
-
+        assert( c  = 0 as Bit'16 ) 


looks much better :D

I agree 😄

emmanuel099 · 2017-11-18T12:32:55Z

test/expression/typecasting/boolean/integer.casm

+rule foo =
+{
+    let x : Integer = undef in assert( x as Boolean = undef )
+    let x : Integer = -123  in assert( x as Boolean = false )


x != 0 -> true

emmanuel099 · 2017-11-18T12:33:08Z

test/expression/typecasting/boolean/integer.casm

+{
+    let x : Integer = undef in assert( x as Boolean = undef )
+    let x : Integer = -123  in assert( x as Boolean = false )
+    let x : Integer = -1    in assert( x as Boolean = false )


x != 0 -> true

emmanuel099 · 2017-11-18T12:33:40Z

test/expression/typecasting/boolean/error/tuple.casm

+rule foo =
+{
+    let x : Tuple< Integer, Integer > = [ 1, 10 ] in
+        assert( x as Boolean = 0 ) //@ ERROR( 1700 )


emmanuel099 · 2017-11-18T12:35:22Z

test/expression/typecasting/bit/integer.casm

+    let x : Integer = -1    in assert( x as Bit = 0b0 )
+    let x : Integer = 0     in assert( x as Bit = 0b0 )
+    let x : Integer = 1     in assert( x as Bit = 0b1 )
+    let x : Integer = 42    in assert( x as Bit = 0b1 )


42 as Bit = 0b1 looks strange

I would expect that this gives a bit-field with the min. number of bits to represent this number.

In this case you have not misunderstood the Bit type, the Bit is a abbreviation for Bit'1, which means a bit-vector of size 1. The conversion from Integer to Bit'{n} is defined as follows: If the content of the Integer can fit into the Bit'n type of bit-size n then the value gets encoded properly, otherwise the conversion shall return a undef. You are definitely right that there is a mistake in the unit test ... I'll fix this ASAP!

From a user's point of view I would expect what I already mentioned above: A missing '{n} means that (if the type system is consistent) the type inference calculates the size automatically because Bit looks like a "partial" type here. Other languages handle it the same way, e.g. in Java List<> = ... where the actual list item type is inferred.

I.e. currently there are two different meanings for a single type name. Bit stands for a single bit while Bit'{n} stands for a bit vector of size n.

I see two possibilities to make it consistent:

Drop Bit in favor of Bit'1

Rename Bit'{n} to e.g. BitVector'{n} and add an alias Bit = BitVector'1

I hope you remember that I warned you about the Bit and Bit'1 problematic about a year ago, right? In this case I would say we drop the Bit in favor for the Bit'1, because I do not want to introduce a camel case type.

PR is coming soon, see issue: sealangdotorg/sea#54

emmanuel099 · 2017-11-18T12:37:01Z

test/expression/typecasting/bit/integer.casm

+
+    let x : Integer = undef in assert( x as Bit'8 = undef )
+    let x : Integer = -24   in assert( x as Bit'8 = 0b0 )
+    let x : Integer = -1    in assert( x as Bit'8 = 0b0 )


Should be 0b11111111

emmanuel099 · 2017-11-18T12:37:27Z

test/expression/typecasting/bit/integer.casm

+    let x : Integer = 42    in assert( x as Bit'1 = 0b1 )
+
+    let x : Integer = undef in assert( x as Bit'8 = undef )
+    let x : Integer = -24   in assert( x as Bit'8 = 0b0 )


Should be 0b11101000

Two's complement -> Same for all other negative numbers

emmanuel099 · 2017-11-18T12:42:35Z

test/expression/typecasting/bit/error/tuple.casm

+    let x : Tuple< Integer, Integer > = [ 1, 10 ] in
+        assert( x as Bit'8 = 0b0 ) //@ ERROR( 1700 )
+
+    let x : Tuple< Integer, Integer > = [ 1, 10 ] in


General remark (for future): 1 representative test case is enough, duplicated test cases make the whole thing harder to change/maintain ;)

Maybe ... maybe not, I think we can not be sure enough to test several cases.

emmanuel099 · 2017-11-18T12:46:14Z

test/expression/typecasting/bit/decimal.casm

+    let x : Decimal = 0.0    in assert( x as Bit'23 = 0b0 )
+    let x : Decimal = 1.1    in assert( x as Bit'23 = 0b1 )
+    let x : Decimal = 42.42  in assert( x as Bit'23 = 0b101010 )
+}


Hmm the whole Decimal to Bit conversion isn't well-defined in my opinion.

The current approach of the type casting from Decimal to Bit'{n} is that only the Integer part gets translated to a proper bit-vector representation.

emmanuel099 · 2017-11-18T12:50:21Z

test/builtin/print/print/03.casm

@@ -45,6 +45,5 @@ CASM init foo

 rule foo =
 {
-    println( undef + asBit( 0, 10 ) )
-
+    println( undef + 0 as Bit'10 )


This and the following print test cases are useless, they all tests the same -> println(undef : Bit)

undef + something is already covered by expression tests.

This should be cleaned up in another PR please ;)

I'm aware of the print test case problematic, there will be a major clean-up PR really soon, the change here is required due to the enforcement of the <expr> as <type> syntax for type castings instead of the as<type>() built-in calls

see: sealangdotorg/sea#53

* related to sealangdotorg/sea#48

* see discussion #15 * related to sealangdotorg/sea#48

ppaulweber · 2017-11-20T16:16:38Z

@emmanuel099 please review

emmanuel099

Nice work!

ppaulweber added 13 commits November 12, 2017 21:07

Moving old built-in 'as*()' to TypeCastingExpression test cases

9121a8f

* related to sealangdotorg/sea#48

TypeCastingExpression: fixed 'as Boolean' test cases

002b1df

* also added missing type test cases like file, port etc. * related to sealangdotorg/sea#48

Description

61ff543

* fixed incorrect test case description

TypeCastingExpression: updated Integer test cases

690cef2

* related to sealangdotorg/sea#48

TypeCastingExpression: updated Bit test cases

83cca57

* related to sealangdotorg/sea#48

TypeCastingExpression: updated String test cases

2a5e093

* related to sealangdotorg/sea#48

TypeCastingExpression: fixed 'as' operator error code

46d1ec8

* related to sealangdotorg/sea#48

TypeCastingExpression: added error test cases

5bebff8

* related to sealangdotorg/sea#48

TypeCastingExpression: removed 'old' casting built-in calls

9d3d905

* related to sealangdotorg/sea#48

TypeCastingExpression: updated String test cases

461c380

* related to sealangdotorg/sea#48

TypeCastingExpression: updated Decimal test cases

3a0caa9

* related to sealangdotorg/sea#48

TypeCastingExpression: updated Rational test cases

7e3bbfe

* related to sealangdotorg/sea#48

TypeCastingExpression: updated Enumeration test cases

0b4d956

* related to sealangdotorg/sea#48

ppaulweber added the needs-review label Nov 18, 2017

ppaulweber self-assigned this Nov 18, 2017

ppaulweber requested a review from emmanuel099 November 18, 2017 01:03

emmanuel099 reviewed Nov 18, 2017

View reviewed changes

ppaulweber added 3 commits November 20, 2017 15:11

TypeCastingExpression: added incorrect result type case

d8ab287

* related to sealangdotorg/sea#48

TypeCastingExpression: updated Integer as Bit unit tests

923fac8

* see discussion #15 * related to sealangdotorg/sea#48

TypeCastingExpression: fixed incorrect tuple error unit test

ade9b1b

TypeCastingExpression: fixed incorrect integer unit test

bac4162

emmanuel099 approved these changes Nov 20, 2017

View reviewed changes

emmanuel099 merged commit d0753ad into master Nov 20, 2017

emmanuel099 deleted the feature/48_as_operator branch November 20, 2017 21:20

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Feature/48 as operator #15

Feature/48 as operator #15

ppaulweber commented Nov 18, 2017

emmanuel099 Nov 18, 2017

ppaulweber Nov 20, 2017

emmanuel099 Nov 18, 2017

emmanuel099 Nov 18, 2017

emmanuel099 Nov 18, 2017

emmanuel099 Nov 18, 2017

ppaulweber Nov 20, 2017

ppaulweber Nov 20, 2017

emmanuel099 Nov 20, 2017

ppaulweber Nov 20, 2017

emmanuel099 Nov 18, 2017

ppaulweber Nov 20, 2017

emmanuel099 Nov 18, 2017

ppaulweber Nov 20, 2017

emmanuel099 Nov 18, 2017

ppaulweber Nov 20, 2017

emmanuel099 Nov 18, 2017

ppaulweber Nov 20, 2017

emmanuel099 Nov 18, 2017

ppaulweber Nov 20, 2017

ppaulweber Nov 20, 2017

ppaulweber commented Nov 20, 2017

emmanuel099 left a comment

Feature/48 as operator #15

Feature/48 as operator #15

Conversation

ppaulweber commented Nov 18, 2017

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ppaulweber commented Nov 20, 2017

emmanuel099 left a comment

Choose a reason for hiding this comment