noteed / language-glsl Goto Github PK

The module Language.GLSL.Tests should be moved to a cabal test suite: http://www.haskell.org/cabal/users-guide/developing-packages.html#test-suites .

This would eliminate the unnecessary dependency on HUnit when installing the library.

The prettyBinary function in Language.GLSL.Pretty sometimes incorrectly omits parentheses.
For instance, (/) is left-associative, so "a / b / c" parses as "(a / b) / c", i.e.,

Div (Div (Variable "a") (Variable "b")) (Variable "c")

and "a / (b / c)" parses as

Div (Variable "a") (Div (Variable "b") (Variable "c"))

However, pretty-printing this last expression yields "a / b / c", which is incorrect.

The problematic code:

prettyBinary :: Pretty a =>
  PrettyLevel -> Rational -> Rational -> String -> a -> a -> Doc
prettyBinary l p op o e1 e2 = prettyParen (p > op) $
  pPrintPrec l op e1 <+> text o <+> pPrintPrec l op e2

To fix this bug, tweak the precedence context when pretty-printing the argument expressions, depending on the operator's associative:

• Left-associative: tighten precedence on the right.
• Right-associative: tighten precedence on the left.
• Non-associative: tighten precedence on both sides.

type Assoc = (Rational -> Rational, Rational -> Rational)

assocLeft, assocRight, assocNone :: Assoc
assocLeft  = (id,bump)
assocRight = (bump,id)
assocNone  = (bump,bump)

bump :: Rational -> Rational
bump = (+ 0.5)

prettyBinary :: Pretty a =>
  PrettyLevel -> Rational -> Rational -> Assoc -> String -> a -> a -> Doc
prettyBinary l p op (lf,rf) o e1 e2 = prettyParen (p > op) $
  pPrintPrec l (lf op) e1 <+> text o <+> pPrintPrec l (rf op) e2

I submitted a pull request.

Hello, I would like to try resolving this issue #4 (or at least enough to support #extension).

I'm stuck trying to build and run the tests for this repo though. When I run cabal test I get this error

PS C:\Users\Martin\Desktop\language-glsl> cabal test
Resolving dependencies...
Warning: solver failed to find a solution:
[__0] trying: language-glsl-0.3.0 (user goal)
line flag, or user target requires opposite flag selection)
[__1] trying: language-glsl:*test
[__2] unknown package: test-framework-hunit (dependency of language-glsl
*test)
[__1] fail (backjumping, conflict set: language-glsl, test-framework-hunit,
language-glsl:test)
After searching the rest of the dependency tree exhaustively, these were the
goals I've had most trouble fulfilling: language-glsl, language-glsl:test,
test-framework-hunit
Trying configure anyway.
Configuring language-glsl-0.3.0...
cabal.exe: Encountered missing dependencies:
HUnit -any,
test-framework -any,
test-framework-hunit -any

(I get the same error for cabal configure and cabal build)

I'm new to haskell so maybe I'm missing something obvious.

GHC version: 8.4.3
cabal-install version 2.2.0.0

http://hydra.cryp.to/build/178490/nixlog/2/raw

<> is now in Prelude, so theres some name space collisions

What's the status of this project? Is it/will it continue to be maintained?

The Language.GLSL.Parser.parse function rejects the following code:

'''
layout (std140) uniform PatternBlock
{ float pattern[100];
};
void main ()
{ gl_FragColor = vec4 (1.0, 1.0, 0.0, 1.0);
}
'''

with the error:

'''
Left "GLSL" (line 2, column 17):
unexpected "1"
expecting "/", space or "]"
'''

From reading a bit in the specification (https://www.opengl.org/registry/doc/GLSLangSpec.1.50.11.pdf), I'm fairly sure that it should accept this syntax.

More precisely, it seems that the 'conditionalExpression' parser does not accept '100', which I believe it should (in spite of the name!), see *.

Looking at the implementation of 'conditionalExpression', it seems quite clear to me that Parsec's 'buildExpressionParser' function may not be the right fit, but I have not looked into why it fails yet.

I intend to investigate this a bit more in-depth shortly.
I'll send you a pull request if I figure out a valid fix.
Just thought I'd let you know in case I never get around to it.

*Note that it is not immediately obvious from the grammar in the spec that '100' should be accepted.
Here are the rewrites to get to '100' starting from conditional_expression:
conditional_expression, logical_or_expression, logical_xor_expression, logical_and_expression, inclusive_or_expression, exclusive_or_expression, and_expression, equality_expression, relational_expression, shift_expression, additive_expression, multiplicative_expression, unary_expression, postfix_expression, primary_expression, which includes INTCONSTANT.

The preprocessor is part of the GLSL spec, section 3.4 of https://www.khronos.org/registry/gles/specs/2.0/GLSL_ES_Specification_1.0.17.pdf

Is it even possible to handle #define and #ifdef properly in language-glsl?

While using WebGL in Elm, I discovered the GLSL parser doesn't seem to infer types of arrays correctly.
For instance: uniform int array[32]; is inferred as having just the type Int.
I'm not sure about Haskell, but in Elm I expect this type to be inferred as Array Int.
Is this simply not supported, or is it a bug?

I have a few questions about using this library:

What versions of GLSL does it support?
Are there any critical portions of GLSL unsupported at this point?
Do you have any tutorials or rough guidelines for using the project?

Thanks!