Recursively checks if the type contains the given basic type

Call for any operation needing GLSL double data-type support.

Call for any operation needing full GLSL integer data-type support.

generic

set new base value as an r-value

If both the id and precision are valid, the id gets tagged with the requested precision. The passed in id is always the returned id, to simplify use p

Map atom -> string.

print usage to stdout

Translate glslang precision to SPIR-V precision decorations.

Install 'p' as the (non-'this') last parameter. Non-'this' parameters are reflected in both the list of parameters and the mangled name.

Call for any operation needing GLSL float32 data-type support.

Call for any operation needing GLSL float64 data-type support.

See if any of the extensions are set to enable, require, or warn.

set new base as an l-value base

Returns the 32-bit floating point value created by calling setFromSignUnbiasedExponentAndNormalizedSignificand on a HexFloat<FloatProxy<float>>

At the end of what block do the next create*() instructions go? Also reset current last DebugScope and current source line to unknown

Link-time warning.

To initialize per-stage shared tables, with the common table already complete.

Call for any operation needing GLSL float16 data-type support.

These will show up in error messages

Low level: add binary node (no promotions or other argument modifications)

retrieve the next character and advance one character

extract integers out of attribute arguments stored in attribute aggregate

Set up the glslang traversal

Returns the raw data.

Return the lowest-level type of scalar that an homogeneous composite is made out of. Typically, this is just to find out if something is made out of i

True if this type of IO is supposed to be arrayed with extra level for per-vertex data

double is used so that unbiased_exponent can be used with the output of ldexp directly.

Map string -> atom. Return 0 if no existing string.