Library
Module
Module type
Parameter
Class
Class type
Settings that control precision, rounding mode, exceptional behaviour, etc.
type round =
| Down
Round towards 0; truncate.
*)| Up
Round away from 0.
*)| Half_up
Round up if last significant digit is >= 5, else round down.
*)| Half_down
Round up if last significant digit is > 5, else round down.
*)| Half_even
Round up if last significant digit is > 5 or next-to-last significant digit is odd, else round down.
*)| Ceiling
Round up if last significant digit is > 0, else no change.
*)| Floor
Round down if last significant digit is > 0, else no change.
*)| Zero_five_up
Round zero or five away from 0.
*)val string_of_round : round -> string
val default : t Stdlib.ref
default
is a reference to the default thread-local context.
val make :
?prec:int ->
?round:round ->
?e_max:int ->
?e_min:int ->
?capitals:bool ->
?clamp:bool ->
unit ->
t
make ?prec ?round ?e_max ?e_min ?capitals ?clamp ()
is a new context value with the given settings and the following traps configured:
conversion_syntax
invalid_operation
div_by_zero
overflow
These may be overridden by using Signal.set (Context.traps context) id bool
, or setting a new default context.
val copy :
orig:t ->
?prec:int ->
?round:round ->
?e_max:int ->
?e_min:int ->
?capitals:bool ->
?clamp:bool ->
?traps:Signal.array ->
?flags:Signal.array ->
unit ->
t
copy ~orig ?prec ?round ?e_max ?e_min ?capitals ?clamp ?traps ?flags
is a deep copy of orig
with the given settings. The optional settings default to the same values as those in orig
.
val prec : t -> int
val e_max : t -> int
val e_min : t -> int
val capitals : t -> bool
val clamp : t -> bool
val traps : t -> Signal.array
traps t
is the set of traps for context t
. Traps may be set to true
or false
using Signal.set
; a trap set to true
raises a runtime exception when its signal is raised. One that's set to false
causes the relevant calculation to silently handle the condition and return a result value (typically, NaN
).
val flags : t -> Signal.array
flags t
is the set of flags for context t
. Flags are set to true
during a calculation in a context t
, when the calculation encounters a condition that triggers the flag. E.g., overflow.
val e_tiny : t -> int
e_tiny t
is e_min t - prec t + 1
, the minimum allowable exponent of context t
.
val e_top : t -> int
e_top t
is the maximum exponent of context t
.
val pp : Stdlib.Format.formatter -> t -> unit
pp f t
pretty-prints the context t
.