fixes #18134; registers formatBiggestFloat in VM (#21299)

fixes #18134; registers formatBiggestFloat in vmops

strformat supports float format in VM

compiler/vmops.nim

@@ -38,6 +38,8 @@ when defined(nimPreviewSlimSystem):
 else:
   from std/formatfloat import addFloatRoundtrip, addFloatSprintf
 
+from std/strutils import formatBiggestFloat, FloatFormatMode
+
 # There are some useful procs in vmconv.
 import vmconv, vmmarshal
 
@@ -374,6 +376,10 @@ proc registerAdditionalOps*(c: PCtx) =
     let x = a.getFloat(1)
     addFloatSprintf(p.strVal, x)
 
+  registerCallback c, "stdlib.strutils.formatBiggestFloat", proc(a: VmArgs) =
+    setResult(a, formatBiggestFloat(a.getFloat(0), FloatFormatMode(a.getInt(1)),
+                                    a.getInt(2), chr(a.getInt(3))))
+
   wrapIterator("stdlib.envvars.envPairsImplSeq"): envPairs()
 
   registerCallback c, "stdlib.marshal.toVM", proc(a: VmArgs) =

lib/pure/strutils.nim

@@ -2413,6 +2413,9 @@ func formatBiggestFloat*(f: BiggestFloat, format: FloatFormatMode = ffDefault,
     doAssert x.formatBiggestFloat() == "123.4560000000000"
     doAssert x.formatBiggestFloat(ffDecimal, 4) == "123.4560"
     doAssert x.formatBiggestFloat(ffScientific, 2) == "1.23e+02"
+  when nimvm:
+    discard "implemented in the vmops"
+  else:
     when defined(js):
       var precision = precision
       if precision == -1:

tests/stdlib/tstrformat.nim

@@ -475,8 +475,10 @@ proc main() =
 
     # Note: times.format adheres to the format protocol. Test that this
     # works:
-
-    var dt = initDateTime(01, mJan, 2000, 00, 00, 00)
+    when nimvm:
+      discard
+    else:
+      var dt = dateTime(2000, mJan, 01, 00, 00, 00)
       check &"{dt:yyyy-MM-dd}", "2000-01-01"
 
       var tm = fromUnix(0)
@@ -558,5 +560,6 @@ proc main() =
     doAssert &"""{(if true: "'" & "'" & ')' else: "")}""" == "'')"
     doAssert &"{(if true: \"\'\" & \"'\" & ')' else: \"\")}" == "'')"
     doAssert fmt"""{(if true: "'" & ')' else: "")}""" == "')"
-# xxx static: main()
+
+static: main()
 main()