$ works for ref|ptr|pointer for all targets (c,cpp,js,vm) + other bugfixes

[timotheecour]

• tests/system/tostring.nim now enabled for js target after doing the minimal modifications to support it; some failing tests indicate pre-existing nim js bugs that should be fixed
• fixes a bug for cast in VM which was causing a crash
• $ now works for ref|ptr|pointer for all targets (c,cpp,js,vm),showing the address, which is useful, eg for determining whether 2 pointers/ref objects are the same (eg inside a log)

$ for ref|ptr|pointer

for nim js it returns a unique object id, eg @123
for all other targets it returns hex address, eg 0x10d766608

the addresses should furthermore be deterministic across re-runs if you disable ASLR, eg by passing --passl:-Wl,-no_pie (see example 2)

see tests in tests/system/tostring.nim

example 1

# D20200421T000023
type
  TFoo = object
    x1: int
    x2: pointer
    x3: seq[Foo]
    x4: ptr TFoo
  Foo = ref TFoo

template fun() =
  var z = 0
  var a = Foo(x1: 1, x2: z.addr)
  a.x3 = @[nil, a, Foo()]
  a.x4 = a[].addr
  echo a[]

fun()
static: fun()

(x1: 1, x2: 0x0, x3: @[nil, 0x10afe4dc8, 0x10b03b808], x4: 0x10afe4dc8)
Hint: ... [SuccessX]
(x1: 1, x2: 0x108e80580, x3: @[nil, 0x108ed2050, 0x108ed2080], x4: 0x108ed2050)

instead of what's printed before this PR:

(x1: 1, x2: ..., x3: ..., x4: ...)
Hint: ... [SuccessX]
(x1: 1, x2: ..., x3: ..., x4: ...)

which doesn't tell you anything useful about reference semantics (eg which pointers are the same)

example 2

git clone https://github.com/timotheecour/vitanim && cd vitanim
nim c -d:case2 --passl:-Wl,-no_pie testcases/t10595.nim
testcases/t10595
testcases/t10595 # shows same addresses

note

• before PR $ doesn't compile for ref,ptr,pointer which is not a very useful behavior, in particular for debugging (or, inside an object/tuple, would show as ... which isn't particularly useful). The current behavior gives you no easy way to check which ref/ptr objects are identical. Showing the addresses allows exactly that, is safe, and also gives a hint as to whether an address is on the stack, heap, or data segment based on its range.

• refactoring and yak shaving. #13687 attempted to make $ recurse inside ref objects however I've explained here Every value should be printable with echo RFCs#203 (comment) why this is a bad idea as a default behavior

• a more sophisticated and customizable pretty-printing (std/prettyprint as I suggested in Every value should be printable with echo RFCs#203 (comment)) doesn't remove the need for $ to work on pointer-like types, since importing std/prettyprint won't work for some modules due to cyclic imports

[Araq]

But I recently rejected any further extensions to $. Also please don't conflate bugfixes with feature additions.

[stale]

This pull request has been automatically marked as stale because it has not had recent activity. If you think it is still a valid PR, please rebase it on the latest devel; otherwise it will be closed. Thank you for your contributions.

[timotheecour]

TODO: move $(ref) to a module other than system

[stale]

This pull request has been automatically marked as stale because it has not had recent activity. If you think it is still a valid PR, please rebase it on the latest devel; otherwise it will be closed. Thank you for your contributions.