[WIP] [TODO] miscellaneous issues

[timotheecour]

see also: label:postponed label:"TODO items remaining" title: [WIP] title: [TODO]

doc

• symbols should have an anchor next to it so it's easy to link to it
  EDIT: partially done, tracked here: docgen doesn't create links for certain symbols #10487
• broken rendering: https://nim-lang.github.io/Nim/contributing.html ; maybe broken after RST parser started accepting partial markdown features:

# from dcstyle.rst
* Inline code should be surrounded by single tick marks (`` like `this` ``). If you would like a character to immediately follow inline code (e.g., "``int8``s are great!"), escape the following character with a backslash (``\``). The preceding is typed as ``` ``int8``\s are great!```.

renders as:

• SPEED:./koch docs runs each nim doc separately, could use a single cmd: nim doc --project fileThatImportsAll.nim + maybe a few other files if needed

CI

in koch.nim: follow up from #10183:

• time each exec command (so we know the bottlenecks)
• add a smoke test (eg smoke_test.nim) that runs as early as possible (before other tests) and whose job is to fail fast to reject PR's in a bad state (eg after a typo) and avoid wasting CI time for these; it shall import as many modules as possible (without running into known limitations) and just try to compile it (maybe via nim --errrormax:0 smoke_test.nim) ; code can be reused with megatest maybe
• with NIM_COMPILE_TO_CPP=true runs in travis/appveyor, the nim mode is probably not yet being honored everywhere (eg nim c hardcoded in places)
• on error, testament doesn't show the relevant err msg if there are many, eg: see D20190111T070023
• report which tests are disabled (in a output json file)

compiler

• update vmgen.codeListing to use compiler/asciitables.nim instead ; it'll generate nicer looking listings
• move compiler/asciitables.nim to stdlib

testament

• use assertEquals and mismatch in testament
• var compilerPrefix* = "compiler" / "nim" ## built via ./koch tests => instead, use nim from PATH, else from —nim; not from compiler" / “nim" and don't rebuild it in ./koch tests D20190205T124107
  => ./koch tests now does not rebuild nim #10572

typetraits.name

• typetraits.name behaves unexpected typetraits.name behaves unexpected #8083
• [typetraits] echo A[B] prints differently depending on unrelated code [typetraits] echo A[B] prints differently depending on unrelated code #8569
• [typetraits] type names should resolve to name of resolved type; currently, not consistent [typetraits] type names should resolve to name of resolved type; currently, not consistent #8570
• Type mismatch error with alias types are useless Type mismatch error with alias types are useless #8339
  => still not completely fixed : see test case in Expand tyAlias when printing nested types #8634 (comment)

runnableExamples

• runnableExamples could probably be run on module scope instead of inside a block; would simplify code (no need to mess with import), and allow certain things that are impossible inside a runnableExamples (eg: X has to be at module scope); I ran into this for several unrelated issues, preventing use of runnnableExamples in those cases; this would fix that; note: no performance penalty now that runnableExamples are all run at once

=> #11732

iterators

• for x in myIter(42)(): ... gives infinite loop; inconsistent with let it = iterGen(42); for x in it(): ... for x in myIter(1)(): ... gives infinite loop; inconsistent with let it = iterGen(1); for x in it(): ... #8775
• [RFC] Better iterators syntax [RFC] Better iterators syntax RFCs#1 (was: #7047)
• closure iterators need while loops and don't work with for closure iterators need while loops and don't work with for #5340
• inline iterators could potentially avoid code bloat by using goto, see [ci skip] document that multiple yield in inline iterator cause code bloat #10553 (comment)

osproc

• osproc: startProcess with poEvalCommand need more documentation and more options osproc: startProcess with poEvalCommand need more documentation and more options #8913

pathutils

• this doesn't make sense:
  os.isAbsolute(f.string) seems odd; and why unix instead of posix?
  $nimc_D/compiler/nim.nim

proc prependCurDir(f: AbsoluteFile): AbsoluteFile =
  when defined(unix):
    if os.isAbsolute(f.string): result = f
    else: result = AbsoluteFile("./" & f.string)
  else:
    result = f

std/macros

• wrong doc for strVal; probably refers to getImpl

proc strVal*(n: NimNode): string  {.magic: "NStrVal", noSideEffect.}
    ## retrieve the implementation of `symbol`. `symbol` can be a
    ## routine or a const.

• deprecate all these: newStrLitNode, newIntLitNode since we can use newLit
  for newFloatLitNode I'm not sure, see comment in code regarding int int32 int64 are all distinct types, but float and float64 are not #5875

• Error: node lacks field: X => should tell the node kind

• remove {.compileTime.} from all procs with a NimNode param since that pragma is automatically inferred by compiler in this case

defer

• settle fate of defer ; I argue we should encourage it's use; see also: result type inference for defer, to help with a very useful debugging / contracts idiom #8793 and https://forum.nim-lang.org/t/4022#25111

misc

• ./doc/nep1.rst f8e219b path is unclear : IMO should be used for a string that can represent both a file and dir, but description not clear
• duplicate changelog entry in https://nim-lang.org/blog/2018/09/26/version-0190-released.html
  Added the parameter val for the CritBitTree[int].inc proc.
  Added the parameter val for the CritBitTree[T].incl proc.
• test_async.txt gets generated and not always deleted by tests/async/tasyncfilewrite.nim
• typo: There will be an outmated process in b138e89
• use aspell as linter to find typos? (and use some logic based on accepting stuff in index as ok workds)
• support clongdouble, see Discrepancy in Documentation About 'f128 Type-Suffix #10213 (comment)
• toInt(1e19) should raise instead of returning -9223372036854775808 ; see toInt doesn't raise an exception #2764 (comment) and Floats are not range checked #7179
• is this a bug? elif defined(nimVM): OsPlatform.nimVM (instead of: when nimvm)

important issues

• [RFC] Cyclic imports and symbol dependencies [RFC] Cyclic imports and symbol dependencies RFCs#6 (was: #3961)