update ci.yml and be more explicit in .nimble

[narimiran]

No description provided.

[tersec]

These --warnings:off in tests seem dodgy. I know it was already there, but probably better to remove it. If it's truly necessary for a specific warning, that warning can be more selectively disabled.

[narimiran]

Agreed!

I've pushed a commit removing that. Let's see if anything pops up.

P.S. EDIT:
No warnings produced, as far as I can tell.

[tersec]

For 2.0:

[NimScript] mkDir: build
[NimScript] exec: nim c   --outdir:build -f -c --hints:off --styleCheck:usages --styleCheck:error --skipParentCfg -d:release -d:kzgExternalBlstNoSha256 -d:kzgExternalBlst --mm:refc tests/test_all
[NimScript] exec: nim c   --outdir:build -f -c --hints:off --styleCheck:usages --styleCheck:error --skipParentCfg -d:release -d:kzgExternalBlstNoSha256 -d:kzgExternalBlst --mm:orc tests/test_all

For 1.6:

[NimScript] mkDir: build
[NimScript] exec: nim c   --outdir:build -f -c --hints:off --styleCheck:usages --styleCheck:error --skipParentCfg -d:release -d:kzgExternalBlstNoSha256 -d:kzgExternalBlst --mm:refc tests/test_all

For devel:

[NimScript] mkDir: build
[NimScript] exec: nim c   --outdir:build -f -c --hints:off --styleCheck:usages --styleCheck:error --skipParentCfg -d:release -d:kzgExternalBlstNoSha256 -d:kzgExternalBlst --mm:refc tests/test_all
[NimScript] exec: nim c   --outdir:build -f -c --hints:off --styleCheck:usages --styleCheck:error --skipParentCfg -d:release -d:kzgExternalBlstNoSha256 -d:kzgExternalBlst --mm:orc tests/test_all

Looks fine, yeah.

In some sense, this isn't actually that interesting a repo for Nim warnings/hints because there's little-to-no nontrivial algorithmic logic. It's mostly C bindings, so not surprising.

It's vaguely unfortunate that the Processing hint means that to avoid un-actionable hint spam, one needs to disable something, and that something can easily end up being all hints.

Hints:
  [ ] Success
  [ ] SuccessX
  [ ] CC
  [ ] XDeclaredButNotUsed
  [ ] DuplicateModuleImport
  [ ] XCannotRaiseY
  [ ] ConvToBaseNotNeeded
  [ ] ConvFromXtoItselfNotNeeded
  [ ] ExprAlwaysX
  [x] QuitCalled
  [x] Processing
  [ ] ProcessingStmt
  [ ] CodeBegin
  [ ] CodeEnd
  [ ] Conf
  [ ] Path
  [ ] CondTrue
  [ ] CondFalse
  [ ] Name
  [ ] Pattern
  [x] Exec
  [ ] Link
  [ ] Dependency
  [ ] Source
  [ ] Performance
  [ ] StackTrace
  [ ] GCStats
  [ ] GlobalVar
  [ ] ExpandMacro
  [ ] AmbiguousEnum
  [x] User
  [ ] UserRaw
  [ ] ExtendedContext
  [ ] MsgOrigin
  [ ] DeclaredLoc
  [ ] UnknownHint

Especially Processing and is basically spam. The rest (Conf, Link, and SuccessX, for example) are at least mostly bounded, happening a small, essentially constant number of times each build.

What does the output look like with --hint:Processing:off instead of --hints:off?

This is a good baseline repo to test this out with precisely because it's kind of a null case, not much going on to flag, to have a reference for what this might do with more involved repos.

[tersec]

e.g., for a single-line echo program, with ORC, the Processing hints:

CC: Nim/lib/system/exceptions.nim
CC: Nim/lib/std/private/since.nim
CC: Nim/lib/system/ctypes.nim
CC: Nim/lib/std/sysatomics.nim
CC: Nim/lib/system/ansi_c.nim
CC: Nim/lib/system/memory.nim
CC: Nim/lib/std/private/syslocks.nim
CC: Nim/lib/std/private/threadtypes.nim
CC: Nim/lib/std/private/digitsutils.nim
CC: Nim/lib/std/private/miscdollars.nim
CC: Nim/lib/std/assertions.nim
CC: Nim/lib/system/iterators.nim
CC: Nim/lib/system/coro_detection.nim
CC: Nim/lib/std/private/dragonbox.nim
CC: Nim/lib/std/private/schubfach.nim
CC: Nim/lib/std/formatfloat.nim
CC: Nim/lib/std/objectdollar.nim
CC: Nim/lib/system/dollars.nim
CC: Nim/lib/std/typedthreads.nim
CC: Nim/lib/system/stacktraces.nim
CC: Nim/lib/std/private/bitops_utils.nim
CC: Nim/lib/system/countbits_impl.nim
CC: Nim/lib/system/repr_v2.nim
CC: Nim/lib/std/widestrs.nim
CC: Nim/lib/std/syncio.nim
CC: Nim/lib/system.nim

refc is better at least:

CC: Nim/lib/system/exceptions.nim
CC: Nim/lib/std/private/digitsutils.nim
CC: Nim/lib/system/dollars.nim
CC: Nim/lib/std/typedthreads.nim
CC: Nim/lib/pure/collections/sharedlist.nim
CC: Nim/lib/system.nim

But the Hints ontology is just not optimal.

[narimiran]

What does the output look like with --hint:Processing:off instead of --hints:off?

It just suppresses processing dots (one line per executed command), hints about used config files are still there.

Using --hint:processing:off --hint:conf:off helps with both.

[tersec]

Using --hint:processing:off --hint:conf:off helps with both.

This seems like a reasonable balance, yeah

[narimiran]

I'd rather keep --hints:off, like it is currently. I don't see any gain in being more specific and/or keeping some of the hints present.