Re: How do i compile asm files with vcc compiler?
You mean masm? That is the Microsoft assembler that comes with VCC. Or does Nim use external assemler now?
Re: Read gzip-compressed file line by line
Since you mention DNA seqs, you might want to take a look at [nim-hts](https://github.com/brentp/hts-nim/). It's easy on unix systems to get htslib, and not too difficult on windows using msys2 You could probably write an iterator around his BGZ type.
Re: migrate oldwinapi to winim
Well . . . it's Microsoft vs the world. I'm inclined to side with Microsoft when it is their library, but then try to hide the irritating proc names in my own module. That has the benefit(?) of clearly marking what is Microsoft stuff as opposed to my own code - when I look back at the code months later.
Re: Trying to learn templates, but its hard to understand.
i had made a PR recently, unfortunately it was closed: [https://github.com/nim-lang/Nim/pull/8528](https://github.com/nim-lang/Nim/pull/8528)
Re: Trying to learn templates, but its hard to understand.
I didn't know you could do this, but apparently this works as well (tried on both 0.18.0 and 0.19.0). type MyType = distinct int template Cast[T](value: typed): T = cast[T](value) let a = Cast[int](3.MyType) echo a # prints 3 Run
Re: How do i compile asm files with vcc compiler?
Looks like a regression. We need to change how `yasm` is called, I think.
Re: migrate oldwinapi to winim
> Mostly it was changing identifiers back to their originals as published by > msdn. I've said it before but `winim` should really have used `lowerCase` for the procs like our style guide recommends. Many Nim programmers read UpperCase as type names nowadays.
Re: Trying to learn templates, but its hard to understand.
With choosenim is 1 command to switch versions, super friendly.
Re: Trying to learn templates, but its hard to understand.
> I want to install 0.19 but i am afraid that whether my current code will work > or not. 0.19 brings lots of improvements and bugfixes, my recommendation would be to use it. And if you want to have multiple versions installed, and you want to easily switch between them: [choosenim](https://github.com/dom96/choosenim/) is your friend.
migrate oldwinapi to winim
I just had great success migrating some of my code from using oldwinapi to
using the newer winm module. **Much applause for the author of winim** ! ! ! !
! ! ! !
Mostly it was changing identifiers back to their originals as published by msdn.
Also since my project does not use unicode (nor utf8) in any way, I needed to
compile using the nim option:
> -d:winascii|
> ---|---
To get the result to link edit when cross compiling from Linux to MS binary, I
changed a line (about #11) in the ---/.nimble/pkgs/winim-2.5.1/winim/inc winim
package (after you install it) from:
> const winimPrivateDir = parentDir(currentSourcePath()) & "/../lib/"
to:
> const winimPrivateDir = parentDir(currentSourcePath()) & r"..lib"
Below is a list of the changes I had to make (to **MY** sources) during the
migration. This no doubt needs to be extended by other programmer's
experiences. Please feel free to do so.
oldwinapi winim module comment
- - --- -
LPITEMIDLISTLPCITEMIDLIST inc/objbase.nim
HINST HINSTANCE inc/windef.nim
WINUINT UINTinc/windef.nim
TopLeft.x leftinc/windef.nim IN RECT
TopLeft.y top inc/windef.nim IN RECT
BottomRight.x right inc/windef.nim IN RECT
BottomRight.y bottom inc/windef.nim IN RECT
TOPENFILENAME OPENFILENAME inc/commdlg.nim
flags Flags inc/commdlg.nim in OPENFILENAME
TLOGFONTLOGFONT inc/wingdi.nim
lfItaliclfItalicinc/wingdi.nim cast to BYTE
TCHOOSEFONTTCHOOSEFONT inc/commdlg.nim NOTE the T prefix is
kept.
Had used UncheckedArray[RGBQUAD] instead of array[0..0,RGBQUAD] in my code
to solve a problem
better solved by {.boundChecks: off.}
make sure 4th parameter of CreateDIBSection is of type: ptr pointer and
just just pointer
When using SendMessage, it may be necessary to coerce the 3rd param into
WPARAM whenever
that param is an int (e.g. any kind of HANDLE). This is because HANDLE is
int whereas
WPARAM is uint32 (when winimCpu32)
See above, but for 4th SendMessage param LPARAM
In SendMessage(---,WM_CTLCOLOREDIT,---,---) return value had to coerce
HBRUSH into LRESULT, because
HBRUSH was HANDLE is int but LRESULT is LONG_PTR is int32 (when winimCpu32)
Run
Re: Trying to learn templates, but its hard to understand.
I want to install 0.19 but i am afraid that whether my current code will work or not.
Re: Trying to learn templates, but its hard to understand.
This is your safest bet:
template Cast(T: typedesc, value: untyped): untyped =
cast[T](value)
echo Cast(int, 3)
Run
To explain the template parameter types, `typedesc` (or just `type` is enough
if you're on 0.19) matches type descriptors, so things like `int`, `Exception`,
`seq`, `Slice[int]` and whatnot. `untyped` matches any AST expression and no
type checking is performed on it, so you can do things like:
template foo(ex: untyped) =
var a {.inject.} = 3 # {.inject.} passes the variable `a` to `ex`
echo ex
foo(a * 4) # 12
Run
On the other hand, templates can also have generics, so you can do:
template Cast[T](_: typedesc[T], value: untyped): T =
cast[T](value)
echo Cast(int, 3)
Run
Like before, if you're on 0.19, you can replace `typedesc[T]` with `type T`.
Trying to learn templates, but its hard to understand.
Hi all, I am trying to learn templates but i coudnt get the idea of templates easily. My aim is to mimic the cast function. Please look this -- template Cast(tp : typed, value : typed) = cast[tp](value) # Failed template Cast[T](tp : T, value : typed) = cast[tp](value) # Failed template Cast(tp : [T], value : typed) = cast[tp](value)# Failed Run All i want to tell the compiler is - template Cast( tp : is_a_generic_type, value : any_typed_value) = cast[ _generic_type](typed_value) Run How to do it ? Any help ?
How do i compile asm files with vcc compiler?
Hey, I am trying following:
assembly.asm:
.code
asm_test PROC
mov rax, 1
Run
assembly.nim:
{.compile: "assembly.asm".}
proc asm_test(): int {.cdecl, importc.}
echo asm_test()
Run
Compilation fails:
C:\nim-coro>nim-0.19.0\bin\nim.exe c --cc:vcc -r assembly.nim
Hint: used config file 'C:\nim-coro\nim-0.19.0\config\nim.cfg' [Conf]
Hint: system [Processing]
Hint: assembly [Processing]
Hint: [Link]
"C:\Windows\system32\cmd.exe" /C ""C:\Program Files (x86)\Microsoft Visual
Studio 14.0\VC\vcvarsall" amd64 && SET"
cl.exe /nologo /DEBUG /Zi /F33554432 /FeC:\nim-coro\assembly.exe
C:\Users\Administrator\nimcache\assembly_d\assembly.asm.obj
C:\Users\Administrator\nimcache\assembly_d\assembly.c.obj
C:\Users\Administrator\nimcache\assembly_d\stdlib_system.c.obj
LINK : fatal error LNK1181: cannot open input file
'C:\Users\Administrator\nimcache\assembly_d\assembly.asm.obj'
Error: execution of an external program failed: 'vccexe.exe
--platform:amd64 /nologo /DEBUG /Zi /F33554432 /FeC:\nim-coro\assembly.exe
C:\Users\Administrator\nimcache\assembly_d\assembly.asm.obj
C:\Users\Administrator\nimcache\assembly_d\assembly.c.obj
C:\Users\Administrator\nimcache\assembly_d\stdlib_system.c.obj '
assembly.asm.obj appears to be never compiled. Am i missing something or is
this a bug?
Re: Opposite of astGenRepr
I should explain exactly what I'm trying to do:
I'm trying to make macros. In my macros, I'd like to start with a base AST,
then modify it according to passed-in arguments. I'm having trouble making that
base AST.
There's `dumpAstGen` which makes it easy to print out the AST-generating code.
For example:
dumpAstGen:
echo something
# produces:
# nnkStmtList.newTree(
# nnkCommand.newTree(
# newIdentNode("echo"),
# newIdentNode("something")
# )
# )
Run
I've had success copying and pasting that output back into my source. But it's
cumbersome to copy and paste a bunch of trees.
So I tried using `quote`. But `quote` produces `newSymNode` in places where
`dumpAstGen` produces `newIdentNode`:
macro usingQuote(): untyped =
var x = quote do:
echo something
echo x.astGenRepr()
usingQuote
# produces:
# nnkCommand.newTree(
# newSymNode("echo"),
# newIdentNode("something")
# )
Run
Unfortunately, `newSymNode` breaks where `newIdentNode` didn't. (I haven't yet
been able to make a small, reproducible example for this breakage).
And that's why I tried making my own kind of `quote` substitute in that first
post. But I can't figure out how to evaluate a string as produced by
`astGenRepr`
Re: Read gzip-compressed file line by line
Storing the whole file will not work very well for the application, where the input file could be 10 or 100 Gb :/ I need an iterator function (that I will create) that reads the file line by line inside the gzip archive and yields each meaningfull chunk (DNA sequences and their associated names and quality) on the fly.
Re: how to increase velocity for merging PRs?
I created this document, but IMO it would be much better if most `.nim` files would get a "Code-owner: Name here" comment instead, who wants to keep these tables up-to-date... [https://raw.githubusercontent.com/nim-lang/Nim/devel/doc/codeowners.rst](https://raw.githubusercontent.com/nim-lang/Nim/devel/doc/codeowners.rst)
Re: `import foo {.private.}` to allows access to private fields (eg: package-level visibility)
Hi The Haxe language has that feature, you can see how they did it here: [https://haxe.org/manual/lf-access-control.html](https://haxe.org/manual/lf-access-control.html) you know, learn from their lessons about intended usage. In general, I think it's a mis-feature, since it can break my design. I only found it useful for unit testing, when the tester is in a different module, and you wish to test all the private functions as well. for what it's worth, C# has a related feature, which is the an access modifier called "internal", which means it's public for everything inside the dll (the same assembly) even if it contains numerous namespaces (modules). I think this is a better way, design-wise, though not as flexible as your suggestion.
Re: `import foo {.private.}` to allows access to private fields (eg: package-level visibility)
I like the idea and would call it `{.exposePrivate.}`, because that is what is
does in my understanding.
