Re: Stream IO errors with -d:release flag
Thank you for your answers. I appreciate that very much. Its a bit funny but i don't use c in this context. The problem I have is that it didn't work with the flag -d: release. I have now tried something out and found out the following: When I create the stream from the previously opened file, these errors occur in the release build, but when I create a new stream from filepath, the program works fine broken: stream = newFileStream(FILE) ... stream.setPosition(blabla) Run working solutiuon: stream = newFileStream(filename) ... stream.setPosition(blabla) Run I'm not sure, maybe I still have a worm somewhere, but when I open the stream from the path it seems to work. It is still strange that it does not work with the stream from the FILE
Re: VSCode debug: strings and booleans are ugly
Sometimes you really want an interactive debugger though
Re: Faster and Safer Raytracing in Nim
> How about to use counter based random number generater like philox? > [https://www.thesalmons.org/john/random123](https://www.thesalmons.org/john/random123)/ This paper is mentioned in JAX (another Google Deep Learning framework), which was mentioned to me here [https://github.com/mratsim/weave/issues/147#issuecomment-631864783](https://github.com/mratsim/weave/issues/147#issuecomment-631864783) [https://github.com/google/jax/blob/master/design_notes/prng.md](https://github.com/google/jax/blob/master/design_notes/prng.md) This is what I'm doing with my `pair` function + reseed. JAX focused more on the splittable RNG part so i may have reinvented the counter RNG :P. > Probability of that happening can be very small, but it looks like > Birthday_problem. You need to use PRNG that have much longer period than > number of random numbers actually used to avoid random number stream overlap. The birthday paradox takes the square root of your probability, i.e. if you have 2^-256, the birthday paradox changes that to 2^-128. I use 2^256 period in my RNGs due to this, Weave is also using 2^256 for task scheduling. Thank you for your suggestions, it's nice to see people interested in RNG quality in the community :)
Re: Doc generation fails
This issue is now fixed on devel, but there's another regression with `nim doc` \+ `runnableExamples`: [https://github.com/nim-lang/Nim/issues/14485](https://github.com/nim-lang/Nim/issues/14485) .
Git mirror for nimble packages https://reg.nim.moe/
Hi, I shared this on IRC a month ago, but wanted to share it here too after it has been running non-stop for a month. **What is it?** This is mostly a cron job that clones all the Git and Mercurial repositories and offers them through a modified package registry. What I'm trying to accomplish here is two things. First, create a mirror of all the repos since previous experience has shown me that repos can be easily taken down either by the authors or Github itself (Github hasn't taken any Nim related repos to the extend of my knowledge). Second, I don't think of myself as being proficient with Mercurial and don't want to mix my packages with other version control systems, the fix being to transform the Mercurial repos to Git repos via HgGit. An example of a transformed Mercurial repo can be found at [https://cgit.nim.moe/strfmt](https://cgit.nim.moe/strfmt) You can find the packages registry at [https://reg.nim.moe/packages.json](https://reg.nim.moe/packages.json) and an overview of the cloned repos served by cgit at [https://cgit.nim.moe/](https://cgit.nim.moe/). The packages list is updated every 10 minutes and all the packages are fetched and if needed pulled from source every 30 minutes. The url of all the packages is modified to target the ones at [https://git.nim.moe/](https://git.nim.moe/). (Why git.nim.moe and not cgit.nim.moe? Because git.nim.moe uses the git http backend while cgit.nim.moe uses a cgit http with similar functionality but with not enough to be fully used by nimble) For a little bit of context, I'm the maintainer/creator/BDFL of [https://github.com/nimgl/nimgl](https://github.com/nimgl/nimgl) and one of the packages that inspired me was glfw by ephja (I haven't for a while used it since I made my own bindings) but one day it simply got deleted after the mass migration to Gitlab. That's what inspired me to do this thing. **Why should I use this?** You shouldn't. The main reason it exists is so packages don't die out after simply being deleted. The only realistic reason is if you find an active Mercurial package which you would like to use without having to interface with Mercurial. If a package you needed was deleted I would recommend to make a clone for yourself from the mirror repo instead of directly referencing it from nim.moe. **I still want to use it** Well, either if for a reason previously stated or because you don't want to trust Github but you do me (I don't know why this would happen). You can add it by following the [Nimble readme](https://github.com/nim-lang/nimble/blob/master/readme.markdown#user-content-configuration). And simply add: [PackageList] name = "official" url = "https://reg.nim.moe/packages.json; Run My biggest concern is that I'm a college student earning exactly $0 dollars a year (By no means a legal tax statement). So all I can do is a $5 droplet in digitalocean and the yearly expenses of the domain. Currently all the packages' data is sitting at a 5.4Gb (21.86% of the droplet's disk) with still room to grow, the other part being that if enough people use it the little baby droplet won't be able to support the load. **I was using X package but it disappeared, do you have it?** This is a project (Baby project? It's little.) that started a month ago, so all the projects that were deleted before this date (20/04/2020) were not able to be cloned. You can find a list of such projects at [https://reg.nim.moe/not_found.txt](https://reg.nim.moe/not_found.txt). **Please** if you have a fork/clone/zip/png of one of this packages let me know either by DM or a comment here so it can be uploaded even if it isn't the latest version. **I like the idea, but I trust no one. Not even me (Yourself)** You can easily create your own instance, I don't have step by step guide on how to create it but all you need is a git http backend and the cron jobs located at [https://github.com/nim-moe/cron](https://github.com/nim-moe/cron). Also Mercurial and the HgGit extension to be able to push the repo from hg to git. Cgit is not necessary since it is only used to give an overview of what packages have been updated and when was the last update loaded into the system. **The future of the project** a.k.a. **I like the domain** I really like the domain, but this was a mini project made in a few hours mostly to try and ignite a conversation on the topics of Nim's package manager and what we as a community will do to prevent packages from being lost into the void. If any of you have any better ideas or want to use the domain name to start a project related to this issues please feel more than free to contact me. And I would even be happy to contribute some code.
Re: Faster and Safer Raytracing in Nim
How about to use counter based random number generater like philox? [https://www.thesalmons.org/john/random123](https://www.thesalmons.org/john/random123)/ It works like a hash function. When you generate a random number, you make a unique number and pass it the function. You can create unique numbers using pixel coordinate, item index, loop counter, line number, etc. Pros: * deterministic as long as input numbers are deterministic. * Stateless. You can use existing variables to generate random numbers and it can reduce memory usage per thread. * Each random number generations can be executed in parallel as it doesn't need PRNG state update. Increase instruction-level parallelism. Cons: * It require more instructions to generate random number from non-random numbers. > I have found a parallel RNG scheme that allows reproducible multithreaded > result for those who wants to do parallel reproducible Monte-Carlo > simulations: Each RNGs generate different RNG streams but I think 2 RNG streams can be overlapped. All RNGs are seeded with unique random number in your code, but the state of one RNG right after being seeded can be same to the state of other RNG after several status update. (In that case, former RNG generate random number stream that is same to the one from later RNG after several state update) Probability of that happening can be very small, but it looks like [Birthday_problem](https://en.wikipedia.org/wiki/Birthday_problem). You need to use PRNG that have much longer period than number of random numbers actually used to avoid random number stream overlap.
Re: Creating a seq or openarray on unmanaged memory
> But Go's slices are memory-safe and zero-copy and can be stored, for example. They are technically memory-safe but prone to aliasing bugs anyway.
Re: Best Nim translation of C {void*, size_t} struct
It's supposed to work, yes and should be documented in the manual.
Re: Creating a seq or openarray on unmanaged memory
> Only Rust has memory safe zero-copy collection that can be stored in a type > (their slice type). Not exactly. Only Rust has this _[plus](https://forum.nim-lang.org/postActivity.xml#plus) compile-time safety checks and zero runtime overhead. But Go's slices are memory-safe and zero-copy and can be stored, for example. It accomplishes this by having the slice type contain a hidden reference to the object that owns the memory; thus the GC ensures that a slice can't outlive its backing store. That would be pretty easy to support in Nim. I could easily make such an object myself, it just wouldn't work the same as an array/seq/string, as I said above. But thanks for the pointers [sic] to the "it" templates and zero-functional; I'll check those out.
Re: Best Nim translation of C {void*, size_t} struct
Just to make sure you're answering the exact question I asked :) — if I declare
a C proc that takes an `openarray`, when Nim calls that function does it pass
that parameter as a pointer followed by an int? For example:
proc set_bytes(bytes: openarray[byte]) {.importc: "set_bytes".}
let data: seq[byte] = @[1, 2, 3, 4]
set_bytes(foo, data)
Run
Will this work going forward? (I could try it out myself, but that only means
it works in Nim 1.2, not that it's expected/supported...)
Re: Is there a command line one-line to start an HTTP server using Nim?
make it depends on libzip optional sounds nice, user may not need full features.
Re: Nim Cheatsheet PDF (English+Spanish+Latex)
That was the idea, use it as template for more cheatsheets or other stuff. :)
Re: Nim Cheatsheet PDF (English+Spanish+Latex)
Nice! I made a copy that's black-and-white, better suited for laser-printing: [https://www.overleaf.com/read/gytjmgdzmcdt](https://www.overleaf.com/read/gytjmgdzmcdt)
Re: Nim Cheatsheet PDF (English+Spanish+Latex)
Iteration order that depends on runtime values is a design bug.
Re: Is there a command line one-line to start an HTTP server using Nim?
After [https://github.com/enthus1ast/nimAsyncHttpTools/pull/2](https://github.com/enthus1ast/nimAsyncHttpTools/pull/2), `simplehttpserver` compiles and runs well on my machine (RHEL 6.8).
Re: Is there a command line one-line to start an HTTP server using Nim?
@bung You do not have an Issues section on your fork. I wanted to ask you if the dependency on libzip can be made optional? (See [my message above](https://forum.nim-lang.org/t/6386#39399) ). The original `nimhttpd` builds and runs fine on my machine (RHEL 6.8), but trying to run your fork fails as I don't have the needed `libzip`.
Re: Nim Cheatsheet PDF (English+Spanish+Latex)
Fixed. Maybe we need that overload (???) ;P
Re: VSCode debug: strings and booleans are ugly
`assercho` is `assert` and `echo` combined, with extra Debug, it shows the Nim code line and the C code line, works with bool and string. [https://github.com/juancarlospaco/nim-contra](https://github.com/juancarlospaco/nim-contra)
Re: Is there a command line one-line to start an HTTP server using Nim?
Awesome! That is the winner! @Araq It would be now cool to bake that into nim :)
Re: Is there a command line one-line to start an HTTP server using Nim?
OK, it builds find after uninstalling `zip` and doing `nimble install zip@#head`. So the correct fix is: 1. Cut a new release of the `zip` nimble package 2. Cut a new release of your `finder` package with updated zip version in its dependency 3. Finally update your `static_server` package with updated finder version in its dependency
Re: Is there a command line one-line to start an HTTP server using Nim?
ok. .now it compiles, but doesn't run: > could not load: libzip(|2).so(|.4|.2|.1|.0) I'll have to figure this out later. I am using an admin managed OS, a very old OS: RHEL 6.8 (Linux).
Re: Is there a command line one-line to start an HTTP server using Nim?
I tried your repo as well. I cloned it and did `nimble build`, but it failed with this: [http://ix.io/2nAa/text](http://ix.io/2nAa/text) It looks like the `psutil` dependency needs to be updated for a backward incompatible change that happened in nim 1.0.0. Now we have `isDigit` taking only `char` arg, not `string`. I'll open a PR on `psutil` to fix this.
Re: Is there a command line one-line to start an HTTP server using Nim?
In case you just need the original nimhttpd like `python3 -m http.server`, run `nimble install nimhttpd` to install the binary and then run `nimhttpd` every time you want to serve static files.
Re: Is there a command line one-line to start an HTTP server using Nim?
The aim of my post was if there's a way that Just Works for kicking off an adhoc server, like `python3 -m http.server` does, **and** then have that baked into nim as well.
Re: Is there a command line one-line to start an HTTP server using Nim?
Hello, I cloned your repo and did `nimble build`. It failed with this error: [http://ix.io/2nA6/text](http://ix.io/2nA6/text) I am using the latest nim devel.
Re: Change Nim colour on GitHub
yellow is difficult to read on a white background
Re: How to bypass a runtime error ?
I have tried it, in many ways, with no luck :( Thanks for trying. ++
Re: incorrect set-to-int conversion
> Is this a bug, or intentional? If the latter, is there a cleaner workaround > like a pragma? It's intentional but it's not part of the spec and we could change it. The workaround is to add an enum field that is mapped to 0.
Re: VSCode debug: strings and booleans are ugly
Oh, bummer. LLDB supports a Python API for pretty printing as well, but I don't think anyone has made a script for Nim yet.
Re: How to bypass a runtime error ?
`try` / `except` or `try`/`finally` [https://nim-lang.org/docs/tut2.html#exceptions-try-statement](https://nim-lang.org/docs/tut2.html#exceptions-try-statement) from strutils import parseInt # read the first two lines of a text file that should contain numbers # and tries to add them var f: File if open(f, "numbers.txt"): try: let a = readLine(f) let b = readLine(f) echo "sum: ", parseInt(a) + parseInt(b) except OverflowError: echo "overflow!" except ValueError: echo "could not convert string to integer" except IOError: echo "IO error!" except: echo "Unknown exception!" # reraise the unknown exception: raise finally: close(f) Run
Re: Stream IO errors with -d:release flag
@Araq, how do you know it's a callback?
When you create a proc to be used in C (callback):
* Don't use exceptions
* Don't use string or sequences
* Deactivate stacktraces
* Use `int32` or `cint` for `int` use `uint` or `csize_t` for `size_t`
* tag `{.cdecl, gcsafe.}` your proc to detect at compile-time the issue
Also one thing I've noticed is that Nim `endOfFile()` is quitting to early
compared to `c_eof` from the `system/ansi_c` module (see my h264 encoder test
case
[https://github.com/mratsim/trace-of-radiance/blob/99f7d85d/trace_of_radiance/io/h264.nim#L278](https://github.com/mratsim/trace-of-radiance/blob/99f7d85d/trace_of_radiance/io/h264.nim#L278))
Example to pass a Nim callback to C code for MP4 muxing:
[https://github.com/mratsim/trace-of-radiance/blob/99f7d85d/trace_of_radiance/io/mp4.nim#L115-L126](https://github.com/mratsim/trace-of-radiance/blob/99f7d85d/trace_of_radiance/io/mp4.nim#L115-L126)
proc MP4E_open(
sequential_mod_flag: cint,
enable_fragmentation: cint,
token: pointer,
write_callback: proc(
offset: int64,
buffer: pointer,
size: csize_t,
token: pointer
): cint{.cdecl, gcsafe.}
): ptr MP4E_mux_t {.importc, header: "minimp4.h".}
## The C code
{.push stackTrace: off.}
proc writeToFile(
offset: int64,
buffer: pointer,
size: csize_t,
token: pointer
): cint {.cdecl, gcsafe.} =
## The callback
let file = cast[File](token)
file.setFilePos(offset)
let bytesWritten = file.writeBuffer(buffer, size)
return cint(bytesWritten.csize_t != size)
{.pop.}
Run
Re: Stream IO errors with -d:release flag
It's likely that your code is buggy. Use `setupForeignThreadGC()` in callbacks that are called from C and run in a different thread.
Re: Change Nim colour on GitHub
Fwiw orange is my favorite color.
How to bypass a runtime error ?
Hello,
I am trying to build a list of mirrors and append the date when the mirror has
been refreshed. (For doing that I take the last-modified header of a file)
The program built without errors. The program begin to parse and print but, it
ends if it meet an error during the parse.
Is there a way to "bypass" the crap mirror and go to the next one ?
Here is the error output :
https://mirror.labkom.id/pub/OpenBSD
2020-05-27 15:45
2020-05-27 20:08
...
...
https://mirrors.nav.ro/pub/OpenBSD
2020-05-27 15:45
2020-05-26 20:01
/home/f/code/mirrosnap/mirrosnap.nim(55) mirrosnap
/home/f/code/mirrosnap/mirrosnap.nim(44) makeDalist
/usr/local/lib/nim/pure/httpclient.nim(254) lastModified
/usr/local/lib/nim/pure/times.nim(2436) parse
/usr/local/lib/nim/pure/times.nim(2412) parse
/usr/local/lib/nim/pure/times.nim(1827) raiseParseException
Error: unhandled exception: Failed to parse '' with format 'ddd, dd MMM
HH:mm:ss 'GMT''. Parsing ended but there was still patterns remaining
[TimeParseError]
Run
Here is the code :
import httpClient
import strutils
import re
import tables
#import osproc
import times
var
client = newHttpClient(timeout = 2)
mirloc = initTable[string, string]() # mirrors table
basemod = initTable[string, string]() # table : url + base freshness
pkgmod = initTable[string, string]() # table : url + packages freshness
kbase: string
kpkg: string
basem: Response
pkgm: Response
let
httpslist: string =
"https://cvsweb.openbsd.org/cgi-bin/cvsweb/~checkout~/www/httpslist;
#(myArchnl, errarch) = execCmdEx("uname -m")
#myArch = myArchnl.replace(re"\n")
proc mirrorsDico(link: string) =
let
contents = client.getContent(link)
mirrors = contents.splitLines()
for m in mirrors:
if m =~ re"(^http.*(BSD|bsd))(\s{2,})(\w.*$)":
# mirror url --- location
mirloc[matches[0]] = matches[3]
mirrorsDico(httpslist)
proc makeDalist(arch: string) =
let
base = "/snapshots/" & $arch & "/SHA256"
pkg = "/snapshots/packages/" & $arch & "/SHA256"
for k, v in mirloc:
kbase = k & base
kpkg = k & pkg
basem = client.get(kbase)
pkgm = client.get(kpkg)
let
bma = basem.lastModified()
pma = pkgm.lastModified()
bm = bma.format("-MM-dd HH:mm")
pm = pma.format("-MM-dd HH:mm")
#basemod[k] = $bm
#pkgmod[k] = $pm
# TODO find a way to handle error and bypass it
echo k
echo bm
echo pm
makeDalist("amd64")
#for mir, bf in basemod:
# echo(mir, " | ", bf)
Run
Don't hesitate to criticize, this is my first try with nim ;)
Regards.
fredg.
Re: VSCode debug: strings and booleans are ugly
Unfortunately this won't work - GDB doesn't work on Mac since 10.14 Mojave
Re: Nim Cheatsheet PDF (English+Spanish+Latex)
thats not correct for i in 9..4: echo i # 9, 8, 7, 6, 5, 4 Run
Re: Is there a command line one-line to start an HTTP server using Nim?
[https://github.com/bung87/nimhttpd](https://github.com/bung87/nimhttpd) you can try my fork version httpd support serve file through file system, memory zip, file zip.
Re: Is there a command line one-line to start an HTTP server using Nim?
i've build one, but it needs some love (especially the part where ip addresses are queried from the os) [https://github.com/enthus1ast/nimAsyncHttpTools/blob/master/src/simplehttpserver.nim](https://github.com/enthus1ast/nimAsyncHttpTools/blob/master/src/simplehttpserver.nim)
