Hello,
First and foremost, thanks for NUlid!
I'd love to use NUlid in one of our "core" libraries.

Not sure if you're familiar with TinyIoC; it has a "simple inclusion model". You basically get its source files, no assemblies to reference.

Would it be possible to have a "source only" package for NUlid?

Thank you, and apologies in advance if opening an "issue" is not the correct way to open requests.

Rob

Hey, thanks for providing this library; it's been very useful to me.

I happened to notice that the Guids that were converted from certain Ulids (and vice-versa) didn't maintain the proper lexicographical sort order that I would have expected. I believe that this is because of Microsoft's screwy way of encoding the byte arrays (https://stackoverflow.com/a/16722909).

I forked and made a commit that appears to fix this at
kdawg14@c5ae7c8

I'm not certain exactly how you like to handle pull requests, or if this was something that was intentionally done the way that it was, so I figured I'd post this here and see if this looks like a good correction or if you maybe have any additional information on the subject.

Thanks!

Is there a means of serializing an NUlid object and still maintaining it's lexicographic ordering? More specifically, is there a way or would it make sense to convert an NUlid object into a representation than can be stored and have it still maintain its order amongst other stored NUlid objects?

The Javascript reference has an example where Ulids generated in the same millisecond are monotonically increasing. Is there a way to support this functionality?

Created a pull request with some potential optimizations/code modernisation for Cysharp/Ulidd. Not sure if any changes would apply here, it looks like NUlid and Ulid serialize to Guid in different ways (not 100% sure only glanced at it).

I meant to check if any improvements would apply to NUlid and create a pr, but I've been busy so I'll just dump the work on you 😆

I ran into this issue while trying to convert a GUID to a ULID, but it happened randomly.

The following code should trigger it

for (var i = 0; i < 100_000; i++)
{
    Ulid.Parse(new Ulid(Guid.NewGuid()).ToString());
}

The stack trace:

System.ArgumentOutOfRangeException : Ticks must be between DateTime.MinValue.Ticks and DateTime.MaxValue.Ticks. (Parameter 'ticks')
   at System.DateTime.ThrowTicksOutOfRange()
   at System.DateTimeOffset..ctor(Int64 ticks, TimeSpan offset)
   at NUlid.Ulid.FromUnixTimeMilliseconds(Int64 milliseconds)
   at NUlid.Ulid.ByteArrayToDateTimeOffset(Byte[] value)
   at NUlid.Ulid.Parse(ReadOnlySpan`1 span)

It is possible to add the nuget package to a .NET 4.5.2 project but since the library is for .NET 4.6 it will not be added to the project references. This is really confusing. I guess it is possible to add some kind of .NET framework compatibility requirement to the package so that projects with < .NET 4.6 can not install the package.

I have seen that the MonotonicUlidRng implements monotonic incrementation within a single machine for the same time. I would like to know if it supports monotonic incrementation across multiple machines for the same time.

Any way you can rewrite this into .NET Core?

Hi,

As I said in the title, a byte array is used as the member variable to hold the value of ULID struct. I (maybe others) expect ULID struct as a value type has a default value without an initialization which should be 00000000-0000-0000-0000-000000000000 in this case. I examine the source of GUID from corefx repo and I saw value types is used for member variables. https://github.com/dotnet/corefx/blob/master/src/Common/src/CoreLib/System/Guid.cs

I hit this problem when I mess around with ULID and EF Core. EFCore create instances of entities to materialize query results but didn't use constructors and throw null reference exceptions for ULID types used in entity classes.

I changed implementation and use 16 different byte variables instead of byte array and achieve expected behavior (at least by me).

@RobThree Do you consider to change your implementation. I didn't keep my changes but can do it again if you prefer a pull request, because I'm intended to use NUlid on a new project.

This is not a joke. Here's the setup:

1. The project is a IIS Asp.net application (WebAPI 2) running on .net 4.6
2. Recycle app pool (or just restart the site in IIS)
3. hit a request that uses Ulid.NewUlid() 2 times, (in my case this is primary key generation)

So what's happening is that the 2 requests are blocked until the Startup has finished after AppPool recycle. After startup finishes both requests are let through the gate. I'm unsure about the internal mechanism might be a semaphore.

Both requests call Ulid.NewUlid() at the same millisecond, so the first part is identical. But how come the random part is the same? Is there a data race in this library?

Looking underneath it uses BaseUlidRng.DEFAULTRNG which initalises a static RNGCryptoServiceProvider could this be the data race?

Hello there,

I'm confused regarding the byte order of the time components in the ULID. The original project states the following:

   The components are encoded as 16 octets.
   Each component is encoded with the MSB first (network byte order).

   0                   1                   2                   3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                      32_bit_uint_time_low                     |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |     16_bit_uint_time_high     |       16_bit_uint_random      |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                       32_bit_uint_random                      |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                       32_bit_uint_random                      |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

What I read from the above would result in the following diff in this function:

diff --git a/NUlid/Ulid.cs b/NUlid/Ulid.cs
index 70fe351..3639faa 100644
--- a/NUlid/Ulid.cs
+++ b/NUlid/Ulid.cs
@@ -147,7 +147,7 @@ namespace NUlid
         private static byte[] DateTimeOffsetToByteArray(DateTimeOffset value)
         {
             var mb = BitConverter.GetBytes(value.ToUnixTimeMilliseconds());
-            return new[] { mb[5], mb[4], mb[3], mb[2], mb[1], mb[0] };                                  // Drop byte 6 & 7
+            return new[] { mb[2], mb[3], mb[4], mb[5], mb[0], mb[1] };                                  // Drop byte 6 & 7
         }

         private static DateTimeOffset ByteArrayToDateTimeOffset(byte[] value)

Can you help me understand this please? Thank you :-)

https://github.com/ulid

Hello!

First, thanks for your job!

When using the latest version 1.7.0 of NUlid, we have this error when generate and then parse NUlid code

Example with this : "01FTX9NB2FVWG9MYN7ZEE03HX0"

'new Ulid("01FTX9NB2FVWG9MYN7ZEE03HX0")' threw an exception of type 'System.ArgumentOutOfRangeException'
ActualValue: null
Data: {System.Collections.ListDictionaryInternal}
HResult: -2146233086
HelpLink: null
InnerException: null
Message: "Index and length must refer to a location within the string.\r\nParameter name: length"
ParamName: "length"
Source: "mscorlib"
StackTrace: " at System.String.Substring(Int32 startIndex, Int32 length)\r\n at NUlid.Ulid.Parse(String s)\r\n at NUlid.Ulid..ctor(String ulid)"
TargetSite: {System.String Substring(Int32, Int32)}

We are using framework .NET 4.7.1 project

Thanks for your help !