Improve paket's conflict reporting during resolution time

[TheAngryByrd]

Paket is being odd and downloading a lot of unnecessary packages. Sorry I can't give you a repro repo right now. I'll try to later tonight.

Paket version 2.17.1.0

[forki]

It doesn't download these packages. It tries to find a resolution. It seems
there are very narrow dependency restrictions.
On Oct 26, 2015 8:30 PM, "Jimmy Byrd" notifications@github.com wrote:

Paket is being odd and downloading a lot of unnecessary packages. Sorry I
can't give you a repro repo right now. I'll try to later tonight.

Paket version 2.17.1.0
[image: screen shot 2015-10-26 at 3 25 46 pm]
https://cloud.githubusercontent.com/assets/1490044/10739982/03f0eef0-7bf6-11e5-8ca9-0cb36fdf0981.png

—
Reply to this email directly or view it on GitHub
#1166.

[isaacabraham]

Turn on verbose mode - usually gives the answer as to why this is.

[TheAngryByrd]

Ah ok. Will do.

[TheAngryByrd]

Ok, so I was pinned to Nancy 1.2.0 in my paket.dependencies but when I added a reference to Nancy.Serialization.JsonNet, it started spinning its wheels looking for the 1.3.0. Updated Nancy to 1.3.0 and it was fine.

[forki]

So I guess if we would report like every nth conflict during resolution
(without using -v) this would already have helped you. Right?
On Oct 26, 2015 8:49 PM, "Jimmy Byrd" notifications@github.com wrote:

Closed #1166 #1166.

—
Reply to this email directly or view it on GitHub
#1166 (comment).

[isaacabraham]

@forki was just thinking this.

[TheAngryByrd]

Yeah that would be a good idea.

[forki]

could you please retry that failing state with paket 2.17.2 and check if it's better?

[TheAngryByrd]

Sure, I'll take a look.

[TheAngryByrd]

It started trying to resolve about 10 minutes ago, and I haven't gotten a warning yet. I'll try to get you a repro repo.

[forki]

Just found my bug. But yes please send repro.
On Oct 27, 2015 14:59, "Jimmy Byrd" notifications@github.com wrote:

It started trying to resolve about 10 minutes ago, and I haven't gotten a
warning yet. I'll try to get you a repro repo.

—
Reply to this email directly or view it on GitHub
#1166 (comment).

[TheAngryByrd]

This is the simplest case I could come up with.
https://github.com/TheAngryByrd/ResolveWarningRepro

I added a few dependencies and pinned Nancy to 1.2.0

nuget FSharp.Formatting
nuget Microsoft.AspNet.SignalR.Core
nuget Microsoft.AspNet.SignalR.JS
nuget Microsoft.Owin.Hosting
nuget Microsoft.Owin.Security.Cookies
nuget Nancy 1.2.0
nuget Nancy.MSOwinSecurity
nuget NUnit 
nuget NUnit.Runners
nuget FAKE
nuget SourceLink.Fake

To reproduce what I was seeing just run

.paket/paket.exe add nuget Nancy.Serialization.JsonNet

The output I got was:

Resolving packages for group Main:
 - Octokit 0.16.0
 - Nancy 1.2.0
 - FAKE 4.7.2
 - FSharp.Formatting 2.12.0
 - Microsoft.AspNet.SignalR.Core 2.2.0
 - Microsoft.AspNet.SignalR.JS 2.2.0
 - Microsoft.Owin.Hosting 3.0.1
 - Microsoft.Owin.Security.Cookies 3.0.1
 - Nancy.MSOwinSecurity 1.0.1
 - Nancy.Serialization.JsonNet 1.3.0
 - NUnit 2.6.4
 - NUnit.Runners 2.6.4
 - SourceLink.Fake 1.1.0
 - FSharpVSPowerTools.Core 2.1.0
 - FSharp.Compiler.Service 1.4.0.6
 - Newtonsoft.Json 7.0.1
 - Microsoft.Owin 3.0.1
 - Microsoft.Owin.Security 3.0.1
 - Owin 1.0.0
 - jQuery 2.1.4
 - jQuery 2.1.3
 - jQuery 2.1.2
 - jQuery 2.1.1
 - jQuery 2.1.0
 - jQuery 2.0.3
 - jQuery 2.0.2
 - jQuery 2.0.1.1
 - jQuery 2.0.1
 - jQuery 2.0.0
 - jQuery 1.11.3
 - jQuery 1.11.2
 - jQuery 1.11.1
 - jQuery 1.11.0
 - jQuery 1.10.2
 - jQuery 1.10.1
 - jQuery 1.10.0.1
 - jQuery 1.10.0
 - jQuery 1.9.1
 - jQuery 1.9.0
 - jQuery 1.8.3
 - jQuery 1.8.2
 - jQuery 1.8.1
 - jQuery 1.8.0
 - jQuery 1.7.2
 - jQuery 1.7.1.1
 - jQuery 1.7.1
 - jQuery 1.7.0
 - jQuery 1.6.4
 - Newtonsoft.Json 6.0.8
 - Newtonsoft.Json 6.0.7
 - Newtonsoft.Json 6.0.6
 - Newtonsoft.Json 6.0.5
 - Newtonsoft.Json 6.0.4
 - SourceLink.Fake 1.0.0
 - SourceLink.Fake 0.6.1
 - SourceLink.Fake 0.6.0
 - SourceLink.Fake 0.5.0
 - SourceLink.Fake 0.4.2
 - SourceLink.Fake 0.4.0
 - SourceLink.Fake 0.3.4
 - SourceLink.Fake 0.3.3
 - SourceLink.Fake 0.3.2
 - SourceLink.Fake 0.3.1
 - SourceLink.Fake 0.3.0
 - NUnit.Runners 2.6.3
 - NUnit.Runners 2.6.2
 - NUnit.Runners 2.6.1
 - NUnit.Runners 2.6.0.12051
 - NUnit 2.6.3
 - NUnit 2.6.2
 - NUnit 2.6.1
 - NUnit 2.6.0.12054
 - NUnit 2.5.10.11092

and I just stopped it after a few minutes, not sure if it would keep going.

[forki]

[forki]

I wonder if we can find a little bit of text that clarifies this even further.
At this stage we can't say for sure that there is absolutely no resolution possible, but as the resolver goes on and on we have less confidence to find a resolution. But as I said to be 100% sure the resolver needs to finish (look at all possible configurations).

Anyone have suggestion for expressing this?

[forki]

funny note: after waiting a couple of minutes the case above actually resolves to.

NUGET
  remote: https://nuget.org/api/v2
  specs:
    FAKE (4.7.2)
    FSharp.Compiler.Service (1.4.0.6)
    FSharp.Formatting (2.12.0)
      FSharp.Compiler.Service (1.4.0.6)
      FSharpVSPowerTools.Core (2.1.0)
    FSharpVSPowerTools.Core (2.1.0)
      FSharp.Compiler.Service (>= 1.4.0.6)
    jQuery (2.1.4)
    Microsoft.AspNet.SignalR.Core (2.2.0)
      Microsoft.Owin (>= 2.1.0)
      Microsoft.Owin.Security (>= 2.1.0)
      Newtonsoft.Json (>= 6.0.4)
      Owin (>= 1.0)
    Microsoft.AspNet.SignalR.JS (2.2.0)
      jQuery (>= 1.6.4)
    Microsoft.Owin (3.0.1)
      Owin (>= 1.0)
    Microsoft.Owin.Hosting (3.0.1)
      Microsoft.Owin (>= 3.0.1)
      Owin (>= 1.0)
    Microsoft.Owin.Security (3.0.1)
      Microsoft.Owin (>= 3.0.1)
      Owin (>= 1.0)
    Microsoft.Owin.Security.Cookies (3.0.1)
      Microsoft.Owin (>= 3.0.1)
      Microsoft.Owin.Security (>= 3.0.1)
      Owin (>= 1.0)
    Nancy (1.2.0)
    Nancy.MSOwinSecurity (1.0.1)
      Microsoft.Owin (> 2.0)
      Nancy (> 0.21)
    Nancy.Serialization.JsonNet (1.2.0)
      Nancy (>= 1.2.0)
      Newtonsoft.Json
    Newtonsoft.Json (7.0.1)
    NUnit (2.6.4)
    NUnit.Runners (2.6.4)
    Owin (1.0.0)
    SourceLink.Fake (1.1.0)

Apart from the reporting, it seems something is wrong with the conflict boosting. ==> created #1169

[TheAngryByrd]

Nice! Probably something along the lines of what you said about resolution not 100% sure.
Quote-alpha-0.0.1

The resolution for %s is taking longer than expected.  This may still resolve, but this might take a while. 

[forki]

#1169 resolves this really really quick. so I need to change that conflict tracing again.