Under unknown circumstances paket adds unlisted package reference

[alexeyzimarev]

Description

Paket 5.91.0

I have an internal package, which depends on MassTransit.SerilogIntegration package. When I add reference to my package in paket.dependencies and run the install, I get this in the logs:

 - FAKE 4.63
 - MassTransit 3.5.7
 - MassTransit.Quartz 3.5.7
 - MassTransit.RabbitMq 3.5.7
 - MassTransit.SerilogIntegration 2013.7.18
     MassTransit.SerilogIntegration 2013.7.18 was unlisted
 - MassTransit.SerilogIntegration 2013.7.15.2
 - OctopusTools 4.22.1
 - RavenDb.Client 3.5.4

Hence the package MassTransit.SerilogIntegration 2013.7.15.2 is also unlisted but it does not get this warning and does not get skipped, just gets installed. This also happens when I add MassTransit.SerilogIntegration to paket.dependencies file.

However, when using direct reference to MassTransit.SerilogIntegration in a small test project works just fine:

Resolving packages for group Main:
 - MassTransit 3.5.7
 - MassTransit.SerilogIntegration 2013.07.18
     MassTransit.SerilogIntegration 2013.07.18 was unlisted
 - MassTransit.SerilogIntegration 2013.07.15.2
     MassTransit.SerilogIntegration 2013.07.15.2 was unlisted
 - MassTransit.SerilogIntegration 2013.07.15
     MassTransit.SerilogIntegration 2013.07.15 was unlisted
 - MassTransit.SerilogIntegration 3.5.7
 - Newtonsoft.Json 10.0.3

I cannot find any reason for this to happen. The workaround now is to use version lock on that package. May be someone who knows the "unlisted" part, can look why can it be ignored?

[forki]

There is one case where we install unlisted packages and that's when it's the only package that makes the resolution valid.
Maybe it's such a case?

[alexeyzimarev]

No, the second log shows correct behavior. There are many listed versions.

On Thu, 31 Aug 2017 at 17:38, Steffen Forkmann ***@***.***> wrote: There is one case where we install unlisted packages and that's when it's the only package that makes the resolution valid. Maybe it's such a case? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#2699 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ACsMVekMDsje7m1FGZhNAumMEurNSxSiks5sdtNbgaJpZM4PJATs> .

-- Med vennligst hilsen / Best regards, Alexey V. Zimarev

[forki]

Ok. Thanks for repro. I will try to find some time to investigate this next week after my parental leave Am 31.08.2017 6:39 nachm. schrieb "Alexey Zimarev" <notifications@github.com

…

: No, the second log shows correct behavior. There are many listed versions. On Thu, 31 Aug 2017 at 17:38, Steffen Forkmann ***@***.***> wrote: > There is one case where we install unlisted packages and that's when it's > the only package that makes the resolution valid. > Maybe it's such a case? > > — > You are receiving this because you authored the thread. > Reply to this email directly, view it on GitHub > <#2699 (comment) >, > or mute the thread > <https://github.com/notifications/unsubscribe-auth/ ACsMVekMDsje7m1FGZhNAumMEurNSxSiks5sdtNbgaJpZM4PJATs> > . > -- Med vennligst hilsen / Best regards, Alexey V. Zimarev — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#2699 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AADgNJV_HioP83u8HRPFDETb-AYpCEv0ks5sduGxgaJpZM4PJATs> .

[forki]

oh I was planning to review this one, but now I see there is no repro since we don't have your package!?

maybe you can mail it to me (remove the libs or something) or create one that has similar deps!?

Or how can I repro?

[alexeyzimarev]

Please don't close it, I will try to figure out how to reproduce it.

[forki]

nonono I didn't want to close it. we just need a way repro on my machine