Badly behaved packages may break nuget helper

[Vidarls]

ref: fsprojects/ProjectScaffold#111

Apparently some packages (cough MySql.Data, 5.1.7.0 cough) for some reason have a nuget.exe in them, and it is not exactly the latest version.

By default, the Fake nuget helper searches subdirectories(?) to find the first nuget.exe there is and uses that. Problems occurs when it is a really old version that crashes for unknown reasons due to complicated new fields like release notes.

One solution would be to have a list of "common places that the real nuget.exe might be expected to be found", IE:

• .nuget/Nuget.exe (standard nuget package restore location)
• packages/Nuget.Commandline/tools/Nuget.exe (where it is found when using paket to get the tools)
• Others

and then add some priority to the paths according to which one is more likely to be updated.

[forki]

I will implement this.
Also it would be good to log an issue to the broken package. They should
depend on nuget.commandline instead of bundling it.
On Oct 22, 2014 12:00 PM, "Vidar Løvbrekke Sømme" [email protected]
wrote:

ref: fsprojects/ProjectScaffold#111
fsprojects/ProjectScaffold#111

Apparently some packages (cough MySql.Data, 5.1.7.0 cough) for some reason
have a nuget.exe in them, and it is not exactly the latest version.

By default, the Fake nuget helper searches subdirectories(?) to find the
first nuget.exe there is and uses that. Problems occurs when it is a really
old version that crashes for unknown reasons due to complicated new fields
like release notes.

One solution would be to have a list of "common places that the real
nuget.exe might be expected to be found", IE:

• .nuget/Nuget.exe (standard nuget package restore location)
• packages/Nuget.Commandline/tools/Nuget.exe (where it is found when
  using paket to get the tools)
• Others

and then add some priority to the paths according to which one is more
likely to be updated.

—
Reply to this email directly or view it on GitHub
#572.

[Vidarls]

Just confirming that all works as expected after the fix.

Thanks 😄

[forki]

Very cool and thanks for reporting.
On Oct 22, 2014 1:01 PM, "Vidar Løvbrekke Sømme" [email protected]
wrote:

Just confirming that all works as expected after the fix.

Thanks [image: 😄]

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