Well, right now I’m using ssh-agent. Which is program that ssh uses (which scp uses) to find an in memory map of your passphrase to key. Then when the server requests your private key, ssh gets it from ssh-agent and no user intervention is needed. This is a standard way of ssh’n around secure machines without requiring typed authentication at each step.

 

If you are not familiar with this approach you may want to look into it.

 

You can setup a script to run ssh-agent before nant is run, making everything work out.

 

-----Original Message-----
From:
Andy Smith [mailto:[EMAIL PROTECTED]]
Sent:
Friday, May 31, 2002 3:22 PM
To: Scott Hernandez; [EMAIL PROTECTED]
Subject: RE: [nant-dev] SCP Task

 

That's definitely a problem...

If a task can't except user input, i don't see any other way but to store some password somewhere.

You could use IsolateStorage to house the encrypted password for final use... but you'll still need some kind of password in the task to decrypt the real password.

 

__
Andy Smith
Chief Code Monkey

-----Original Message-----
From: Scott Hernandez [mailto:[EMAIL PROTECTED]]
Sent: Friday, May 31, 2002 4:14 PM
To: [EMAIL PROTECTED]
Subject: [nant-dev] SCP Task

I’ve written a simple SCPTask, that wraps scp.exe. For it to work, you need to have a setup that doesn’t require any password/pass-phrase to be typed. I’m thinking about adding this task to the newly (soon to be) formed NAntContrib project.

 

I’m thinking about ways of encrypting your password to a file so we can use that when prompted, but right it doesn’t work yet. This will hide your password outside the buildfile, but doesn’t really protect the contents much. I would assume this file would not be something that is actually checked into source control.

 

I’m assuming that build files should be completely independent of user input. At no time should a task require the user to type anything, right?

 

Later,

Scott

Reply via email to