[Rubygems-developers] gem list as input to gem install?

Hugh Sasse hgs at dmu.ac.uk
Mon Feb 1 16:07:27 EST 2010

On Mon, 1 Feb 2010, Chad Woolley wrote:

> On Mon, Feb 1, 2010 at 11:31 AM, Hugh Sasse <hgs at dmu.ac.uk> wrote:
> > On the other hand, I don't see how Bundler would help with this.  [Which
> > is fair enough, it is outside the scope of its design (for recording
> > gems on a per project basis)].
> Well, assuming that Sexp#sexp_type comes from a gem, Bundler would
> help in this case.
> Assuming that you are both using Bundler correctly, and it is working
> on Kevin Rutherford's system but not yours, you can be sure that it is
> not a problem with different gems or gem versions.  This is because
> bundler locks down the *entire* gem dependency tree.
> So, if you were using Bundler in this case, and still had an error
> which was reproducible on only one system, you'd be pretty sure it is
> a problem related to a native gem, or your ruby interpreter, or
> something else that differs in your environment - not because you are
> using different gems or versions.
> Let me know if I'm missing something...

I see the above would be true if it were in use.  My point is that his
project doesn't use it, and I can't impose it on him, or any arbitrary
project.  So I agree that it will be a useful thing for new projects, 
but for existing code bases which don't use bundler, I would
need something else.

Or are you telling me that I can wrap a code base with Bundler, leaving
the code base unchanged?  Then the Bundler functionality can be used
to verify things by the author, even when they don't want that dependency,
within their project?

I would still find this formatting of the output useful, because it
would allow replication of the environment, in order to replicate
bugs, and in order to ensure consistency across a group of machines.
Those are the only use cases I can think of at the moment, but the
ability of the software to manipulate data about (other instances
of) itself seems to be a theme in computing: lisp code is lisp data,
YAML is specified in YAML.  

> -- Chad

        Thank you,

More information about the Rubygems-developers mailing list