Received: by 10.42.78.194 with SMTP id o2mr5734658ick.12.1347214718619;
        Sun, 09 Sep 2012 11:18:38 -0700 (PDT)
X-BeenThere: python-virtualenv@googlegroups.com
Received: by 10.50.36.129 with SMTP id q1ls2770460igj.0.gmail; Sun, 09 Sep
 2012 11:18:38 -0700 (PDT)
Received: by 10.43.53.144 with SMTP id vq16mr5667793icb.26.1347214718139;
        Sun, 09 Sep 2012 11:18:38 -0700 (PDT)
Received: by 10.43.53.144 with SMTP id vq16mr5667790icb.26.1347214718124;
        Sun, 09 Sep 2012 11:18:38 -0700 (PDT)
Return-Path: <p.f.mo...@gmail.com>
Received: from mail-iy0-f178.google.com (mail-iy0-f178.google.com [209.85.210.178])
        by gmr-mx.google.com with ESMTPS id rf8si2318439igb.2.2012.09.09.11.18.38
        (version=TLSv1/SSLv3 cipher=OTHER);
        Sun, 09 Sep 2012 11:18:38 -0700 (PDT)
Received-SPF: pass (google.com: domain of p.f.mo...@gmail.com designates 209.85.210.178 as permitted sender) client-ip=209.85.210.178;
Authentication-Results: gmr-mx.google.com; spf=pass (google.com: domain of p.f.mo...@gmail.com designates 209.85.210.178 as permitted sender) smtp.mail=p.f.mo...@gmail.com; dkim=pass header...@gmail.com
Received: by mail-iy0-f178.google.com with SMTP id h11so1113214iae.9
        for <python-virtualenv@googlegroups.com>; Sun, 09 Sep 2012 11:18:38 -0700 (PDT)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed;
        d=gmail.com; s=20120113;
        h=mime-version:in-reply-to:references:date:message-id:subject:from:to
         :content-type;
        bh=I9+j0USLN7r3Ju3pO1PJaZk2F2T3DtreyejRXh3mKR4=;
        b=SU3nNf23QBEWhonx97nLjfcKE5AmjjeP+EDy1Dh09li8CeF7z11RrnOK9DhSTsGgpW
         t4EgTdlDSAraH8tvRR645eYSasKF+Nse4y9mftBxPd3G92CExAxY36qJd8u/vimFrP3d
         Lh4j5AhAZQjXxBkktO2LI1IswG1gSqkMg74PHYCPCjEACR9umSuOGlyPyOc+Etm78Egc
         cQvFSzuQoUvgUXVWd74zTTn1xfsOmZel8RuFJlS26sSpQGXLutobsF+qSIQRXbBbIh49
         //MYMEzUKVA84xVaZ9JKnl0JBB8/WwOPTtcR4o4n/J0BZE02N7wM5DqlnSqjgkCK0oaE
         g+1w==
MIME-Version: 1.0
Received: by 10.50.156.170 with SMTP id wf10mr7504864igb.27.1347214717978;
 Sun, 09 Sep 2012 11:18:37 -0700 (PDT)
Received: by 10.64.56.34 with HTTP; Sun, 9 Sep 2012 11:18:37 -0700 (PDT)
In-Reply-To: <504CC9E8.5010...@oddbird.net>
References: <CACac1F9ve48WpjWFW3qEOMRHrz5paVY99TRy8ET5j_CcmsJ...@mail.gmail.com>
	<504CC9E8.5010...@oddbird.net>
Date: Sun, 9 Sep 2012 19:18:37 +0100
Message-ID: <CACac1F8Vp+2FBb2B1tLytyPMu1UVOJu82XZJh4ciOddXXL7...@mail.gmail.com>
Subject: Re: [venv] pip bundles
From: Paul Moore <p.f.mo...@gmail.com>
To: python-virtualenv@googlegroups.com
Content-Type: text/plain; charset=ISO-8859-1

On 9 September 2012 17:55, Carl Meyer <c...@oddbird.net> wrote:
>> (As a side note, why doesn't --find-links accept local directory
>> names, rather than requiring me to make them into file URLs, which are
>> a pain as they don't tab-complete, they are hard to remember, and I
>> can't get relative names to work...)
>
> Totally agree. There's a bug open for this, all that's missing is the patch.

Looks like I've found my next thing to work on, then :-)

To be honest, I suspect people working in Unix probably have less
trouble with file URLs because they are basically just an absolute
path with file:/ tacked on the front. So there's not much motivation
to worry about it. On Windows, you have to switch backslash for
forward slash and worry about which variation of syntax for drive
letters you should use this time. So waiting till a Windows user like
me is annoyed enough to write a patch is by far the most effective
approach :-)

>> So I thought I'd create a bundle from this. Following the manual, I
>> assumed I should be able to do
>>
>> pip bundle prereqs.bundle --no-index --find-links
>> file://path/to/my/directory -r dependencies.req
>>
>> and then in a new virtualenv, just pip install prereqs.bundle. But
>> when I tried this, I got an error:
>
> I don't use or recommend the use of pip's bundle feature (it's poorly
> tested, not widely used, has significant known bugs, and isn't necessary
> -- just distribute your directory of sdists and use --find-links
> instead). We've had discussions for a couple years now about deprecating
> and eventually removing it; I've just added a note to the docs to that
> effect.

My original intention was to distribute the sdists like you suggest.
The only reason I tried the bundle thing is because I spotted it in
the manual and it avoids the file URL madness :-)

> But from looking at the docs and a quick grep of the source code, it
> appears that the ".pybundle" extension is special-cased, so it's not
> optional. (Note that it's ".pybundle", not ".bundle"). Maybe try using
> that extension?

That worked - who'da thought it?

Thanks. That's really helpful.

Paul.