Debian Bug report logs - #833340
mini-buildd: please make man page mini-buildd.8 reproducible

Toggle useless messages

Message #5 received at submit@bugs.debian.org (full text, mbox, reply):

From: Boyuan Yang <073plan@gmail.com>

To: Debian Bug Tracking System <submit@bugs.debian.org>

Subject: mini-buildd: please make the build reproducible

Date: Wed, 03 Aug 2016 17:20:47 +0800

[Message part 1 (text/plain, inline)]

Source: mini-buildd
Version: 1.0.16
Severity: wishlist
Tags: patch

Hi,

The current unreproducible status is due to different env HOME value generated
by help2man.

An example patch is provided to fix make the package reproducible.

Thanks.

[mini-buildd.diff.txt (text/x-diff, attachment)]

Message #12 received at 833340@bugs.debian.org (full text, mbox, reply):

From: Chris Lamb <lamby@debian.org>

To: 833340@bugs.debian.org

Cc: 833340-submitter@bugs.debian.org

Subject: Re: mini-buildd: please make the build reproducible

Date: Wed, 03 Aug 2016 14:14:16 +0200

user reproducible-builds@lists.alioth.debian.org
usertags 833340 timestamps
thanks

Hi,

> mini-buildd: please make the build reproducible

- default=os.getenv("HOME"),
+ default="$HOME",

Unless I'm missing something, this won't work as $VARS are not expanded
in Python..

(What you probably want to do is set the default to ~/.mini-buildd and
patch the callsites to use os.path.expanduser at "real" runtime.)


Regards,

-- 
      ,''`.
     : :'  :     Chris Lamb
     `. `'`      lamby@debian.org / chris-lamb.co.uk
       `-

Message #20 received at 833340@bugs.debian.org (full text, mbox, reply):

From: Boyuan Yang <073plan@gmail.com>

To: Chris Lamb <lamby@debian.org>

Cc: 833340@bugs.debian.org, 833340-submitter@bugs.debian.org

Subject: Re: Bug#833340: mini-buildd: please make the build reproducible

Date: Wed, 3 Aug 2016 20:37:53 +0800

Control: user reproducible-builds@lists.alioth.debian.org
Control: usertags -1 environment
X-Debbugs-Cc: reproducible-builds@lists.alioth.debian.org

2016-08-03 20:14 GMT+08:00 Chris Lamb <lamby@debian.org>:
> - default=os.getenv("HOME"),
> + default="$HOME",
>
> Unless I'm missing something, this won't work as $VARS are not expanded
> in Python..
>
> (What you probably want to do is set the default to ~/.mini-buildd and
> patch the callsites to use os.path.expanduser at "real" runtime.)

I intended to use "$HOME" string literally since I did not come up
with a better solution. :-(

The original author used os.getenv() in python argparse, so the output
of `mini-buildd --help' really depend on $HOME env var. Unfortunately
the man page is generated using help2man (or similar tool) [1] during
building process and reproducible build website will alter $HOME env
var to test reproducibility. So in my opinion there is no better
option except fixed text string. Mini-buildd script won't and
shouldn't detect if itself is running for help2man or for end-user.

[1] https://tests.reproducible-builds.org/debian/rb-pkg/unstable/armhf/diffoscope-results/mini-buildd.html

Thanks,
Boyuan

Message #28 received at 833340@bugs.debian.org (full text, mbox, reply):

From: Chris Lamb <lamby@debian.org>

To: Boyuan Yang <073plan@gmail.com>

Cc: 833340@bugs.debian.org, 833340-submitter@bugs.debian.org

Subject: Re: Bug#833340: mini-buildd: please make the build reproducible

Date: Wed, 03 Aug 2016 15:11:58 +0200

[Message part 1 (text/plain, inline)]

> The original author used os.getenv() in python argparse, so the output
> of `mini-buildd --help' really depend on $HOME env var

I agree that upstream are using an environment variable called HOME and
note that it ends up in the manpage.

However, this does not imply that the string "$HOME" will be interpolated,
so your patch breaks mini-buildd.

> Unfortunately in my opinion there is no better option except fixed text
> string.

A fixed text string is indeed required, but "$HOME" is not that string :)
What's wrong with my expanduser suggestion?

Patch attached.


Regards,

-- 
      ,''`.
     : :'  :     Chris Lamb
     `. `'`      lamby@debian.org / chris-lamb.co.uk
       `-

[mini-buildd.diff.txt (text/plain, attachment)]

Message #36 received at 833340@bugs.debian.org (full text, mbox, reply):

From: Boyuan Yang <073plan@gmail.com>

To: Chris Lamb <lamby@debian.org>

Cc: 833340@bugs.debian.org, 833340-submitter@bugs.debian.org

Subject: Re: Bug#833340: mini-buildd: please make the build reproducible

Date: Wed, 3 Aug 2016 21:27:45 +0800

2016-08-03 21:11 GMT+08:00 Chris Lamb <lamby@debian.org>:
>> The original author used os.getenv() in python argparse, so the output
>> of `mini-buildd --help' really depend on $HOME env var
>
> I agree that upstream are using an environment variable called HOME and
> note that it ends up in the manpage.
>
> However, this does not imply that the string "$HOME" will be interpolated,
> so your patch breaks mini-buildd.

Sorry, it is indeed my mistake.

> What's wrong with my expanduser suggestion?
>
> Patch attached.

If we need to expand earlier or later, what is the difference between
current situation and patched version? Home directory will eventually
expand to difference strings due to different $HOME env var, and the
package will be unreproducible. Also, this patch should not be alone,
since we need to add os.path.expanduser() somewhere else.

> ... patch the callsites to use os.path.expanduser at "real" runtime

How do we know if the script is running at "real" runtime?

Thanks,
Boyuan

Message #44 received at 833340@bugs.debian.org (full text, mbox, reply):

From: Chris Lamb <lamby@debian.org>

To: Boyuan Yang <073plan@gmail.com>

Cc: 833340@bugs.debian.org, 833340-submitter@bugs.debian.org

Subject: Re: Bug#833340: mini-buildd: please make the build reproducible

Date: Wed, 03 Aug 2016 16:50:40 +0200

> If we need to expand earlier or later, what is the difference between
> current situation and patched version?

With my patch, the manpage contains the string "~" instead of the
value of the HOME environment variable. ie. it is reproducible.


Regards,

-- 
      ,''`.
     : :'  :     Chris Lamb
     `. `'`      lamby@debian.org / chris-lamb.co.uk
       `-

Message #52 received at 833340@bugs.debian.org (full text, mbox, reply):

From: Boyuan Yang <073plan@gmail.com>

To: Chris Lamb <lamby@debian.org>

Cc: 833340@bugs.debian.org, 833340-submitter@bugs.debian.org

Subject: Re: Bug#833340: mini-buildd: please make the build reproducible

Date: Wed, 3 Aug 2016 23:11:13 +0800

2016-08-03 22:50 GMT+08:00 Chris Lamb <lamby@debian.org>:
>> If we need to expand earlier or later, what is the difference between
>> current situation and patched version?
>
> With my patch, the manpage contains the string "~" instead of the
> value of the HOME environment variable. ie. it is reproducible.

OK now I got your idea. That should be a good solution, but we may
need to call os.path.expanduser() somewhere else.

Thanks,
Boyuan

Message #60 received at 833340@bugs.debian.org (full text, mbox, reply):

From: Chris Lamb <lamby@debian.org>

To: Boyuan Yang <073plan@gmail.com>

Cc: 833340@bugs.debian.org, 833340-submitter@bugs.debian.org

Subject: Re: Bug#833340: mini-buildd: please make the build reproducible

Date: Wed, 03 Aug 2016 17:34:53 +0200

> OK now I got your idea. That should be a good solution, but we may
> need to call os.path.expanduser() somewhere else.

I deliberately expand it immediately after command-line parsing to
avoid this.

(Did you see my patch?)


Regards,

-- 
      ,''`.
     : :'  :     Chris Lamb
     `. `'`      lamby@debian.org / chris-lamb.co.uk
       `-

Message #68 received at 833340@bugs.debian.org (full text, mbox, reply):

From: Boyuan Yang <073plan@gmail.com>

To: Chris Lamb <lamby@debian.org>

Cc: 833340@bugs.debian.org, 833340-submitter@bugs.debian.org

Subject: Re: Bug#833340: mini-buildd: please make the build reproducible

Date: Wed, 3 Aug 2016 23:46:51 +0800

2016-08-03 23:34 GMT+08:00 Chris Lamb <lamby@debian.org>:
> I deliberately expand it immediately after command-line parsing to
> avoid this.
>
> (Did you see my patch?)

Really sorry I overlook the last several lines. It is a good solution.

Thanks,
Boyuan

Message #76 received at 833340@bugs.debian.org (full text, mbox, reply):

From: Stephan Sürken <absurd@olurdix.de>

To: Boyuan Yang <073plan@gmail.com>, 833340@bugs.debian.org, Chris Lamb <lamby@debian.org>

Cc: 833340-submitter@bugs.debian.org

Subject: Re: Bug#833340: mini-buildd: please make the build reproducible

Date: Fri, 05 Aug 2016 17:09:42 +0200

Hi guys,

On Mi, 2016-08-03 at 23:46 +0800, Boyuan Yang wrote:
> 2016-08-03 23:34 GMT+08:00 Chris Lamb <lamby@debian.org>:
> > 
> > I deliberately expand it immediately after command-line parsing to
> > avoid this.
> > 
> > (Did you see my patch?)
> Really sorry I overlook the last several lines. It is a good
> solution.

hmm thanks ;). However, I already did a patch (actually before the bug
popped up). It's already pushed (1.0.x branch), so maybe someone may
check if that still has a flaw ;).

Thx!

S

Message #84 received at 833340@bugs.debian.org (full text, mbox, reply):

From: Stephan Sürken <absurd@olurdix.de>

To: 833340@bugs.debian.org, Boyuan Yang <073plan@gmail.com>, Chris Lamb <lamby@debian.org>

Subject: Re: Bug#833340: mini-buildd: please make the build reproducible

Date: Fri, 05 Aug 2016 21:04:30 +0200

Chris, Boyang,

On Fr, 2016-08-05 at 17:09 +0200, Stephan Sürken wrote:
(...)
> > > 
> > > (Did you see my patch?)
> > Really sorry I overlook the last several lines. It is a good
> > solution.
> hmm thanks ;). However, I already did a patch (actually before the
> bug
> popped up). It's already pushed (1.0.x branch), so maybe someone may
> check if that still has a flaw ;).

fwiw, which would have been

https://anonscm.debian.org/git/collab-maint/mini-buildd.git/commit/?h=1.0.x&id=b513a9e398f2c05ec9265a627171eb2281c4d974

This does work alright, but I actually like Chris' patch better, so I
have swapped them.

Still pending, so hopefully the next upload will actually be
reproducible ;).

Thx,

S

Message #89 received at 833340@bugs.debian.org (full text, mbox, reply):

From: Chris Lamb <lamby@debian.org>

To: Stephan Sürken <absurd@olurdix.de>, 833340@bugs.debian.org, Boyuan Yang <073plan@gmail.com>

Subject: Re: Bug#833340: mini-buildd: please make the build reproducible

Date: Fri, 05 Aug 2016 21:16:21 +0200

> This does work alright, but I actually like Chris' patch better, so I
> have swapped them.

:) 


Regards,

-- 
      ,''`.
     : :'  :     Chris Lamb
     `. `'`      lamby@debian.org / chris-lamb.co.uk
       `-

Message #94 received at 833340-close@bugs.debian.org (full text, mbox, reply):

From: Stephan Sürken <absurd@debian.org>

To: 833340-close@bugs.debian.org

Subject: Bug#833340: fixed in mini-buildd 1.0.17

Date: Sun, 04 Sep 2016 19:32:17 +0000

Source: mini-buildd
Source-Version: 1.0.17

We believe that the bug you reported is fixed in the latest version of
mini-buildd, which is due to be installed in the Debian FTP archive.

A summary of the changes between this version and the previous one is
attached.

Thank you for reporting the bug, which will now be closed.  If you
have further comments please address them to 833340@bugs.debian.org,
and the maintainer will reopen the bug report if appropriate.

Debian distribution maintenance software
pp.
Stephan Sürken <absurd@debian.org> (supplier of updated mini-buildd package)

(This message was generated automatically at their request; if you
believe that there is a problem with it please contact the archive
administrators by mailing ftpmaster@ftp-master.debian.org)


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

Format: 1.8
Date: Sun, 04 Sep 2016 18:06:18 +0000
Source: mini-buildd
Binary: python-mini-buildd mini-buildd mini-buildd-common
Architecture: source
Version: 1.0.17
Distribution: unstable
Urgency: medium
Maintainer: Stephan Sürken <absurd@debian.org>
Changed-By: Stephan Sürken <absurd@debian.org>
Description:
 mini-buildd - minimal build daemon - daemon
 mini-buildd-common - minimal build daemon - 08x upgrade dummy package
 python-mini-buildd - minimal build daemon - python library and user tool
Closes: 797585 833338 833340 834321 834329 834683
Changes:
 mini-buildd (1.0.17) unstable; urgency=medium
 .
   gnupg 1.4/2.1 compatibility release. This also adds the new 'hotfix'
   suite to the default layout (see docs) and various fixes...
 .
   * [b513a9e] mini-buildd: Generate reproducible mini-buildd man page.
     (Closes: 833340)
   * [c3969de] control: Vcs-*: Update to anonscm.debian.org. Thanks to
     Boyuan Yang (Closes: 833338)
   * [fd99541] Revert "mini-buildd: Generate reproducible mini-buildd man
     page."
   * [2194991] mini-buildd: man page/reproducible builds: Use Chris' patch
     (which is nicer than my previous patch).
   * [460a6e1] wizards: Add "hotfix" suite to default layout.
   * [7ab128f] gnupg.py: Add flavor() method (compute 'gpg flavor'). Not
     used currently.
   * [88fc314] gnupg.py: Use extra methods to get key and fingerprint form
     1st secret key.
   * [2b6cb38] gnupg.py: Simplify internal impl.
   * [9daee35] gnupg.py: Add flavor class variable (and fail-proof flavor
     getter).
   * [e993dc5] gnupg.py: GPG 2.1: Add '%no-protection' to key generator
     template.
   * [8ee94bc] gnupg.py: Add extra method to get sec user id. Fixes doctest
     for GPG 2.1. Thanks to Santiago Vila (Closes: 834683)
   * [2b29745] misc.py, builder.py: Try to log calls more suitable for cut-
     and-paste debugging. Thanks to Marc Haber (Closes: 797585)
   * [3320484] control: Add dirmngr depends for mini-buildd (needed for GPG
     2.1 'recv-keys' functionality).
   * [61302c6] gnupg.py: Add newline for (2.1) template flavor option
     (needed for 'gen-key' from within mini-buildd).
   * [3047d2f] control: Depend on sbuild >= 0.71. Fixes (1) builds in
     chroots >= stretch (where gpg is gpg 2.1)) and (2) mini-buildd's
     sbuild key workaround (via gpg 2.1 fixup in sbuild, and revertion of
     ascii keyrings in sbuild 0.71). Thanks to Marc Haber (Closes: 834321,
     834329)
   * [969b12a] misc.py: Add compat for urlopen with python versions < 2.7.9
     (relevant for wheezy installations only).
   * [e149d03] gnupg.py: For '--with-colons' calls, use arguments that
     deliver the same result for GPG 1.4 and 2.1.
   * [a0c941d] models/gnupg.py: Fix for slightly changed GPG 2.1-style
     colons output.
   * [c058f99] builder.py: Cleanup subprocess sequence code for sbuild
     call.
Checksums-Sha1:
 818386e578e3c46d7b1c2a17f2a5da0bc94c587c 2154 mini-buildd_1.0.17.dsc
 d107b7cc67f71b0f72c875c739030ca58dbcafca 860276 mini-buildd_1.0.17.tar.xz
Checksums-Sha256:
 2e67a7ac60af6d9976f026da27a20ebcd8ff2a429b71e3dfb36d37ddd7deb9e2 2154 mini-buildd_1.0.17.dsc
 9eb0e80935e6e7cfad288853346102f8e0905211365eef6653c27f590e538711 860276 mini-buildd_1.0.17.tar.xz
Files:
 f461822b7e415981b656308cd0f31492 2154 devel extra mini-buildd_1.0.17.dsc
 0c7c33e7a27320a3553982a03868e8e0 860276 devel extra mini-buildd_1.0.17.tar.xz

-----BEGIN PGP SIGNATURE-----

iQIvBAEBCAAZBQJXzGdNEhxhYnN1cmRAZGViaWFuLm9yZwAKCRAGbegBfith2QyX
EACjnWISllzTpnYrqKSuqW0fzY35k6epDm7vb1zu2Wvhv3zdOXWiVOvBbIl2x9yr
6BPHY72WbxZtslruZsL3vEA+13Qfuurj0zNzhAz+gMJFgVZIo2NOMRmq2OSGnqrC
qfrGpW82VbJF7qvvj5r1mZJyXZc5aXGEF/NyeDWQkNsBorpTkpnIYey6wdcVt6TK
J6xkopc5zh1mq+EpcPrWBEB0cUDkbDNIjVZlxqsd2mMxLzKIck5DPiimZGj+Dhpa
z23yENMDDqGE/NHFBaIaEzRdZp8yE+JgN9pq8qFQFl0YisLXBCBvRLIvZAX+G/6H
WM3SA81IlZTCmfulY30d96AJBS+CO/Ark1PC5HKXwbYl9TZfpcR4eQs1yUV6+pNn
JTzH5uS0MglAYOTFUdtWwsoJeSfSirPTz38/yb1gFsEBKgB6Qd6sbPy8iHNai9NX
MFBKMTp8SH8OmF7t7Q6pTAuQTkqr7EIPexEqfBprtJcTkquYjnev4Fg0yXirrCZF
CDe2woMnyV7TUp42D/d7WZJnrHdVKJtq9KOeDv2obaQw8uFhCv9skO1h8c0n/WhG
RfPn4YK5EJu8E/6NaWa3TTYS3EsB8/Xl8vEArgoyey23ylMKLciDueGAQfalWxEz
Ei6gny3wbmxItPpPUYGk9/xRkxbrFKDfWpy/1mB7JGiL5w==
=F5V9
-----END PGP SIGNATURE-----

Send a report that this bug log contains spam.