Bug#831587: stressapptest: please make the build reproducible

[Chris Lamb]

Chris Lamb wrote:

> [..]

Friendly ping on this?


Regards,

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

Bug#831587: stressapptest: please make the build reproducible

[Chris Lamb]

> Would you consider applying this patch and uploading?

Friendly ping on this :)


Best wishes,

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

Bug#831587: stressapptest: please make the build reproducible

[Chris Lamb]

Blanc Pierre wrote:

> The respective dsc file can be found at:
> https://mentors.debian.net/debian/pool/main/s/stressapptest/stressapptest_1.0.6-3.dsc

Uploaded.

In ideal world you would have used the value of SOURCE_DATE_EPOCH in
some way that makes sense when displayed in the application. (I don't
actually know where this timestamp is used).

Seeing "SOURCE_DATE_EPOCH set" is unlikely to be very informative to
end users. Please consider this for your next upload. :)


Regards,

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

Bug#831587: stressapptest: please make the build reproducible

[Blanc Pierre]

Hello,

> I would set it to *something*, not an empty string otherwise wherever
> that string is used in the program itself it will look broken.
I have just updated the package.
https://anonscm.debian.org/git/collab-maint/stressapptest.git/tree/debian/patches/reproducible_build

It has built on mentor:
https://mentors.debian.net/package/stressapptest

The respective dsc file can be found at:
https://mentors.debian.net/debian/pool/main/s/stressapptest/stressapptest_1.0.6-3.dsc

Thank you for your patience.

-- 
Cordialement,
Blanc Pierre

Bug#831587: stressapptest: please make the build reproducible

[Chris Lamb]

> -AC_DEFINE_UNQUOTED([STRESSAPPTEST_TIMESTAMP],
> +if test -n "$SOURCE_DATE_EPOCH"; then
> +  AC_DEFINE(STRESSAPPTEST_TIMESTAMP,"")

I would set it to *something*, not an empty string otherwise wherever
that string is used in the program itself it will look broken.


Regards,

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

Bug#831587: stressapptest: please make the build reproducible

[Blanc Pierre]

Hello,

> - This won't be reproducible as $username is not fixed too.
Right.

> - I don't think always setting "hostname" to "localhost" is right.
>   Feels wrong.
Ok, you right. The better is to remove information

>
> Why not follow my previous suggestion of wrapping the:
>
>  AC_DEFINE_UNQUOTED([STRESSAPPTEST_TIMESTAMP],
>
> when SOURCE_DATE_EPOCH is set?
Something like that:

-AC_DEFINE_UNQUOTED([STRESSAPPTEST_TIMESTAMP],
+if test -n "$SOURCE_DATE_EPOCH"; then
+  AC_DEFINE(STRESSAPPTEST_TIMESTAMP,"")
+else
+  AC_DEFINE_UNQUOTED([STRESSAPPTEST_TIMESTAMP],
"$username @ $hostname on $timestamp",
[Timestamp when ./configure was executed])
+fi
+

Regards

Bug#831587: stressapptest: please make the build reproducible

[Chris Lamb]

> https://anonscm.debian.org/git/collab-maint/stressapptest.git/commit/?id=57885d7144ac4eb9d3861cbf579a0b45a07f9ca3

- This won't be reproducible as $username is not fixed too. 

- I don't think always setting "hostname" to "localhost" is right.
  Feels wrong.

- Using "Author: [me]" when I did not write the patch is... strange.

Why not follow my previous suggestion of wrapping the:

 AC_DEFINE_UNQUOTED([STRESSAPPTEST_TIMESTAMP],

when SOURCE_DATE_EPOCH is set?

:)


Regards,

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

Bug#831587: stressapptest: please make the build reproducible

[Blanc Pierre]

Hi Chris,

> Well, I am 100% certain you do not want to be patching the generated
> ./configure file so I would not upload this.
You have totally right.

I used you previous patch.
https://anonscm.debian.org/git/collab-maint/stressapptest.git/commit/?id=57885d7144ac4eb9d3861cbf579a0b45a07f9ca3

And this is the package uploaded on mentors:
https://mentors.debian.net/package/stressapptest

Best regards

Bug#831587: stressapptest: please make the build reproducible

[Chris Lamb]

Hi Blanc,

> I checked your patch, are you sure that the fault is the timestamp ?

Well, it was some time ago so I'm looking into it again now...

> So, I have created this patch to fix the issue :
> https://anonscm.debian.org/git/collab-maint/stressapptest.git/commit/?id=3be7d9f2c4d434ad412838d764b104fa659212b6

Well, I am 100% certain you do not want to be patching the generated
./configure file so I would not upload this.

Not only is this very ugly, the changes would simply be overwritten
automatically and silently on an autoreconf as that happens after the
patch phase.

Also, this still leaves the timestamp.

I suggest we patch this:

  AC_DEFINE_UNQUOTED([STRESSAPPTEST_TIMESTAMP],
 "$username @ $hostname on $timestamp",
 [Timestamp when ./configure was executed])

.. in configure.ac for the case when SOURCE_DATE_EPOCH is set.


Regards,

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

Bug#831587: stressapptest: please make the build reproducible

[Blanc Pierre]

Hi Chris,

> No problem; did you get a chance to have a look at this yet? :)

First, I am sincerely sorry for the delay.

I checked your patch, are you sure that the fault is the timestamp ?

The error on Jenkins is
"user_hostname_manually_added_requiring_further_investigation"
https://tests.reproducible-builds.org/debian/rb-pkg/unstable/amd64/stressapptest.html

So, I have created this patch to fix the issue :
https://anonscm.debian.org/git/collab-maint/stressapptest.git/commit/?id=3be7d9f2c4d434ad412838d764b104fa659212b6

I also built it on mentors
https://mentors.debian.net/package/stressapptest

What is your opinion ?

-- 
Cordialement,
Blanc Pierre

Bug#831587: stressapptest: please make the build reproducible

[Chris Lamb]

Hi Blanc,

> I will check and update this bug during this week.
> I'm late, sorry. I'll do my best.

No problem; did you get a chance to have a look at this yet? :)


Regards,

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

Bug#831587: stressapptest: please make the build reproducible

[Chris Lamb]

Dear Maintainer,

> Source: stressapptest
> Version: 1.0.6-2
> Tags: patch

There hasn't seem to be any update on this bug in 64 days, in which
time the Reproducible Builds effort has come on a long way. :)

Would you consider applying this patch and uploading?


Regards,

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

Bug#831587: stressapptest: please make the build reproducible

[Chris Lamb]

Source: stressapptest
Version: 1.0.6-2
Severity: wishlist
Tags: patch
User: reproducible-bui...@lists.alioth.debian.org
Usertags: timestamps
X-Debbugs-Cc: reproducible-bui...@lists.alioth.debian.org

Hi,

Whilst working on the "reproducible builds" effort [0], we noticed
that stressapptest could not be built reproducibly.

Patch attached.

 [0] https://wiki.debian.org/ReproducibleBuilds


Regards,

-- 
  ,''`.
 : :'  : Chris Lamb
 `. `'`  la...@debian.org / chris-lamb.co.uk
   `-
--- a/debian/patches/reproducible_build 1970-01-01 02:00:00.0 +0200
--- b/debian/patches/reproducible_build 2016-07-17 15:55:23.021436026 +0200
@@ -0,0 +1,18 @@
+Author: Chris Lamb 
+Last-Update: 2016-07-17
+
+--- stressapptest-1.0.6.orig/configure.ac
 stressapptest-1.0.6/configure.ac
+@@ -95,7 +95,11 @@ hostname=$(uname -n)
+ AC_MSG_CHECKING([host name])
+ AC_MSG_RESULT([$hostname])
+ 
+-timestamp=$(date)
++if test -n "$SOURCE_DATE_EPOCH"; then
++  timestamp=$(LC_ALL=C date --utc --date=@$SOURCE_DATE_EPOCH)
++else
++  timestamp=$(date)
++fi
+ AC_MSG_CHECKING([current timestamp])
+ AC_MSG_RESULT([$timestamp])
+ 
--- a/debian/patches/series 2016-07-17 15:32:30.793480316 +0200
--- b/debian/patches/series 2016-07-17 15:55:15.493349335 +0200
@@ -1,3 +1,4 @@
 armhf_support
 support_i486_builds
 support_i586_builds
+reproducible_build